| ANIMA WG | T. Eckert, Ed. |
| Internet-Draft | Futurewei USA |
| Intended status: Standards Track | M. Behringer, Ed. |
| Expires: September 10, 2020 | |
| S. Bjarnason | |
| Arbor Networks | |
| March 9, 2020 |
An Autonomic Control Plane (ACP)
draft-ietf-anima-autonomic-control-plane-24
Autonomic functions need a control plane to communicate, which depends on some addressing and routing. This Autonomic Control Plane should ideally be self-managing, and as independent as possible of configuration. This document defines such a plane and calls it the "Autonomic Control Plane", with the primary use as a control plane for autonomic functions. It also serves as a "virtual out-of-band channel" for Operations, Administration and Management (OAM) communications over a network that provides automatically configured hop-by-hop authenticated and encrypted communications via automatically configured IPv6 even when the network is not configured, or misconfigured.
This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.
Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet-Drafts is at https://datatracker.ietf.org/drafts/current/.
Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress."
This Internet-Draft will expire on September 10, 2020.
Copyright (c) 2020 IETF Trust and the persons identified as the document authors. All rights reserved.
This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (https://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License.
Autonomic Networking is a concept of self-management: Autonomic functions self-configure, and negotiate parameters and settings across the network. [RFC7575] defines the fundamental ideas and design goals of Autonomic Networking. A gap analysis of Autonomic Networking is given in [RFC7576]. The reference architecture for Autonomic Networking in the IETF is specified in the document [I-D.ietf-anima-reference-model].
Autonomic functions need an autonomically built communications infrastructure. This infrastructure needs to be secure, resilient and re-usable by all autonomic functions. Section 5 of [RFC7575] introduces that infrastructure and calls it the Autonomic Control Plane (ACP). More descriptively it would be the "Autonomic communications infrastructure for OAM and Control". For naming consistency with that prior document, this document continues to use the name ACP though.
Today, the OAM and control plane of networks typically uses a routing and forwarding table which is dependent on correct configuration and routing. Misconfigurations or routing problems can disrupt OAM and control channels. Traditionally, an out-of-band network has been used to avoid or allow recovery from such problems, or personnel are sent on site to access devices through out-of-band management ports (also called craft ports, serial console, management ethernet port). However, both options are expensive.
In increasingly automated networks either centralized management systems or distributed autonomic service agents in the network require a control plane which is independent of the configuration of the network they manage, to avoid impacting their own operations through the configuration actions they take.
This document describes a modular design for a self-forming, self-managing and self-protecting ACP, which is a virtual in-band network designed to be as independent as possible of configuration, addressing and routing problems. The details how this is achieved are described in Section 6. The ACP is designed to remain operational even in the presence of configuration errors, addressing or routing issues, or where policy could inadvertently affect connectivity of both data packets or control packets.
This document uses the term "Data-Plane" to refer to anything in the network nodes that is not the ACP, and therefore considered to be dependent on (mis-)configuration. This Data-Plane includes both the traditional forwarding-plane, as well as any pre-existing control-plane, such as routing protocols that establish routing tables for the forwarding plane.
The Autonomic Control Plane serves several purposes at the same time:
This document describes these purposes as use cases for the ACP in Section 3, it defines the requirements in Section 4. Section 5 gives an overview how the ACP is constructed.
The normative part of this document starts with Section 6, where the ACP is specified. Section 7 defines normative how to support ACP on L2 switches. Section 8 explains normative how non-ACP nodes and networks can be integrated.
The remaining sections are non-normative: Section 10 reviews benefits of the ACP (after all the details have been defined), Section 9 provides operational recommendations, Appendix A provides additional explanations and describes additional details or future standard or propriety extensions that were considered not to be appropriate for standardization in this document but were considered important to document. There are no dependencies against Appendix A to build a complete working and interoperable ACP according to this document.
The ACP provides secure IPv6 connectivity, therefore it can be used not only as the secure connectivity for self-management as required for the ACP in [RFC7575], but it can also be used as the secure connectivity for traditional (centralized) management. The ACP can be implemented and operated without any other components of autonomic networks, except for the GRASP protocol. ACP relies on per-link DULL GRASP (see Section 6.3) to autodiscover ACP neighbors, and includes the ACP GRASP instance to provide service discovery for clients of the ACP (see Section 6.8) including for its own maintenance of ACP certificates.
The document "Using Autonomic Control Plane for Stable Connectivity of Network OAM" describes how the ACP alone can be used to provide secure and stable connectivity for autonomic and non-autonomic OAM applications. That document also explains how existing management solutions can leverage the ACP in parallel with traditional management models, when to use the ACP and how to integrate with potentially IPv4 only OAM backends.
Combining ACP with Bootstrapping Remote Secure Key Infrastructures (BRSKI), see [I-D.ietf-anima-bootstrapping-keyinfra]) results in the "Autonomic Network Infrastructure" (ANI) as defined in [I-D.ietf-anima-reference-model], which provides autonomic connectivity (from ACP) with secure zero-touch (automated) bootstrap from BRSKI. The ANI itself does not constitute an Autonomic Network, but it allows the building of more or less autonomic networks on top of it - using either centralized, Software Defined Networking- (SDN-)style (see [RFC7426]) automation or distributed automation via Autonomic Service Agents (ASA) / Autonomic Functions (AF) - or a mixture of both. See [I-D.ietf-anima-reference-model] for more information.
Please see the following Terminology section (Section 2) for explanations of terms used in this section.
The design of the ACP as defined in this document is considered to be applicable to all types of "professionally managed" networks: Service Provider, Local Area Network (LAN), Metro(politan networks), Wide Area Network (WAN), Enterprise Information Technology (IT) and ->"Operational Technology" (OT) networks. The ACP can operate equally on layer 3 equipment and on layer 2 equipment such as bridges (see Section 7). The hop-by-hop authentication, integrity-protection and confidentiality mechanism used by the ACP is defined to be negotiable, therefore it can be extended to environments with different protocol preferences. The minimum implementation requirements in this document attempt to achieve maximum interoperability by requiring support for multiple options depending on the type of device: IPsec, see [RFC4301], and datagram Transport Layer Security (DTLS) version 1.2, see [RFC6347]).
The implementation footprint of the ACP consists of Public Key Infrastructure (PKI) code for the ACP certificate, the GRASP protocol, UDP, TCP and TLS (for security and reliability of GRASP), the ACP secure channel protocol used (such as IPsec or DTLS), and an instance of IPv6 packet forwarding and routing via the Routing Protocol for Low-power and Lossy Networks (RPL), see [RFC6550], that is separate from routing and forwarding for the Data-Plane (user traffic).
The ACP uses only IPv6 to avoid complexity of dual-stack ACP operations (IPv6/IPv4). Nevertheless, it can without any changes be integrated into even otherwise IPv4-only network devices. The Data-Plane itself would not need to change, it could continue to be IPv4 only. For such IPv4 only devices, the IPv6 protocol itself would be additional implementation footprint only used for the ACP.
The protocol choices of the ACP are primarily based on wide use and support in networks and devices, well understood security properties and required scalability. The ACP design is an attempt to produce the lowest risk combination of existing technologies and protocols to build a widely applicable operational network management solution:
RPL was chosen because it requires a smaller routing table footprint in large networks compared to other routing protocols with an autonomically configured single area. The deployment experience of large scale Internet of Things (IoT) networks serves as the basis for wide deployment experience with RPL. The profile chosen for RPL in the ACP does not leverage any RPL specific forwarding plane features (IPv6 extension headers), making its implementation a pure control plane software requirement.
GRASP is the only completely novel protocol used in the ACP, and this choice was necessary because there is no existing suitable protocol to provide the necessary functions to the ACP, so GRASP was developed to fill that gap.
The ACP design can be applicable to (cpu, memory) constrained devices and (bitrate, reliability) constrained networks, but this document does not attempt to define the most constrained type of devices or networks to which the ACP is applicable. RPL and DTLS for ACP secure channels are two protocol choices already making ACP more applicable to constrained environments. Support for constrained devices in this specification is opportunistic, but not complete, because the reliable transport for GRASP (see Section 6.8.2) only specifies TCP/TLS). See Appendix A.9 for discussions about how future standards or proprietary extensions/variations of the ACP could better meet different expectations from those on which the current design is based including supporting constrained devices better.
[RFC Editor: WG/IETF/IESG review of the terms below asked for references between these terms when they refer to each other. The only option I could fin RFC/XML to point to a hanging text acronym definition that also displays the actual term is the format="title" version, which leads to references such as '->"ACP domain certificate" ()'. I found no reasonable way to eliminate the trailing '()' generated by this type of cross references. Can you please take care of removing these artefacts during editing (after conversion to nroff ?). I also created a ticket to ask for an xml2rfc enhancement to avoid this in the future: https://trac.tools.ietf.org/tools/xml2rfc/trac/ticket/347.
[RFC Editor: Question: Is it possible to change the first occurrences of [RFCxxxx] references to "rfcxxx title" [RFCxxxx]? the XML2RFC format does not seem to offer such a format, but I did not want to duplicate 50 first references - one reference for title mentioning and one for RFC number.]
This document serves both as a normative specification for how ACP nodes have to behave as well as describing requirements, benefits, architecture and operational aspects to explain the context. Normative sections are labelled "(Normative)" and use [RFC2119]/[RFC8174] keywords. Other sections are labelled "(Informative)" and do not use those normative keywords.
In the rest of the document we will refer to systems using the ACP as "nodes". Typically such a node is a physical (network equipment) device, but it can equally be some virtualized system. Therefore, we do not refer to them as devices unless the context specifically calls for a physical system.
This document introduces or uses the following terms (sorted alphabetically). Terms introduced are explained on first use, so this list is for reference only.
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 [RFC2119],[RFC8174] when, and only when, they appear in all capitals, as shown here.
This section summarizes the use cases that are intended to be supported by an ACP. To understand how these are derived from and relate to the larger set of use cases for autonomic networks, please refer to [RFC8316].
Autonomic Functions need a stable infrastructure to run on, and all autonomic functions should use the same infrastructure to minimize the complexity of the network. In this way, there is only need for a single discovery mechanism, a single security mechanism, and single instances of other processes that distributed functions require.
Today, bootstrapping a new node typically requires all nodes between a controlling node such as an SDN controller ("Software Defined Networking", see [RFC7426]) and the new node to be completely and correctly addressed, configured and secured. Bootstrapping and configuration of a network happens in rings around the controller - configuring each ring of devices before the next one can be bootstrapped. Without console access (for example through an out-of-band network) it is not possible today to make devices securely reachable before having configured the entire network leading up to them.
With the ACP, secure bootstrap of new devices and whole new networks can happen without requiring any configuration of unconfigured devices along the path: As long as all devices along the path support ACP and a zero-touch bootstrap mechanism such as BRSKI, the ACP across a whole network of unconfigured devices can be brought up without operator/provisioning intervention. The ACP also provides additional security for any bootstrap mechanism, because it can provide encrypted discovery (via ACP GRASP) of registrars or other bootstrap servers by bootstrap proxies connecting to nodes that are to be bootstrapped and the ACP encryption hides the identities of the communicating entities (pledge and registrar), making it more difficult to learn which network node might be attackable. The ACP domain certificate can also be used to end-to-end encrypt the bootstrap communication between such proxies and server. Note that bootstrapping here includes not only the first step that can be provided by BRSKI (secure keys), but also later stages where configuration is bootstrapped.
Today, most critical control plane protocols and OAM protocols are using the Data-Plane of the network. This leads to often undesirable dependencies between control and OAM plane on one side and the Data-Plane on the other: Only if the forwarding and control plane of the Data-Plane are configured correctly, will the Data-Plane and the OAM/control plane work as expected.
Data-Plane connectivity can be affected by errors and faults, for example misconfigurations that make AAA (Authentication, Authorization and Accounting) servers unreachable or can lock an administrator out of a device; routing or addressing issues can make a device unreachable; shutting down interfaces over which a current management session is running can lock an admin irreversibly out of the device. Traditionally only out-of-band access can help recover from such issues (such as serial console or ethernet management port).
Data-Plane dependencies also affect applications in a Network Operations Center (NOC) such as SDN controller applications: Certain network changes are today hard to implement, because the change itself may affect reachability of the devices. Examples are address or mask changes, routing changes, or security policies. Today such changes require precise hop-by-hop planning.
Note that specific control plane functions for the Data-Plane often want to depend on forwarding of their packets via the Data-Plane: Aliveness and routing protocol signaling packets across the Data-Plane to verify reachability across the Data-Plane, using IPv4 signaling packets for IPv4 routing vs. IPv6 signaling packets for IPv6 routing.
Assuming appropriate implementation (see Section 6.12.2 for more details), the ACP provides reachability that is independent of the Data-Plane. This allows the control plane and OAM plane to operate more robustly:
The document "Using Autonomic Control Plane for Stable Connectivity of Network OAM" explains this use case for the ACP in significantly more detail and explains how the ACP can be used in practical network operations.
The following requirements were identified for the design of the ACP based on the above use-cases (Section 3). These requirements are informative. The ACP as specified in the normative parts of this document is meeting or exceeding these use-case requirements:
Explanation for ACP4: In a fully autonomic network (AN), newly written ASA could potentially all communicate exclusively via GRASP with each other, and if that was assumed to be the only requirement against the ACP, it would not need to provide IPv6 layer connectivity between nodes, but only GRASP connectivity. Nevertheless, because ACP also intends to support non-AN networks, it is crucial to support IPv6 layer connectivity across the ACP to support any transport and application layer protocols.
The ACP operates hop-by-hop, because this interaction can be built on IPv6 link local addressing, which is autonomic, and has no dependency on configuration (requirement 1). It may be necessary to have ACP connectivity across non-ACP nodes, for example to link ACP nodes over the general Internet. This is possible, but introduces a dependency against stable/resilient routing over the non-ACP hops (see Section 8.2).
The Autonomic Control Plane is constructed in the following way (for details, see Section 6):
Note:
The following figure illustrates the ACP.
ACP node 1 ACP node 2
................... ...................
secure . . secure . . secure
channel: +-----------+ : channel : +-----------+ : channel
..--------| ACP VRF |---------------------| ACP VRF |---------..
: / \ / \ <--routing--> / \ / \ :
: \ / \ / \ / \ / :
..--------| Loopback |---------------------| Loopback |---------..
: | interface | : : | interface | :
: +-----------+ : : +-----------+ :
: : : :
: Data-Plane :...............: Data-Plane :
: : link : :
:.................: :.................:
Figure 1: ACP VRF and secure channels
The resulting overlay network is normally based exclusively on hop-by-hop tunnels. This is because addressing used on links is IPv6 link local addressing, which does not require any prior set-up. In this way the ACP can be built even if there is no configuration on the node, or if the Data-Plane has issues such as addressing or routing problems.
This section describes the components and steps to set up an ACP and highlights the key properties which make it "indestructible" against many inadvertent changes to the Data-Plane, for example caused by misconfigurations.
An ACP node can be a router, switch, controller, NMS host, or any other IP capable node. Initially, it MUST have its ACP domain certificate, as well as an (empty) ACP Adjacency Table (described in Section 6.2). It then can start to discover ACP neighbors and build the ACP. This is described step by step in the following sections:
The ACP relies on group security. An ACP domain is a group of nodes that trust each other to participate in ACP operations such as creating ACP secure channels in an autonomous peer-to-peer fashion between ACP domain members via protocols such as IPsec. To establish trust, each ACP member requires keying material: An ACP node MUST have a Local Device IDentity certificate (LDevID) and a Trust Anchor (TA) consisting of a certificate (chain) used to sign the LDevID of all ACP domain members. The LDevID is used to cryptographically authenticate the membership of its owner node in the ACP domain to other ACP domain members. The TA is used to authenticate the ACP domain membership of other nodes (see Section 6.1.3). This document calls the LDevID used for the the ACP the ACP certificate.
Manual keying via shared secrets is not usable for an ACP domain because it would require a single shared secret across all current and future ACP domain members to meet the expectation of autonomous, peer-to-peer establishment of ACP secure channels between any ACP domain members. Such a single shared secret would be an inacceptable security weakness. Asymmetric keying material (public keys) without certificates does not provide the mechanisms to authenticate ACP domain membership in an autonomous, peer-to-peer fashion for current and future ACP domain members.
The LDevID is called the ACP domain certificate, the TA is the Certificate Authority (CA) root certificate of the ACP domain.
The ACP does not mandate specific mechanisms by which this keying material is provisioned into the ACP node. It only requires the certificate to comply with Section 6.1.1, specifically the Domain information field as specified in Section 6.1.2 in its domain certificate as well as those of candidate ACP peers. See Appendix A.2 for more information about enrollment or provisioning options.
This document uses the term ACP in many places where the Autonomic Networking reference documents [RFC7575] and [I-D.ietf-anima-reference-model] use the word autonomic. This is done because those reference documents consider (only) fully autonomic networks and nodes, but support of ACP does not require support for other components of autonomic networks except for relying on GRASP and providing security and transport for GRASP. Therefore the word autonomic might be misleading to operators interested in only the ACP.
[RFC7575] defines the term "Autonomic Domain" as a collection of autonomic nodes. ACP nodes do not need to be fully autonomic, but when they are, then the ACP domain is an autonomic domain. Likewise, [I-D.ietf-anima-reference-model] defines the term "Domain Certificate" as the certificate used in an autonomic domain. The ACP domain certificate is that domain certificate when ACP nodes are (fully) autonomic nodes. Finally, this document uses the term ACP network to refer to the network created by active ACP nodes in an ACP domain. The ACP network itself can extend beyond ACP nodes through the mechanisms described in Section 8.1.
ACP domain certificates MUST be [RFC5280] compliant X.509 certificates.
ACP nodes MUST support RSA and Elliptic Curve (ECC) public keys in ACP certificates. ACP certificates wth ECC keys MUST indicate to be Elliptic Curve Diffie-Hellman capable (ECDH) if X.509 v3 keyUsage and extendedKeyUsage are included in the certificate.
ACP nodes MUST support Certificates RSA keys with no less than 2048 bit key length and ECC keys with NIST P-256, P-384 and P-521 key length (and curve) or better. ACP nodes MUST support SHA-256, SHA-384, SHA-512 or better signatures for ACP certificates with RSA key and the same RSA signatures plus ECDSA signatures for ACP certificates with ECC key.
The ACP certificate SHOULD use an RSA key and an RSA signature when the ACP certificate is intended to be used not only for ACP authentication but also for other purposes. The ACP certificate MAY use an ECC key and an ECDSA signature if the ACP certificate is only used for ACP and ANI authentication and authorization.
Any secure channel protocols used for the ACP as specified in this document or extensions of this document MUST therefore support authentication (e.g.:signing) starting with these type of certificates. See [RFC4492] for more information.
The reason for these choices are as follows: As of 2020, RSA is still more widely used than ECC, therefore the MUST for RSA. ECC offers equivalent security at (logarithmically) shorter key lengths (see [RFC4492]). This can be beneficial especially in the presence of constrained bandwidth or constrained nodes in an ACP/ANI network. Some ACP functions such as GRASP peer-2-peer across the ACP require end-to-end/any-to-any authentication/authorization, therefore ECC can only reliably be used in the ACP when it MUST be supported on all ACP nodes. RSA signatures are mandatory to be supported also for ECC certificates because CAs themselves may not support ECC yet.
For further certificate details, ACP certificates may follow the recommendations from [CABFORUM].
The ACP domain certificate SHOULD be used for any authentication between nodes with ACP domain certificates (ACP nodes and NOC nodes) where the required authorization condition is ACP domain membership, such as ACP node to NOC/OAM end-to-end security and ASA to ASA end-to-end security. Section 6.1.3 defines this "ACP domain membership check". The uses of this check that are standardized in this document are for the establishment of hop-by-hop ACP secure channels (Section 6.6) and for ACP GRASP (Section 6.8.2) end-to-end via TLS ([RFC5246]).
The ACP domain membership check requires a minimum amount of elements in a certificate as described in Section 6.1.3. All elements are [RFC5280] compliant. The identity of a node in the ACP is carried via the ACP Domain Information Field as defined in Section 6.1.2 which is encoded as an rfc822Name field.
Any other field of the ACP domain certificate is to be populated as required by [RFC5280] or desired by the operator of the ACP domain ACP registrars/CA and required by other purposes that the ACP domain certificate is intended to be used for.
For diagnostic and other operational purposes, it is beneficial to copy the device identifying fields of the node's IDevID into the ACP domain certificate, such as the "serialNumber" (see [I-D.ietf-anima-bootstrapping-keyinfra] section 2.3.1). This can be done for example if it would be acceptable for the devices "serialNumber" to be signalled via the Link Layer Discovery Protocol (LLDP, [LLDP]) because like LLDP signalled information, the ACP certificate information can be retrieved bei neighboring nodes without further authentication and be used either for beneficial diagnostics or for malicious attacks. Retrieval of the ACP certificate is possible via a (failing) attempt to set up an ACP secure channel, and the "serialNumber" contains usually device type information that may help to faster determine working exploits/attacks against the device.
Note that there is no intention to constrain authorization within the ACP or autonomic networks using the ACP to just the ACP domain membership check as defined in this document. It can be extended or modified with future requirements. Such future authorizations can use and require additional elements in certificates or policies or even additional certificates. For an example, see Appendix A.10.5.
Information about the domain MUST be encoded in the domain certificate in a subjectAltName / rfc822Name field according to the following ABNF ([RFC5234]) definition:
[RFC Editor: Please substitute SELF in all occurrences of rfcSELF in this document with the RFC number assigned to this document and remove this comment line]
domain-information = local-part "@" acp-domain-name
local-part = key [ "+" local-info ]
key = "rfcSELF"
local-info = [ acp-address ] [ "+" rsub extensions ]
HEXLC = DIGIT / "a" / "b" / "c" / "d" / "e" / "f"
; DIGIT as of RFC5234 section B.1
acp-address = 32HEXLC | "0"
rsub = [ <subdomain> ] ; <subdomain> as of RFC1034, section 3.5
acp-domain-name = ; <domain> ; as of RFC 1034, section 3.5
extensions = *( "+" extension )
extension = ; future standard definition.
; Must fit RFC5322 simple dot-atom format.
routing-subdomain = [ rsub "." ] acp-domain-name
Example:
given an ACP address of fd89:b714:f3db:0:200:0:6400:0000
and an ACP domain-name of acp.example.com
and an rsub extenstion of area51.research
then this results in:
domain-information = rfcSELF+fd89b714F3db00000200000064000000
+area51.research@acp.example.com
acp-domain-name = acp.example.com
routing-subdomain = area51.research.acp.example.com
Figure 2: ACP Domain Information Field ABNF
domain-information is the encoded information that is put into the ACP domain certificates subjectAltName / rfc822Name field. routing-subdomain is a string that can be constructed from the domain-information, and it is used in the hash-creation of the ULA (see below). The requirements and sementics of the parts of this information are explained in the following paragraphs:
Nodes complying with this specification MUST be able to receive their ACP address through the domain certificate, in which case their own ACP domain certificate MUST have the 32HEXDIG "acp-address" field. Nodes complying with this specification MUST also be able to authenticate nodes as ACP domain members or ACP secure channel peers when they have a 0-value acp-address field and as ACP domain members (but not as ACP secure channel peers) when they have an empty acp-address field. See Section 6.1.3.
"acp-domain-name" is used to indicate the ACP Domain across which all ACP nodes trust each other and are willing to build ACP channels to each other. See Section 6.1.3. Acp-domain-name SHOULD be the FQDN of a DNS domain owned by the operator assigning the certificate. This is a simple method to ensure that the domain is globally unique and collision of ACP addresses would therefore only happen due to ULA hash collisions (see Section 6.10.2). If the operator does not own any FQDN, it should choose a string (in FQDN format) that it intends to be equally unique.
To keep the encoding simple, there is no consideration for internationalized acp-domain-names. The ACP domain information is not intended for enduser consumption, and there is no protection against someone not owning a domain name to simpy choose it. Instead, it only serves as a hash seed for the ULA and for diagnostics to the operator. Therefore, any operator owning only an internationalized domain name should be able to pick an equivalently unique 7-bit ASCII acp-domain-name string representing the internationalized domain name.
"routing-subdomain" is a heuristic that allows a Registrar to consistently generate a unique 48-bit ULA prefix for ACP addresses. The presence of the "rsub" component allows to a single ACP domain to employ multiple /48 ULA prefixes. See Appendix A.7 for example use-cases.
The optional "extensions" field is used for future standardized extensions to this specification. It MUST be ignored if present and not understood.
Formatting notes:
The subjectAltName / rfc822Name encoding of the ACP domain name and ACP address is used for the following reasons:
Rfc822name encoding for the ACP domain information was choosen for the following reasons.
Compatibilities:
See section 4.2.1.6 of [RFC5280] for details on the subjectAltName field.
The following points constitute the ACP domain membership check of a candidate peer via its certificate:
When checking a candidate peer's certificate for the purpose of establishing an ACP secure channel, one additional check is performed:
Technically, ACP secure channels can only be built with nodes that have an acp-address. Rule 6 ensures that this is taken into account during ACP domain membership check.
Nodes with an empty acp-address field can only use their ACP domain certificate for non-ACP-secure channel authentication purposes. This includes for example NMS type nodes permitted to communicate into the ACP via ACP connect
The special value 0 in an ACP certificates acp-address field is used for nodes that can and should determine their ACP address through other mechanisms than learning it through the acp-address field in their ACP domain certificate. These ACP nodes are permitted to establish ACP secure channels. Mechanisms for those nodes to determine their ACP address are outside the scope of this specification, but this option is defined here so that any ACP nodes can build ACP secure channels to them according to Rule 6.
In summary:
For brevity, the remainder of this document refers to this process only as authentication instead of as authentication and authorization.
An ACP node with a realtime clock in which it has confidence, MUST check the time stamps when performing ACP domain membership check such as as the certificate validity period in step 1. and the respective times in step 4 for revocation information (e.g.: signingTimes in CMS signatures).
An ACP node without such a realtime clock MAY ignore those time stamp validation steps if it does not know the current time. Such an ACP node SHOULD obtain the current time in a secured fashion, such as via a Network Time Protocol signalled through the ACP. It then ignores time stamp validation only until the current time is known. In the absence of implementing a secured mechanism, such an ACP node MAY use a current time learned in an insecured fashion in the ACP domain membership check.
Beside ACP domain membership check, the ACP itself has no dependency against knowledge of the current time, but protocols and services using the ACP will likley have the need to know the current time. For example event logging.
ACP nodes need Trust Point (TP) certificates to perform certificate path validation as required by Section 6.1.3, rule 3. Trust Point(s) MUST be provisioned to an ACP node (together with its ACP domain certificate) by an ACP Registrar during initial enrolment of a candidate ACP node. ACP nodes MUST also support renewal of TPs via Enrollment over Secure Transport (EST, see [RFC7030]), as described below in Section 6.1.5.
Trust Point is the term used in this document for a CA and its associated set of certificates. Multiple certificates are required for a CA to deal with CA certificate renewals as explained in Section 4.4 of CMP ([RFC4210]).
A certificate path is a chain of certificates starting at the ACP certificate (leaf/end-entity) followed by zero or more intermediate Trust Point or sub-CA certificates and ending with a self-signed certificate of a so called root-CA or Trust Anchor. Certificate path validation authenticates that the ACP certificate is signed by a Trust Anchor, directly or indirectly via one or more intermediate Trust Points.
Note that different ACP nodes may have different Trust Points and even different Trust Anchors in their certificate path, as long as the set of Trust Points for all ACP node includes the same set of Trust Anchors (usually 1), and each ACP nodes set of Trust Anchors includes the intermediate Trust Points for its own ACP domain certificate. The protocols through which ACP domain membership check rules 1-4 are performed therefore need to support the exchange not only of the ACP nodes certificates, but also their intermediate Trust Points.
ACP nodes MUST support for the ACP domain membership check the certificate path validation with 0 or 1 intermediate Trust Points. They SHOULD support 2 intermediate Trust Points and two Trust Anchors (to permit migration to different root-CAs).
Trust Points for ACP domain certificates are trusted to sign certificates with valid ACP domain information fields only for trusted ACP registrars of that domain. This can be achieved by using Trust Anchors private to the owner of the ACP domain or potentially through appropriate contractual agreements between the involved parties. Public CA without such obligations and guarantees can not be used.
A single owner can operate multiple independent ACP domains from the same set of private trust anchors (CAs) when the ACP Registrars are trusted not to permit certificates with incorrect ACP information fields to be signed, such as ACP information with a wrong acp-domain field. In this case, CAs can be completely unaware of ACP specifics, so that it should be possible to use any existing CA software. When ACP Registrars are not to be trusted, the correctness of the ACP domain information field for the candidate ACP node has to be verified by the CA signing the ACP domain certificate.
ACP nodes MUST support renewal of their Certificate and TPs via EST ("Enrollment over Secure Transport", see [RFC7030]) and MAY support other mechanisms. An ACP network MUST have at least one ACP node supporting EST server functionality across the ACP so that EST renewal is useable.
ACP nodes SHOULD be able to remember the IPv6 locator parameters of the O_IPv6_LOCATOR in GRASP of the EST server from which they last renewed their ACP domain certificate. They SHOULD provide the ability for these EST server parameters to also be set by the ACP Registrar (see Section 6.10.7) that initially enrolled the ACP device with its ACP domain certificate. When BRSKI (see [I-D.ietf-anima-bootstrapping-keyinfra]) is used, the IPv6 locator of the BRSKI registrar from the BRSKI TLS connection SHOULD be remembered and used for the next renewal via EST if that registrar also announces itself as an EST server via GRASP (see next section) on its ACP address.
The EST server MUST present a certificate that is passing ACP domain membership check in its TLS connection setup (Section 6.1.3, rules 1..5, not rule 6 as this is not for an ACP secure channel setup). The EST server certificate MUST also contain the id-kp-cmcRA [RFC6402] extended key usage extension and the EST client MUST check its presence.
The additional check against the id-kp-cmcRA extended key usage extension field ensures that clients do not fall prey to an illicit EST server. While such illicit EST servers should not be able to support certificate signing requests (as they are not able to elicit a signing response from a valid CA), such an illicit EST server would be able to provide faked CA certificates to EST clients that need to renew their CA certificates when they expire.
Note that EST servers supporting multiple ACP domains will need to have for each of these ACP domains a separate certificate and respond on a different transport address (IPv6 address and/or TCP port), but this is easily automated on the EST server as long as the CA does not restrict registrars to request certificates with the id-kp-cmcRA extended usage extension for themselves.
ACP nodes that are EST servers MUST announce their service via GRASP in the ACP through M_FLOOD messages. See [I-D.ietf-anima-grasp], section 2.8.11 for the definition of this message type:
Example:
[M_FLOOD, 12340815, h'fd89b714f3db0000200000064000001', 210000,
["SRV.est", 4, 255 ],
[O_IPv6_LOCATOR,
h'fd89b714f3db0000200000064000001', TCP, 443]
]
Figure 3: GRASP SRV.est example
The formal definition of the objective in Concise data definition language (CDDL) (see [RFC8610]) is as follows:
flood-message = [M_FLOOD, session-id, initiator, ttl,
+[objective, (locator-option / [])]]
; see example above and explanation
; below for initiator and ttl
objective = ["SRV.est", objective-flags, loop-count,
objective-value]
objective-flags = sync-only ; as in GRASP spec
sync-only = 4 ; M_FLOOD only requires synchronization
loop-count = 255 ; recommended as there is no mechanism
; to discover network diameter.
objective-value = any ; reserved for future extensions
Figure 4: GRASP SRV.est definition
The objective name "SRV.est" indicates that the objective is an [RFC7030] compliant EST server because "est" is an [RFC6335] registered service name for [RFC7030]. Objective-value MUST be ignored if present. Backward compatible extensions to [RFC7030] MAY be indicated through objective-value. Non [RFC7030] compatible certificate renewal options MUST use a different objective-name. Non-recognized objective-values (or parts thereof if it is a structure partially understood) MUST be ignored.
The M_FLOOD message MUST be sent periodically. The default SHOULD be 60 seconds, the value SHOULD be operator configurable but SHOULD be not smaller than 60 seconds. The frequency of sending MUST be such that the aggregate amount of periodic M_FLOODs from all flooding sources cause only negligible traffic across the ACP. The time-to-live (ttl) parameter SHOULD be 3.5 times the period so that up to three consecutive messages can be dropped before considering an announcement expired. In the example above, the ttl is 210000 msec, 3.5 times 60 seconds. When a service announcer using these parameters unexpectedly dies immediately after sending the M_FLOOD, receivers would consider it expired 210 seconds later. When a receiver tries to connect to this dead service before this timeout, it will experience a failing connection and use that as an indication that the service instance is dead and select another instance of the same service instead (from another GRASP announcement).
When performing renewal, the node SHOULD attempt to connect to the remembered EST server. If that fails, it SHOULD attempt to connect to an EST server learned via GRASP. The server with which certificate renewal succeeds SHOULD be remembered for the next renewal.
Remembering the last renewal server and preferring it provides stickiness which can help diagnostics. It also provides some protection against off-path compromised ACP members announcing bogus information into GRASP.
Renewal of certificates SHOULD start after less than 50% of the domain certificate lifetime so that network operations has ample time to investigate and resolve any problems that causes a node to not renew its domain certificate in time - and to allow prolonged periods of running parts of a network disconnected from any CA.
The ACP node SHOULD support revocation through CRL(s) via HTTP from one or more CRL Distribution Points (CDPs). The CDP(s) MUST be indicated in the Domain Certificate when used. If the CDP URL uses an IPv6 address (ULA address when using the addressing rules specified in this document), the ACP node will connect to the CDP via the ACP. If the CDP uses a domain name, the ACP node will connect to the CDP via the Data-Plane.
It is common to use domain names for CDP(s), but there is no requirement for the ACP to support DNS. Any DNS lookup in the Data-Plane is not only a possible security issue, but it would also not indicate whether the resolved address is meant to be reachable across the ACP. Therefore, the use of an IPv6 address versus the use of a DNS name doubles as an indicator whether or not to reach the CDP via the ACP.
A CDP can be reachable across the ACP either by running it on a node with ACP or by connecting its node via an ACP connect interface (see Section 8.1). The CDP SHOULD use an ACP domain certificate for its HTTPs connections. The connecting ACP node SHOULD verify that the CDP certificate used during the HTTPs connection has the same ACP address as indicated in the CDP URL of the node's ACP domain certificate if the CDP URL uses an IPv6 address.
Certificate lifetime may be set to shorter lifetimes than customary (1 year) because certificate renewal is fully automated via ACP and EST. The primary limiting factor for shorter certificate lifetimes is load on the EST server(s) and CA. It is therefore recommended that ACP domain certificates are managed via a CA chain where the assigning CA has enough performance to manage short lived certificates. See also Section 9.2.4 for discussion about an example setup achieving this. See also [I-D.ietf-acme-star].
When certificate lifetimes are sufficiently short, such as few hours, certificate revocation may not be necessary, allowing to simplify the overall certificate maintenance infrastructure.
See Appendix A.2 for further optimizations of certificate maintenance when BRSKI can be used ("Bootstrapping Remote Secure Key Infrastructures", see [I-D.ietf-anima-bootstrapping-keyinfra]).
An ACP node may determine that its ACP domain certificate has expired, for example because the ACP node was powered down or disconnected longer than its certificate lifetime. In this case, the ACP node SHOULD convert to a role of a re-enrolling candidate ACP node.
In this role, the node does maintain the trust anchor and certificate chain associated with its ACP domain certificate exclusively for the purpose of re-enrollment, and attempts (or waits) to get re-enrolled with a new ACP certificate. The details depend on the mechanisms/protocols used by the ACP registrars.
Please refer to Section 6.10.7 and [I-D.ietf-anima-bootstrapping-keyinfra] for explanations about ACP registrars and vouchers as used in the following text. When ACP is intended to be used without BRSKI, the details about BRSKI and vouchers in the following text can be skipped.
When BRSKI is used (i.e.: on ACP nodes that are ANI nodes), the re-enrolling candidate ACP node would attempt to enroll like a candidate ACP node (BRSKI pledge), but instead of using the ACP nodes IDevID, it SHOULD first attempt to use its ACP domain certificate in the BRSKI TLS authentication. The BRSKI registrar MAY honor this certificate beyond its expiration date purely for the purpose of re-enrollment. Using the ACP node's domain certificate allows the BRSKI registrar to learn that node's ACP domain information field, so that the BRSKI registrar can re-assign the same ACP address information to the ACP node in the new ACP domain certificate.
If the BRSKI registrar denies the use of the old ACP domain certificate, the re-enrolling candidate ACP node MUST re-attempt re-enrollment using its IDevID as defined in BRSKI during the TLS connection setup.
Both when the BRSKI connection is attempted with the old ACP domain certificate or the IDevID, the re-enrolling candidate ACP node SHOULD authenticate the BRSKI registrar during TLS connection setup based on its existing trust anchor/certificate chain information associated with its old ACP certificate. The re-enrolling candidate ACP node SHOULD only fall back to requesting a voucher from the BRSKI registrar when this authentication fails during TLS connection setup.
When other mechanisms than BRSKI are used for ACP domain certificate enrollment, the principles of the re-enrolling candidate ACP node are the same. The re-enrolling candidate ACP node attempts to authenticate any ACP registrar peers during re-enrollment protocol/mechanisms via its existing certificate chain/trust anchor and provides its existing ACP domain certificate and other identification (such as the IDevID) as necessary to the registrar.
Maintaining existing trust anchor information is especially important when enrollment mechanisms are used that unlike BRSKI do not leverage a voucher mechanism to authenticate the ACP registrar and where therefore the injection of certificate failures could otherwise make the ACP node easily attackable remotely.
When using BRSKI or other protocol/mechanisms supporting vouchers, maintaining existing trust anchor information allows for re-enrollment of expired ACP certificates to be more lightweight, especially in environments where repeated acquisition of vouchers during the lifetime of ACP nodes may be operationally expensive or otherwise undesirable.
An ACP domain certificate is called failing in this document, if/when the ACP node to which the certificate was issued can determine that it was revoked (or explicitly not renewed), or in the absence of such explicit local diagnostics, when the ACP node fails to connect to other ACP nodes in the same ACP domain using its ACP certificate. For connection failures to determine the ACP domain certificate as the culprit, the peer should pass the domain membership check (Section 6.1.3) and other reasons for the connection failure can be excluded because of the connection error diagnostics.
This type of failure can happen during setup/refresh of a secure ACP channel connections or any other use of the ACP domain certificate, such as for the TLS connection to an EST server for the renewal of the ACP domain certificate.
Example reasons for failing certificates that the ACP node can only discover through connection failure are that the domain certificate or any of its signing certificates could have been revoked or may have expired, but the ACP node cannot self-diagnose this condition directly. Revocation information or clock synchronization may only be available across the ACP, but the ACP node cannot build ACP secure channels because ACP peers reject the ACP node's domain certificate.
ACP nodes SHOULD support the option to determines whether its ACP certificate is failing, and when it does, put itself into the role of a re-enrolling candidate ACP node as explained above (Section 6.1.5.5).
To know to which nodes to establish an ACP channel, every ACP node maintains an adjacency table. The adjacency table contains information about adjacent ACP nodes, at a minimum: Node-ID (identifier of the node inside the ACP, see Section 6.10.3 and Section 6.10.5), interface on which neighbor was discovered (by GRASP as explained below), link-local IPv6 address of neighbor on that interface, certificate (including domain information field). An ACP node MUST maintain this adjacency table. This table is used to determine to which neighbor an ACP connection is established.
Where the next ACP node is not directly adjacent (i.e., not on a link connected to this node), the information in the adjacency table can be supplemented by configuration. For example, the Node-ID and IP address could be configured. See Section 8.2.
The adjacency table MAY contain information about the validity and trust of the adjacent ACP node's certificate. However, subsequent steps MUST always start with the ACP domain membership check against the peer (see Section 6.1.3).
The adjacency table contains information about adjacent ACP nodes in general, independently of their domain and trust status. The next step determines to which of those ACP nodes an ACP connection should be established.
[RFC Editor: GRASP draft is in RFC editor queue, waiting for dependencies, including ACP. Please ensure that references to I-D.ietf-anima-grasp that include section number references (throughout this document) will be updated in case any last-minute changes in GRASP would make those section references change.
Discovery Unsolicited Link-Local (DULL) GRASP is a limited subset of GRASP intended to operate across an insecure link-local scope. See section 2.5.2 of [I-D.ietf-anima-grasp] for its formal definition. The ACP uses one instance of DULL GRASP for every L2 interface of the ACP node to discover link level adjacent candidate ACP neighbors. Unless modified by policy as noted earlier (Section 5 bullet point 2.), native interfaces (e.g., physical interfaces on physical nodes) SHOULD be initialized automatically to a state in which ACP discovery can be performed and any native interfaces with ACP neighbors can then be brought into the ACP even if the interface is otherwise not configured. Reception of packets on such otherwise not configured interfaces MUST be limited so that at first only IPv6 StateLess Address Auto Configuration (SLAAC - [RFC4862]) and DULL GRASP work and then only the following ACP secure channel setup packets - but not any other unnecessary traffic (e.g., no other link-local IPv6 transport stack responders for example).
Note that the use of the IPv6 link-local multicast address (ALL_GRASP_NEIGHBORS) implies the need to use Multicast Listener Discovery Version 2 (MLDv2, see [RFC3810]) to announce the desire to receive packets for that address. Otherwise DULL GRASP could fail to operate correctly in the presence of MLD snooping, non-ACP enabled L2 switches ([RFC4541]) - because those would stop forwarding DULL GRASP packets. Switches not supporting MLD snooping simply need to operate as pure L2 bridges for IPv6 multicast packets for DULL GRASP to work.
ACP discovery SHOULD NOT be enabled by default on non-native interfaces. In particular, ACP discovery MUST NOT run inside the ACP across ACP virtual interfaces. See Section 9.3 for further, non-normative suggestions on how to enable/disable ACP at node and interface level. See Section 8.2.2 for more details about tunnels (typical non-native interfaces). See Section 7 for how ACP should be extended on devices operating (also) as L2 bridges.
Note: If an ACP node also implements BRSKI to enroll its ACP domain certificate (see Appendix A.2 for a summary), then the above considerations also apply to GRASP discovery for BRSKI. Each DULL instance of GRASP set up for ACP is then also used for the discovery of a bootstrap proxy via BRSKI when the node does not have a domain certificate. Discovery of ACP neighbors happens only when the node does have the certificate. The node therefore never needs to discover both a bootstrap proxy and ACP neighbor at the same time.
An ACP node announces itself to potential ACP peers by use of the "AN_ACP" objective. This is a synchronization objective intended to be flooded on a single link using the GRASP Flood Synchronization (M_FLOOD) message. In accordance with the design of the Flood message, a locator consisting of a specific link-local IP address, IP protocol number and port number will be distributed with the flooded objective. An example of the message is informally:
[M_FLOOD, 12340815, h'fe80000000000000c0011001feef0000', 210000,
["AN_ACP", 4, 1, "IKEv2" ],
[O_IPv6_LOCATOR,
h'fe80000000000000c0011001feef0000', UDP, 15000]
["AN_ACP", 4, 1, "DTLS" ],
[O_IPv6_LOCATOR,
h'fe80000000000000c0011001feef0000', UDP, 17000]
]
Figure 5: GRASP AN_ACP example
The formal CDDL definition is:
flood-message = [M_FLOOD, session-id, initiator, ttl,
+[objective, (locator-option / [])]]
objective = ["AN_ACP", objective-flags, loop-count,
objective-value]
objective-flags = sync-only ; as in the GRASP specification
sync-only = 4 ; M_FLOOD only requires synchronization
loop-count = 1 ; limit to link-local operation
objective-value = method
method = "IKEv2" / "DTLS" ; or future standard methods
Figure 6: GRASP AN_ACP definition
The objective-flags field is set to indicate synchronization.
The loop-count is fixed at 1 since this is a link-local operation.
In the above example the RECOMMENDED period of sending of the objective is 60 seconds. The indicated ttl of 210000 msec means that the objective would be cached by ACP nodes even when two out of three messages are dropped in transit.
The session-id is a random number used for loop prevention (distinguishing a message from a prior instance of the same message). In DULL this field is irrelevant but has to be set according to the GRASP specification.
The originator MUST be the IPv6 link local address of the originating ACP node on the sending interface.
The 'objective-value' parameter is a string indicating the protocol available at the specified or implied locator. It is a protocol supported by the node to negotiate a secure channel. IKEv2 as shown above is the protocol used to negotiate an IPsec secure channel.
The locator-option is optional and only required when the secure channel protocol is not offered at a well-defined port number, or if there is no well-defined port number.
IKEv2 is the actual protocol used to negotiate an Internet Protocol security architecture (IPsec) connection. GRASP therefore indicates "IKEv2" and not "IPsec". If "IPsec" was used, this too could mean use of the obsolete older version IKE (v1) ([RFC2409]). IKEv2 has am IANA assigned port number 500, but in the above example, the candidate ACP neighbor is offering ACP secure channel negotiation via IKEv2 on port 15000 (purely to show through the example that GRASP allows to indicate the port number and it does not have to be the IANA assigned one).
"DTLS" indicates DTLS version 1.2. This can also be a newer version of the protocol as long as it can negotiate down to version 1.2 in the presence of a peer only speaking DTLS version 1.2. There is no default UDP port for DTLS, it is always locally assigned by the node. For details, see Section 6.7.4.
If a locator is included, it MUST be an O_IPv6_LOCATOR, and the IPv6 address MUST be the same as the initiator address (these are DULL requirements to minimize third party DoS attacks).
The secure channel methods defined in this document use the objective-values of "IKEv2" and "DTLS". There is no distinction between IKEv2 native and GRE-IKEv2 because this is purely negotiated via IKEv2.
A node that supports more than one secure channel protocol method needs to flood multiple versions of the "AN_ACP" objective so that each method can be accompanied by its own locator-option. This can use a single GRASP M_FLOOD message as shown in Figure 5.
Note that a node serving both as an ACP node and BRSKI Join Proxy may choose to distribute the "AN_ACP" objective and the respective BRSKI in the same M_FLOOD message, since GRASP allows multiple objectives in one message. This may be impractical though if ACP and BRSKI operations are implemented via separate software modules / ASAs.
The result of the discovery is the IPv6 link-local address of the neighbor as well as its supported secure channel protocols (and non-standard port they are running on). It is stored in the ACP Adjacency Table (see Section 6.2), which then drives the further building of the ACP to that neighbor.
Note that the DULL GRASP objective described does intentionally not include ACP nodes ACP domain certificate even though this would be useful for diagnostics and to simplify the security exchange in ACP secure channel security association protocols (see Section 6.7). The reason is that DULL GRASP messages are periodically multicasted across IPv6 subnets and full certificates could easily lead to fragmented IPv6 DULL GRASP multicast packets due to the size of a certificate. This would be highly undesirable.
An ACP node determines to which other ACP nodes in the adjacency table it should attempt to build an ACP connection. This is based on the information in the ACP Adjacency table.
The ACP is established exclusively between nodes in the same domain. This includes all routing subdomains. Appendix A.7 explains how ACP connections across multiple routing subdomains are special.
The result of the candidate ACP neighbor selection process is a list of adjacent or configured autonomic neighbors to which an ACP channel should be established. The next step begins that channel establishment.
To avoid attacks, initial discovery of candidate ACP peers cannot include any non-protected negotiation. To avoid re-inventing and validating security association mechanisms, the next step after discovering the address of a candidate neighbor can only be to try first to establish a security association with that neighbor using a well-known security association method.
From the use-cases it seems clear that not all type of ACP nodes can or need to connect directly to each other or are able to support or prefer all possible mechanisms. For example, code space limited IoT devices may only support DTLS because that code exists already on them for end-to-end security, but low-end in-ceiling L2 switches may only want to support Media Access Control Security (MacSec, see 802.1AE ([MACSEC]) because that is also supported in their chips. Only a flexible gateway device may need to support both of these mechanisms and potentially more. Note that MacSec is not required by any profiles of the ACP in this specification but just mentioned as a likely next interesting secure channel protocol.
To support extensible secure channel protocol selection without a single common mandatory to implement (MTI) protocol, ACP nodes MUST try all the ACP secure channel protocols it supports and that are feasible because the candidate ACP neighbor also announced them via its AN_ACP GRASP parameters (these are called the "feasible" ACP secure channel protocols).
To ensure that the selection of the secure channel protocols always succeeds in a predictable fashion without blocking, the following rules apply:
For example, originally Bob could have been the initiator of one ACP secure channel protocol that Bob prefers and the security association succeeded. The roles of Bob and Alice are then assigned and the connection setup is completed. The protocol could for example be IPsec via IKEv2 ("IP security", see [RFC4301] and "Internet Key Exchange protocol version 2", see [RFC7296]. It is now up to Alice to decide how to proceed. Even if the IPsec connection from Bob succeeded, Alice might prefer another secure protocol over IPsec (e.g., FOOBAR), and try to set that up with Bob. If that preference of Alice succeeds, she would close the IPsec connection. If no better protocol attempt succeeds, she would keep the IPsec connection.
[1] Node 1 sends GRASP AN_ACP message to announce itself
[2] Node 2 sends GRASP AN_ACP message to announce itself
[3] Node 2 receives [1] from Node 1
[4:C1] Because of [3], Node 2 starts as initiator on its
preferred secure channel protocol towards Node 1.
Connection C1.
[5] Node 1 receives [2] from Node 2
[6:C2] Because of [5], Node 1 starts as initiator on its
preferred secure channel protocol towards Node 2.
Connection C2.
[7:C1] Node1 and Node2 have authenticated each others
certificate on connection C1 as valid ACP peers.
[8:C1] Node 1 certificate has lower ACP Node-ID than Node2,
therefore Node 1 considers itself Bob and Node 2 Alice
on connection C1. Connection setup C1 is completed.
[9] Node 1 (Bob)) refrains from attempting any further secure
channel connections to Node 2 (Alice) as learned from [2]
because it knows from [8:C1] that it is Bob relative
to Node 1.
[10:C2] Node1 and Node2 have authenticated each others
certificate on connection C2 (like [7:C1]).
[11:C2] Node 1 certificate has lower ACP Node-ID than Node2,
therefore Node 1 considers itself Bob and Node 2 Alice
on connection C1, but they also identify that C2 is to the
same mutual peer as their C1, so this has no further impact.
[12:C2] Node 1 (Alice) closes C1. Because of [8:C1], Node 2 (Bob)
expected this.
[13] Node 1 (Alice) and Node 2 (Bob) start data transfer across
C2, which makes it become a secure channel for the ACP.
Figure 7: Secure Channel sequence of steps
The following sequence of steps show this example in more detail. Each step is tagged with [<step#>{:<connection>}]. The connection is included to easier distinguish which of the two competing connections the step belong to, one initiated by Node 1, one initiated by Node 2.
All this negotiation is in the context of an "L2 interface". Alice and Bob will build ACP connections to each other on every "L2 interface" that they both connect to. An autonomic node MUST NOT assume that neighbors with the same L2 or link-local IPv6 addresses on different L2 interfaces are the same node. This can only be determined after examining the certificate after a successful security association attempt.
Independent of the security association protocol chosen, candidate ACP neighbors need to be authenticated based on their domain certificate. This implies that any secure channel protocol MUST support certificate based authentication that can support the ACP domain membership check as defined in Section 6.1.3. If it fails, the connection attempt is aborted and an error logged. Attempts to reconnect MUST be throttled. The RECOMMENDED default is exponential base 2 backoff with a minimum delay of 10 seconds and a maximum delay of 640 seconds.
This section describes how ACP nodes establish secured data connections to automatically discovered or configured peers in the ACP. Section 6.3 above described how IPv6 subnet adjacent peers are discovered automatically. Section 8.2 describes how non IPv6 subnet adjacent peers can be configured.
Section 6.12.5.2 describes how secure channels are mapped to virtual IPv6 subnet interfaces in the ACP. The simple case is to map every ACP secure channel into a separate ACP point-to-point virtual interface Section 6.12.5.2.1. When a single subnet has multiple ACP peers this results in multiple ACP point-to-point virtual interfaces across that underlying multi-party IPv6 subnet. This can be optimized with ACP multi-access virtual interfaces Section 6.12.5.2.2 but the benefits of that optimization may not justify the complexity of that option.
Due to Channel Selection, ACP can support an evolving set of security association protocols. These protocols only need to be used to establish secure channels with L2 adjacent ACP neighbors and only optionally (where needed) across non-ACP capable L3 network (see Section 8.2). Therefore, there is architecturally no need for any network wide MTI security association protocols. Instead, ACP nodes only need to implement those protocols required to interoperate with their candidate peers, not with potentially any node in the ACP domain. See Section 6.7.5 for an example of this.
The degree of security required on every hop of an ACP network needs to be consistent across the network so that there is no designated "weakest link" because it is that "weakest link" that would otherwise become the designated point of attack. When the secure channel protection on one link is compromised, it can be used to send/receive packets across the whole ACP network. Therefore, even though the security association protocols can be different, their minimum degree of security should be comparable.
Secure channel protocols do not need to always support arbitrary L3 connectivity between peers, but can leverage the fact that the standard use case for ACP secure channels is an L2 adjacency. Hence, L2 mechanism dependent mechanisms could be adopted for use as secure channel association protocols:
L2 mechanisms such as strong encrypted radio technologies or [MACSEC] may offer equivalent encryption and the ACP security association protocol may only be required to authenticate ACP domain membership of a peer and/or derive a key for the L2 mechanism. Mechanisms to auto-discover and associate ACP peers leveraging such underlying L2 security are possible and desirable to avoid duplication of encryption, but none are specified in this document.
Strong physical security of a link may stand in where cryptographic security is infeasible. As there is no secure mechanism to automatically discover strong physical security solely between two peers, it can only be used with explicit configuration and that configuration too could become an attack vector. This document therefore only specifies with ACP connect one explicitly configured mechanism without any secure channel association protocol - for the case where both the link and the nodes attached to it have strong physical security.
The authentication of peers in any security association protocol MUST use the ACP domain certificate according to Section 6.1.3. Because auto-discovery of candidate ACP neighbors via GRASP (see Section 6.3) as specified in this document does not communicate the neighbors ACP domain certificate, and ACP nodes may not (yet) have any other network connectivity to retrieve certificates, any security association protocol MUST use a mechanism to communicate the certificate directly instead of relying on a referential mechanism such as communicating only a hash and/or URL for the certificate.
A security association protocol MUST use Forward Secrecy (whether inherently or as part of a profile of the security association protocol).
Because the ACP payload of legacy protocol payloads inside the ACP and hop-by-hop ACP flooded GRASP information is unencrypted, the ACP secure channel protocol requires confidentiality. Symmetric encryption for the transmission of secure channel data MUST use encryption schemes considered to be security wise equal to or better than AES256. There MUST NOT be support for NULL encryption.
An ACP secure channel MUST immediately be terminated when the lifetime of any certificate in the chain used to authenticate the neighbor expires or becomes revoked. This may not be standard behavior in secure channel protocols because the certificate authentication may only influences the setup of the secure channel in these protocols, but may not be re-validated during the lifetime of the secure connection in the absence of this requirement.
When introducing the profile for a security association protocol in support of the ACP, protocol options SHOULD be eliminated that do not provide benefits for devices that should be able to support the ACP. For example, definitions for security protocols often include old/inferior security options required only to interoperate with existing devices that will not be a able to update to the currently preferred security options. Such old/inferior security options do not need to be supported when a security association protocol is first specified for the ACP, strengthening the "weakest link" and simplifying ACP implementation overhead.
An ACP node announces its ability to support IPsec, negotiated via IKEv2, as the ACP secure channel protocol using the "IKEv2" objective-value in the "AN_ACP" GRASP objective.
The ACP usage of IPsec and IKEv2 mandates a narrow profile of the current standards-track usage guidance for IPsec [RFC8221] and IKEv2 [RFC8247]. This profile provides for stringent security properties and can exclude deprecated/legacy algorithms because there is no need for interoperability with legacy equipment for ACP secure channels. Any such backward compatibility would lead only to increased attack surface and implementation complexity, for no benefit.
An ACP node that is supporting native IPsec MUST use IPsec in tunnel mode, negotiated via IKEv2, and with IPv6 payload (e.g., ESP Next Header of 41). It MUST use local and peer link-local IPv6 addresses for encapsulation. Manual keying MUST NOT be used, see Section 6.1.
IPsec tunnel mode is required because the ACP will route/forward packets received from any other ACP node across the ACP secure channels, and not only its own generated ACP packets. With IPsec transport mode, it would only be possible to send packets originated by the ACP node itself.
ACP IPsec implementations MUST comply with [RFC8221] (and its updates). The requirements from above and this section amend and superceed its requirements.
AH MUST NOT be used (because it does not provide confidentiality).
For the required ESP encryption algorithms in section 5 of [RFC8221] the following guidance applies:
IKEv2 indicates an order for the offered algorithms. The algorithms SHOULD be ordered by performance. The first algorithm supported by both sides is generally choosen.
Explanations:
[RFC8221] allows for 128-bit or 256-bit AES keys. This document mandates that only 256-bit AES keys MUST be supported.
When [RFC8221] is updated, ACP implementations will need to consider legacy interoperability, and the IPsec WG has generally done a very good job of taking that into account in its recommendations.
[RFC8247] provides a baseline recommendation for mandatory to implement ciphers, integrity checks, pseudo-random-functions and Diffie-Hellman mechanisms. Those recommendations, and the recommendations of subsequent documents apply well to the ACP. Because IKEv2 for ACP secure channels is sufficient to be implemented in control plane software, rather than in ASIC hardware, and ACP nodes supporting IKEv2 are not assumed to be code-space constrained, and because existing IKEv2 implementations are expected to support [RFC8247] recommendations, this documents makes no attempt to simplify its recommendations for use with the ACP.
This document does establish a policy statement as permitted by [RFC8247] for the specific case of ACP traffic.
The IKEv2 Diffie-Hellman key exchange group 19 (256-bit random ECP), listed as a SHOULD, is to be configured, along with the 2048-bit MODP (group 14). ECC provides a similar security level to finite-field (MODP) key exchange with a shorter key length, so is generally preferred absent other considerations.
IKEv2 authentication MUST use the ACP domain certificates. The Certificate Encoding "PKCS #7 wrapped X.509 certificate" (1) MUST be supported. See [IKEV2IANA] for this and other IANA IKEv2 parameter names used in this text.
A certificate payload with the ACP certificate MUST be included during IKEv2 authentication to support the ACP domain membership check as described in Section 6.1.3, because it is using additional elements of the ACP certificates.
If certificate chains are used, all intermediate certificates up to, and including the locally provisioned trust anchor certificate MUST be signaled. See Section 6.10.7 for the sub-CA example in which certificate chains are used.
While the top-trust anchor will never be used by an ACP peer with proper provisioned ACP TAs, it can provide a significant amount of useful information to debug an ACP network. There is some small concern that this might provide useful information to an attacker about who the network is, but the other certificates that are sent already clearly indicate this information.
In IKEv2, ACP nodes are identified by their ACP address. The ID_IPv6_ADDR IKEv2 identification payload MUST be used and MUST convey the ACP address. If the peer's ACP domain certificate includes an ACP address in the ACP domain information field (not "0" or empty), the address in the IKEv2 identification payload MUST match it. See Section 6.1.3 for more information about "0" or empty ACP address fields in the ACP domain information field.
IKEv2 authentication MUST use authentication method 14 ("Digital Signature") for ACP certificates; this authentication method can be used with both RSA and ECDSA certificates, as indicated by a PKIX-style OID.
The Digital Signature hash SHA2-512 MUST be supported (in addition to SHA2-256).
In network devices it is often more common to implement high performance virtual interfaces on top of GRE encapsulation than on top of a "native" IPsec association (without any other encapsulation than those defined by IPsec). On those devices it may be beneficial to run the ACP secure channel on top of GRE protected by the IPsec association.
The requirements for ESP/IPsec/IKEv2 are the same as for native IPsec (see Section 6.7.3.1) except that IPsec transport mode and next protocol GRE (47) are to be negotiated. Tunnel mode is not required because of GRE.
If IKEv2 initiator and responder support IPsec over GRE, it has to be preferred over native IPsec. The ACP IPv6 traffic has to be carried across GRE according to [RFC7676].
We define the use of ACP via DTLS in the assumption that it is likely the first transport encryption supported in some classes of constrained devices because DTLS is already used in those devices but IPsec is not, and code-space may be limited.
An ACP node announces its ability to support DTLS v1.2 compliant with the requirements defined in this document as an ACP secure channel protocol in GRASP through the "DTLS" objective-value in the "AN_ACP" objective.
To run ACP via UDP and DTLS v1.2 [RFC6347], a locally assigned UDP port is used that is announced as a parameter in the GRASP AN_ACP objective to candidate neighbors. This port can also be any newer version of DTLS as long as that version can negotiate a DTLS v1.2 connection in the presence of an DTLS v1.2 only peer.
All ACP nodes supporting DTLS as a secure channel protocol MUST adhere to the DTLS implementation recommendations and security considerations of BCP 195 except with respect to the DTLS version. ACP nodes supporting DTLS MUST support DTLS 1.2. They MUST NOT support older versions of DTLS. Implementation MUST comply with BCP 195, [RFC7525].
Unlike for IPsec, no attempts are made to simplify the requirements of the BCP 195 recommendations because the expectation is that DTLS would be using software-only implementations where the ability to reuse of widely adopted implementations is more important than minizing the complexity of a hardware accelerated implementation which is known to be important for IPsec.
DTLS v1.3 ([I-D.ietf-tls-dtls13]) is "backward compatible" with DTLS v1.2 (see section 1. of DTLS v1.3): A DTLS implementation supporting both DTLS v1.2 and DTLS v1.3 does comply with the above requirements of negoting to DTLS v1.2 in the presence of a DTLS v1.2 only peer, but using DTLS v1.3 when booth peers support it.
Version v1.2 is the MTI version of DTLS in this specification because
Nevertheless, newer versions of DTLS, such as DTLS v1.3 have more strict security requirements and use of the latest standard protocol version is for IETF security standards in general recommended. Therefore, ACP implementations are advised to support all the newer versions of DTLS that can still negotiate down to DTLS v1.2.
[RFC-editor: if by the time of AUTH48, DTLS 1.3 would have evolved to be an RFC, then not only would the references to the DTLS v1.3 draft be changed to the RFC number, but that RFC is then going to be put into the normative list of references and the above paragraph is going to be amended to say: Implementations SHOULD support [DTLSv1.3-RFC]. This is not done right now, because there is no benefit in potentially waiting in RFC-editor queue for that RFC given how the text alreayd lays out a non-nrmative desire to support DTLSv1.3.]
There is no additional session setup or other security association besides this simple DTLS setup. As soon as the DTLS session is functional, the ACP peers will exchange ACP IPv6 packets as the payload of the DTLS transport connection. Any DTLS defined security association mechanisms such as re-keying are used as they would be for any transport application relying solely on DTLS.
As explained in the beginning of Section 6.5, there is no single secure channel mechanism mandated for all ACP nodes. Instead, this section defines two ACP profiles (baseline and constrained) for ACP nodes that do introduce such requirements.
A baseline ACP node MUST support IPsec natively and MAY support IPsec via GRE. A constrained ACP node that cannot support IPsec MUST support DTLS. An ACP node connecting an area of constrained ACP nodes with an area of baseline ACP nodes needs to support IPsec and DTLS and supports therefore the baseline and constrained profile.
Explanation: Not all type of ACP nodes can or need to connect directly to each other or are able to support or prefer all possible secure channel mechanisms. For example, code space limited IoT devices may only support DTLS because that code exists already on them for end-to-end security, but high-end core routers may not want to support DTLS because they can perform IPsec in accelerated hardware but would need to support DTLS in an underpowered CPU forwarding path shared with critical control plane operations. This is not a deployment issue for a single ACP across these type of nodes as long as there are also appropriate gateway ACP nodes that support sufficiently many secure channel mechanisms to allow interconnecting areas of ACP nodes with a more constrained set of secure channel protocols. On the edge between IoT areas and high-end core networks, general-purpose routers that act as those gateways and that can support a variety of secure channel protocols is the norm already.
ACP nodes need to specify in documentation the set of secure ACP mechanisms they support and should declare which profile they support according to above requirements.
The ACP MUST run an instance of GRASP inside of it. It is a key part of the ACP services. The function in GRASP that makes it fundamental as a service of the ACP is the ability to provide ACP wide service discovery (using objectives in GRASP).
ACP provides IP unicast routing via the RPL routing protocol (see Section 6.11).
The ACP does not use IP multicast routing nor does it provide generic IP multicast services (the handling of GRASP link-local multicast messages is explained in Section 6.8.2). Instead, the ACP provides service discovery via the objective discovery/announcement and negotiation mechanisms of the ACP GRASP instance (services are a form of objectives). These mechanisms use hop-by-hop reliable flooding of GRASP messages for both service discovery (GRASP M_DISCOVERY messages) and service announcement (GRASP M_FLOOD messages).
See Appendix A.5 for discussion about this design choice of the ACP.
In the terminology of GRASP ([I-D.ietf-anima-grasp]), the ACP is the security and transport substrate for the GRASP instance run inside the ACP ("ACP GRASP").
This means that the ACP is responsible for ensuring that this instance of GRASP is only sending messages across the ACP GRASP virtual interfaces. Whenever the ACP adds or deletes such an interface because of new ACP secure channels or loss thereof, the ACP needs to indicate this to the ACP instance of GRASP. The ACP exists also in the absence of any active ACP neighbors. It is created when the node has a domain certificate, and continues to exist even if all of its neighbors cease operation.
In this case ASAs using GRASP running on the same node would still need to be able to discover each other's objectives. When the ACP does not exist, ASAs leveraging the ACP instance of GRASP via APIs MUST still be able to operate, and MUST be able to understand that there is no ACP and that therefore the ACP instance of GRASP cannot operate.
The following explanation how ACP acts as the security and transport substrate for GRASP is visualized in Figure 8 below.
GRASP unicast messages inside the ACP always use the ACP address. Link-local addresses from the ACP VRF MUST NOT be used inside objectives. GRASP unicast messages inside the ACP are transported via TLS which MUST comply with [RFC7525] execept that only TLS version 1.2 ([RFC5246]) is REQUIRED and TLS 1.3 ([RFC8446] is RECOMMENDED. There is no need for older version backward compatibility in the new use-case of ACP. Mutual authentication MUST use the ACP domain membership check defined in (Section 6.1.3).
GRASP link-local multicast messages are targeted for a specific ACP virtual interface (as defined Section 6.12.5) but are sent by the ACP into an ACP GRASP virtual interface that is constructed from the TCP connection(s) to the IPv6 link-local neighbor address(es) on the underlying ACP virtual interface. If the ACP GRASP virtual interface has two or more neighbors, the GRASP link-local multicast messages are replicated to all neighbor TCP connections.
TLS for GRASP MUST offer TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384 and TLS_ECDHE_ECDSA_WITH_AES_256_GCM_SHA384 and MUST NOT offer options with less than 256bit AES or less than SHA384. TLS for GRASP MUST also include the "Supported Elliptic Curves" extension, it MUST support support the NIST P-256 (secp256r1) and P-384 (secp384r1(24)) curves [RFC4492]. In addition, GRASP TLS clients SHOULD send an ec_point_formats extension with a single element, "uncompressed". For further interoperability recommendations, GRASP TLS implementations SHOULD follow [RFC7525].
TCP and TLS connections for GRASP in the ACP use the IANA assigned TCP port for GRASP (7107). Effectively the transport stack is expected to be TLS for connections from/to the ACP address (e.g., global scope address(es)) and TCP for connections from/to link-local addresses on the ACP virtual interfaces. The latter ones are only used for flooding of GRASP messages.
..............................ACP..............................
. .
. /-GRASP-flooding-\ ACP GRASP instance .
. / \ A
. GRASP GRASP GRASP C
. link-local unicast link-local P
. multicast messages multicast .
. messages | messages .
. | | | .
...............................................................
. v v v ACP security and transport .
. | | | substrate for GRASP .
. | | | .
. | ACP GRASP | - ACP GRASP A
. | Loopback | Loopback interface C
. | interface | - ACP-cert auth P
. | TLS | .
. ACP GRASP | ACP GRASP - ACP GRASP virtual .
. subnet1 | subnet2 virtual interfaces .
. TCP | TCP .
. | | | .
...............................................................
. | | | ^^^ Users of ACP (GRASP/ASA) .
. | | | ACP interfaces/addressing .
. | | | .
. | | | A
. | ACP-Loopback Interf.| <- ACP Loopback interface C
. | ACP-address | - address (global ULA) P
. subnet1 | subnet2 <- ACP virtual interfaces .
. link-local | link-local - link-local addresses .
...............................................................
. | | | ACP VRF .
. | RPL-routing | virtual routing and forwarding .
. | /IP-Forwarding\ | A
. | / \ | C
. ACP IPv6 packets ACP IPv6 packets P
. |/ \| .
. IPsec/DTLS IPsec/DTLS - ACP-cert auth .
...............................................................
| | Data-Plane
| |
| | - ACP secure channel
link-local link-local - encapsulation addresses
subnet1 subnet2 - Data-Plane interfaces
| |
ACP-Nbr1 ACP-Nbr2
Figure 8: ACP as security and transport substrate for GRASP
TCP encapsulation for GRASP M_DISCOVERY and M_FLOOD link local messages is used because these messages are flooded across potentially many hops to all ACP nodes and a single link with even temporary packet loss issues (e.g., WiFi/Powerline link) can reduce the probability for loss free transmission so much that applications would want to increase the frequency with which they send these messages. Such shorter periodic retransmission of datagrams would result in more traffic and processing overhead in the ACP than the hop-by-hop reliable retransmission mechanism by TCP and duplicate elimination by GRASP.
TLS is mandated for GRASP non-link-local unicast because the ACP secure channel mandatory authentication and encryption protects only against attacks from the outside but not against attacks from the inside: Compromised ACP members that have (not yet) been detected and removed (e.g., via domain certificate revocation / expiry).
If GRASP peer connections were to use just TCP, compromised ACP members could simply eavesdrop passively on GRASP peer connections for whom they are on-path ("Man In The Middle" - MITM) or intercept and modify them. With TLS, it is not possible to completely eliminate problems with compromised ACP members, but attacks are a lot more complex:
Eavesdropping/spoofing by a compromised ACP node is still possible because in the model of the ACP and GRASP, the provider and consumer of an objective have initially no unique information (such as an identity) about the other side which would allow them to distinguish a benevolent from a compromised peer. The compromised ACP node would simply announce the objective as well, potentially filter the original objective in GRASP when it is a MITM and act as an application level proxy. This of course requires that the compromised ACP node understand the semantics of the GRASP negotiation to an extent that allows it to proxy it without being detected, but in an ACP environment this is quite likely public knowledge or even standardized.
The GRASP TLS connections are run the same as any other ACP traffic through the ACP secure channels. This leads to double authentication/encryption, which has the following benefits:
The ACP is in a separate context from the normal Data-Plane of the node. This context includes the ACP channels' IPv6 forwarding and routing as well as any required higher layer ACP functions.
In classical network system, a dedicated VRF is one logical implementation option for the ACP. If possible by the systems software architecture, separation options that minimize shared components are preferred, such as a logical container or virtual machine instance. The context for the ACP needs to be established automatically during bootstrap of a node. As much as possible it should be protected from being modified unintentionally by ("Data-Plane") configuration.
Context separation improves security, because the ACP is not reachable from the Data-Plane routing or forwarding table(s). Also, configuration errors from the Data-Plane setup do not affect the ACP.
The channels explained above typically only establish communication between two adjacent nodes. In order for communication to happen across multiple hops, the autonomic control plane requires ACP network wide valid addresses and routing. Each ACP node creates a Loopback interface with an ACP network wide unique address inside the ACP context (as explained in in Section 6.9). This address may be used also in other virtual contexts.
With the algorithm introduced here, all ACP nodes in the same routing subdomain have the same /48 ULA prefix. Conversely, ULA global IDs from different domains are unlikely to clash, such that two ACP networks can be merged, as long as the policy allows that merge. See also Section 10.1 for a discussion on merging domains.
Links inside the ACP only use link-local IPv6 addressing, such that each node's ACP only requires one routable virtual address.
The ACP is based exclusively on IPv6 addressing, for a variety of reasons:
The Base ULA addressing scheme for ACP nodes has the following format:
8 40 2 78
+--+-------------------------+------+------------------------------+
|fd| hash(routing-subdomain) | Type | (sub-scheme) |
+--+-------------------------+------+------------------------------+
Figure 9: ACP Addressing Base Scheme
The first 48-bits follow the ULA scheme, as defined in [RFC4193], to which a type field is added:
The sub-scheme may imply a range or set of addresses assigned to the node, this is called the ACP address range/set and explained in each sub-scheme.
Please refer to Section 6.10.7 and Appendix A.1 for further explanations why the following Sub-Addressing schemes are used and why multiple are necessary.
+------+-----+----------------+-------+------------+ | Type | Z | name | F-bit | V-bit size | +------+-----+----------------+-------+------------+ | 0x00 | 0 | ACP Zone | N/A | 1 bit | +------+-----+----------------+-------+------------+ | 0x00 | 1 | ACP Manual | N/A | 1 bit | +------+-----+----------------+-------+------------+ | 0x01 | N/A | VLong-ASA | 0 | 8-bits | +------+-----+----------------+-------+------------+ | 0x01 | N/A | VLong-ACP-edge | 1 | 16-bits | +------+-----+----------------+-------+------------+
Figure 10: Addressing schemes
The following summarizes the addressing Sub-Schemes:
This sub-scheme is used when the Type field of the base scheme is 0x00 and the Z bit is 0x0.
64 64
+-----------------+---+---------++-----------------------------+---+
| (base scheme) | Z | Zone-ID || Node-ID |
| | | || Registrar-ID | Node-Number| V |
+-----------------+---+---------++--------------+--------------+---+
50 1 13 48 15 1
Figure 11: ACP Zone Addressing Sub-Scheme
The fields are defined as follows:
The 64-bit Node-ID is derived and composed as follows:
In the ACP Zone Addressing Sub-Scheme, the ACP address in the certificate has Zone-ID and V fields as all zero bits. The ACP address set includes addresses with any Zone-ID value and any V value.
The "Node-ID" itself is unique in a domain (i.e., the Zone-ID is not required for uniqueness). Therefore, a node can be addressed either as part of a flat hierarchy (Zone-ID = 0), or with an aggregation scheme (any other Zone-ID). An address with Zone-ID = 0 is an identifier, with a Zone-ID !=0 it is a locator. See Section 6.10.3.1 for more details.
The Virtual bit in this sub-scheme allows the easy addition of the ACP as a component to existing systems without causing problems in the port number space between the services in the ACP and the existing system. V:0 is the ACP router (autonomic node base system), V:1 is the host with pre-existing transport endpoints on it that could collide with the transport endpoints used by the ACP router. The ACP host could for example have a p2p virtual interface with the V:0 address as its router into the ACP. Depending on the software design of ASAs, which is outside the scope of this specification, they may use the V:0 or V:1 address.
The location of the V bit(s) at the end of the address allows the announcement of a single prefix for each ACP node. For example, in a network with 20,000 ACP nodes, this avoid 20,000 additional routes in the routing table.
The Zone-ID allows for the introduction of route prefixes in the addressing scheme.
Zone-ID = 0 is the default addressing scheme in an ACP domain. Every ACP node with a Zone Addressing Sub-Scheme address MUST respond to its ACP address with Zone-ID = 0. Used on its own this leads to a non-hierarchical address scheme, which is suitable for networks up to a certain size. Zone-ID = 0 addresses act as identifiers for the nodes, and aggregation of these address in the ACP routing table is not possible.
If aggregation is required, the 13-bit Zone-ID value allows for up to 8191 zones. The allocation of Zone-ID's may either happen automatically through a to-be-defined algorithm; or it could be configured and maintained explicitly.
If a node learns (see Appendix A.10.1) that it is part of a zone, it MUST also respond to its ACP address with that Zone-ID. In this case the ACP Loopback is configured with two ACP addresses: One for Zone-ID = 0 and one for the assigned Zone-ID. This method allows for a smooth transition between a flat addressing scheme and a hierarchical one.
A node knowing it is in a zone MUST use that Zone-ID != 0 address in GRASP locator fields. This eliminates the use of the identifier address (Zone-ID = 0) in forwarding and the need for network wide reachability of those non-aggregable identifier addresses. Zone-ID != 0 addresses are assumed to be aggregable in routing/forwarding based on how they are allocated in the ACP topology.
Note: The Zone-ID is one method to introduce structure or hierarchy into the ACP. Another way is the use of the routing subdomain field in the ACP that leads to multiple /48 Global IDs within an ACP domain.
Note: Zones and Zone-ID as defined here are not related to [RFC4007] zones or zone_id. ACP zone addresses are not scoped (reachable only from within an RFC4007 zone) but reachable across the whole ACP. An RFC4007 zone_id is a zone index that has only local significance on a node, whereas an ACP Zone-ID is an identifier for an ACP zone that is unique across that ACP.
This sub-scheme is used when the Type field of the base scheme is 0x00 and the Z bit is 0x1.
64 64
+---------------------+---+----------++-----------------------------+
| (base scheme) | Z | Subnet-ID|| Interface Identifier |
+---------------------+---+----------++-----------------------------+
50 1 13
Figure 12: ACP Manual Addressing Sub-Scheme
The fields are defined as follows:
This sub-scheme is meant for "manual" allocation to subnets where the other addressing schemes cannot be used. The primary use case is for assignment to ACP connect subnets (see Section 8.1.1).
"Manual" means that allocations of the Subnet-ID need to be done today with pre-existing, non-autonomic mechanisms. Every subnet that uses this addressing sub-scheme needs to use a unique Subnet-ID (unless some anycast setup is done).
The Z bit field was added to distinguish Zone addressing and manual addressing sub-schemes without requiring one more bit in the base scheme and therefore allowing for the Vlong scheme (described below) to have one more bit available.
Manual addressing sub-scheme addresses SHOULD NOT be used in ACP domain certificates. Any node capable to build ACP secure channels and permitted by Registrar policy to participate in building ACP secure channels SHOULD receive an ACP address (prefix) from one of the other ACP addressing sub-schemes. Nodes not capable (or permitted) to participate in ACP secure channels can connect to the ACP via ACP connect interfaces of ACP edge nodes (see Section 8.1), without setting up an ACP secure channel. Their ACP domain certificate MUST include an empty acp-address to indicate that their ACP domain certificate is only usable for non- ACP secure channel authentication, such as end-to-end transport connections across the ACP or Data-Plane.
Address management of ACP connect subnets is done using traditional assignment methods and existing IPv6 protocols. See Section 8.1.3 for details.
This sub-scheme is used when the Type field of the base scheme is 0x01.
50 78
+---------------------++-----------------------------+----------+
| (base scheme) || Node-ID |
| || Registrar-ID |F| Node-Number| V |
+---------------------++--------------+--------------+----------+
50 46 1 23/15 8/16
Figure 13: ACP Vlong Addressing Sub-Scheme
This addressing scheme foregoes the Zone-ID field to allow for larger, flatter routed networks (e.g., as in IoT) with 8421376 Node-Numbers (2^23+2^15). It also allows for up to 2^16 (i.e. 65536) different virtualized addresses within a node, which could be used to address individual software components in an ACP node.
The fields are the same as in the Zone-ID sub-scheme with the following refinements:
The F=0 bit format addresses are intended to be used for "general purpose" ACP nodes that would potentially have a limited number (< 256) of clients (ASA/Autonomic Functions or legacy services) of the ACP that require separate V(irtual) addresses.
The F=1 bit Node-Numbers are intended for ACP nodes that are ACP edge nodes (see Section 8.1.1) or that have a large number of clients requiring separate V(irtual) addresses. For example large SDN controllers with container modular software architecture (see Section 8.1.2).
In the Vlong addressing sub-scheme, the ACP address in the certificate has all V field bits as zero. The ACP address set for the node includes any V value.
Before further addressing sub-schemes are defined, experience with the schemes defined here should be collected. The schemes defined in this document have been devised to allow hopefully sufficiently flexible setup of ACPs for a variety of situation. These reasons also lead to the fairly liberal use of address space: The Zone Addressing Sub-Scheme is intended to enable optimized routing in large networks by reserving bits for Zone-ID's. The Vlong addressing sub-scheme enables the allocation of 8/16-bit of addresses inside individual ACP nodes. Both address spaces allow distributed, uncoordinated allocation of node addresses by reserving bits for the registrar-ID field in the address.
IANA is asked need to assign a new "type" for each new addressing sub-scheme. With the current allocations, only 2 more schemes are possible, so the last addressing scheme MUST provide further extensions (e.g., by reserving bits from it for further extensions).
ACP registrars are responsible to enroll candidate ACP nodes with ACP domain certificates and associated trust point(s). They are also responsible that an ACP domain information field is included in the ACP domain certificate carrying the ACP domain name and the ACP nodes ACP address prefix. This address prefix is intended to persist unchanged through the lifetime of the ACP node.
Because of the ACP addressing sub-schemes, an ACP domain can have multiple distributed ACP registrars that do not need to coordinate for address assignment. ACP registrars can also be sub-CAs, in which case they can also assign ACP domain certificates without dependencies against a (shared) root-CA (except during renewals of their own certificates).
ACP registrars are PKI registration authorities (RA) enhanced with the handling of the ACP domain certificate specific fields. They request certificates for ACP nodes from a Certificate Authority through any appropriate mechanism (out of scope in this document, but required to be BRSKI for ANI registrars). Only nodes that are trusted to be compliant with the requirements against registrar described in this section can be given the necessary credentials to perform this RA function, such as credentials for the BRSKI connection to the CA for ANI registrars.
Any protocols or mechanisms may be used as ACP registrars, as long as the resulting ACP certificate and trust anchors allow to perform the ACP domain membership described in Section 6.1.3 with other ACP domain members, and meet the ACP addressing requirements for its ACP domain information field as described further below in this section.
An ACP registrar could be a person deciding whether to enroll a candidate ACP node and then orchestrating the enrollment of the ACP certificate and associated trust anchor, using command line or web based commands on the candidate ACP node and trust anchor to generate and sign the ACP domain certificate and configure certificate and trust anchors onto the node.
The only currently defined protocol for ACP registrars is BRSKI ([I-D.ietf-anima-bootstrapping-keyinfra]). When BRSKI is used, the ACP nodes are called ANI nodes, and the ACP registrars are called BRSKI or ANI registrars. The BRSKI specification does not define the handling of the ACP domain information field because the rules do not depend on BRSKI but apply equally to any protocols/mechanisms an ACP registrar may use.
ACP registrars MUST NOT allocate ACP address prefixes to ACP nodes via the ACP domain information field that would collide with the ACP address prefixes of other ACP nodes in the same ACP domain. This includes both prefixes allocated by the same ACP registrar to different ACP nodes as well as prefixes allocated by other ACP registrars for the same ACP domain.
To support such unique address allocation, an ACP registrar MUST have one or more 46-bit identifiers unique across the ACP domain which is called the Registrar-ID. Allocation of Registrar-ID(s) to an ACP registrar can happen through OAM mechanisms in conjunction with some database / allocation orchestration.
ACP registrars running on physical devices with known globally unique EUI-48 MAC address(es) can use the lower 46 bits of those address(es) as unique Registrar-IDs without requiring any external signaling/configuration (the upper two bits, V and U are not uniquely assigned but functional). This approach is attractive for distributed, non-centrally administered, lightweight ACP registrar implementations. There is no mechanism to deduce from a MAC address itself whether it is actually uniquely assigned. Implementations need to consult additional offline information before making this assumption. For example by knowing that a particular physical product/MIC-chip is guaranteed to use globally unique assigned EUI-48 MAC address(es).
When the candidate ACP device (called Pledge in BRSKI) is to be enrolled into an ACP domain, the ACP registrar needs to allocate a unique ACP address to the node and ensure that the ACP certificate gets a domain information field (Section 6.1.2) with the appropriate information - ACP domain-name, ACP-address, and so on. If the ACP registrar uses BRSKI, it signals the ACP domain information field to the Pledge via the EST /csrattrs command (see [I-D.ietf-anima-bootstrapping-keyinfra], section 5.9.2 - "EST CSR Attributes").
[RFC Editor: please update reference to section 5.9.2 accordingly with latest BRSKI draft at time of publishing, or RFC]
The ACP registrar selects for the candidate ACP node a unique address prefix from an appropriate ACP addressing sub-scheme, either a zone addressing sub-scheme prefix (see Section 6.10.3), or a Vlong addressing sub-scheme prefix (see Section 6.10.5). The assigned ACP address prefix encoded in the domain information field of the ACP domain certificate indicates to the ACP node its ACP address information. The sub-addressing scheme indicates the prefix length: /127 for zone address sub-scheme, /120 or /112 for Vlong address sub-scheme. The first address of the prefix is the ACP address. All other addresses in the prefix are for other uses by the ACP node as described in the zone and Vlong addressing sub scheme sections. The ACP address prefix itself is then signaled by the ACP node into the ACP routing protocol (see Section 6.11) to establish IPv6 reachability across the ACP.
The choice of addressing sub-scheme and prefix-length in the Vlong address sub-scheme is subject to ACP registrar policy. It could be an ACP domain wide policy, or a per ACP node or per ACP node type policy. For example, in BRSKI, the ACP registrar is aware of the IDevID of the candidate ACP node, which contains a "serialNnumber" that is typically indicating the node's vendor and device type and can be used to drive a policy selecting an appropriate addressing sub-scheme for the (class of) node(s).
ACP registrars SHOULD default to allocate ACP zone sub-address scheme addresses with Zone-ID 0. Allocation and use of zone sub-addresses with Zone-ID != 0 is outside the scope of this specification because it would need to go along with rules for extending ACP routing to multiple zones, which is outside the scope of this specification.
ACP registrars that are aware of can use the IDevID of a candidate ACP device SHOULD be able to choose the zone vs. Vlong sub-address scheme for ACP nodes based on the "serialNumber" of the IDevID, for example by the PID (Product Identifier) part which identifies the product type, or the complete "serialNumber".
In a simple allocation scheme, an ACP registrar remembers persistently across reboots its currently used Registrar-ID and for each addressing scheme (zone with Zone-ID 0, Vlong with /112, Vlong with /120), the next Node-Number available for allocation and increases it during successful enrollment to an ACP node. In this simple allocation scheme, the ACP registrar would not recycle ACP address prefixes from no longer used ACP nodes.
When an ACP domain certificate is renewed or rekeyed via EST or other mechanisms, the ACP address/prefix in the ACP domain information field MUST be maintained unless security issues or violations of the unique address assignment requirements exist or are suspected by the ACP registrar.
ACP address information SHOULD be maintained even when the renewing/rekeying ACP registrar is not the same as the one that enrolled the prior ACP certificate. See Section 9.2.4 for an example.
ACP address information SHOULD also be maintained even after an ACP certificate did expire or failed. See Section 6.1.5.5 and Section 6.1.5.6.
Section 9.2 discusses further informative details of ACP registrars: What interactions registrars need, what parameters they require, certificate renewal and limitations, use of sub-CAs on registrars and centralized policy control.
Once ULA address are set up all autonomic entities should run a routing protocol within the autonomic control plane context. This routing protocol distributes the ULA created in the previous section for reachability. The use of the autonomic control plane specific context eliminates the probable clash with Data-Plane routing tables and also secures the ACP from interference from the configuration mismatch or incorrect routing updates.
The establishment of the routing plane and its parameters are automatic and strictly within the confines of the autonomic control plane. Therefore, no explicit configuration is required.
All routing updates are automatically secured in transit as the channels of the ACP are encrypted, and this routing runs only inside the ACP.
The routing protocol inside the ACP is RPL ([RFC6550]). See Appendix A.4 for more details on the choice of RPL.
RPL adjacencies are set up across all ACP channels in the same domain including all its routing subdomains. See Appendix A.7 for more details.
The following is a description of the RPL profile that ACP nodes need to support by default. The format of this section is derived from draft-ietf-roll-applicability-template.
The choosen RPL profile is one that expects a fairly reliable network with reasonably fast links so that RPL convergence will be triggered immediately upon recognition of link failure/recovery.
The profile is also designed to not require any RPL Data-Plane artifacts (such as defined in [RFC6553]). This is largely driven by the desire to avoid introducing the required Hop-by-Hop headers into the ACP forwarding plane, especially to support devices with silicon forwarding planes that cannot support insertion/removal of these headers in silicon or hop-by-hop forwarding based on them. Note: Insertion/removal of headers by a (potentially silicon based) ACP node would be be necessary when senders/receivers of ACP packets are legacy NOC devices connected via ACP connect (see Section 8.1.1 to the ACP. Their connectivity can be handled in RPL as non-RPL-aware leafs (or "Internet") according to the Data-Plane architecture explained in [I-D.ietf-roll-useofrplinfo].
This RPL profile avoids the use of Data-Plane artefacts (RPL data packet headers, see Section 6.11.1.13), because hardware accelerated forwarding planes most likely can not support them today. To achieve this, the profile uses a simple destination prefix based routing/forwarding table. To achieve this, the profiles uses only one RPL instanceID. This single instanceID can contain only one Destination Oriented Directed Acyclic Graph (DODAG), and the routing/forwarding table can therefore only calculate a single class of service ("best effort towards the primary NOC/root") and cannot create optimized routing paths to accomplish latency or energy goals between any two nodes.
Consider a network that has multiple NOCs in different locations. Only one NOC will become the DODAG root. Traffic to and from other NOCs has to be sent through the DODAG (shortest path tree) rooted in the primary NOC. Depending on topology, this can be an annoyance from a latency point of view or from minimizing network path resources, but this is deemed to be acceptable given how ACP traffic is "only" network management/control traffic.
Using a single instanceID/DODAG does not introduce a single point of failure, as the DODAG will reconfigure itself when it detects data-plane forwarding failures including choosing a different root when the primary one fails. See Appendix A.10.4 for more details.
The benefit of this profile, especially compared to other IGPs is that it does not calculate routes for node reachable through the same interface as the DODAG root. This RPL profile can therefore scale to much larger number of ACP nodes in the same amount of compute and memory than other routing protocols. Especially on nodes that are leafs of the topology or those close to those leafs.
The lack of RPL Packet Information (see Section 6.11.1.13), means that the Data-Plane will have no rank value that can be used to detect loops. As a result, traffic may loop until the time-to-live (TTL) of the packet reaches zero. This is the same behavior as that of other IGPs that do not have the Data-Plane options of RPL.
Since links in the ACP are assumed to be mostly reliable (or have link layer protection against loss) and because there is no stretch according to Section 6.11.1.7, loops caused by RPL routing packet loss should be exceedingly rare.
There are a variety of mechanisms possible in RPL to further avoid temporary loops: DODAG Information Objects (DIOs) SHOULD be sent 2...3 times to inform children when losing the last parent. The technique in [RFC6550] section 8.2.2.6. (Detaching) SHOULD be favored over that in section 8.2.2.5., (Poisoning) because it allows local connectivity. Nodes SHOULD select more than one parent, at least 3 if possible, and send Destination Advertisement Objects (DAO)s to all of them in parallel.
Additionally, failed ACP tunnels can be quickly discovered trough the secure channel protocol mechanisms such as IKEv2 Dead Peer Detection. This can function as a replacement for a Low-power and Lossy Networks' (LLN's) Expected Transmission Count (ETX) feature that is not used in this profile. A failure of an ACP tunnel should imediately signal the RPL control plane to pick a different parent.
Single RPL instance. Default RPLInstanceID = 0.
RPL Mode of Operations (MOP): MUST support mode 2 - "Storing Mode of Operations with no multicast support". Implementations MAY support mode 3 ("... with multicast support" as that is a superset of mode 2). Note: Root indicates mode in DIO flow.
Proactive, aggressive DAO state maintenance:
Hopcount.
Objective Function (OF): Use OF0 [RFC6552]. No use of metric containers.
rank_factor: Derived from link speed: <= 100Mbps: LOW_SPEED_FACTOR(5), else HIGH_SPEED_FACTOR(1)
Global Repair: we assume stable links and ranks (metrics), so no need to periodically rebuild DODAG. DODAG version only incremented under catastrophic events (e.g., administrative action).
Local Repair: As soon as link breakage is detected, send No-Path DAO for all the targets that were reachable only via this link. As soon as link repair is detected, validate if this link provides you a better parent. If so, compute your new rank, and send new DIO that advertises your new rank. Then send a DAO with a new path sequence about yourself.
stretch_rank: none provided ("not stretched").
Data Path Validation: Not used.
Trickle: Not used.
Not used yet but possible because of the selected mode of operations.
[RFC6550] security not used, substituted by ACP security.
Because the ACP links already include provisions for confidentiality and integrity protection, their usage at the RPL layer would be redundant, and so RPL security is not used.
Not used.
Every ACP node (RPL node) announces an IPv6 prefix covering the address(es) used in the ACP node. The prefix length depends on the chosen addressing sub-scheme of the ACP address provisioned into the certificate of the ACP node, e.g., /127 for Zone Addressing Sub-Scheme or /112 or /120 for Vlong addressing sub-scheme. See Section 6.10 for more details.
Every ACP node MUST install a black hole (aka null) route for whatever ACP address space that it advertises (i.e.: the /96 or /127). This is avoid routing loops for addresses that an ACP node has not (yet) used.
Administrative Preference ([RFC6550], 3.2.6 – to become root): Indicated in DODAGPreference field of DIO message.
RPL Packet Information (RPI) defined in [RFC6550], section 11.2 defines the data packet artefacts required or beneficial in forwarding of those data packets when their routing information is derived from RPL. This profile does not use RPI for better compatibility with accelerated hardeware forwarding planes and achieves this for the following reasons.
One RPI option is the RPL Source Routing Header (SRH) [RFC6554] which is not necessary in this profile because it uses storing mode where each hop has the necessary next-hop forwarding information.
The simpler RPL Option header [RFC6553] is also not necessary in this profile, because it uses a single RPL instance and data path validation is also not used.
Because RPL minimizes the size of the routing and forwarding table, prefixes reachable through the same interface as the RPL root are not known on every ACP node. Therefore traffic to unknown destination addresses can only be discovered at the RPL root. The RPL root SHOULD have attach safe mechanisms to operationally discover and log such packets.
As this requirement raises additional data plane requirements, it does not apply to nodes where the administrative parameter to become root (Section 6.11.1.12) can always only be 0b001, e.g.: the node does not support explicit configuration to be root, or to be ACP registrar or to have ACP-connect functionality. If an ACP network is degraded to the point where there are no nodes that could be configured roots, ACP registrars or ACP-connect nodes, traffic to unknown destinations could not be diagnosed, but in the absence of any intelligent nodes supporting other than 0b001 administrative preference, there is likely also no diagnostic function possible.
Since channels are by default established between adjacent neighbors, the resulting overlay network does hop-by-hop encryption. Each node decrypts incoming traffic from the ACP, and encrypts outgoing traffic to its neighbors in the ACP. Routing is discussed in Section 6.11.
There are no performance requirements against ACP implementations defined in this document because the performance requirements depend on the intended use case. It is expected that full autonomic node with a wide range of ASA can require high forwarding plane performance in the ACP, for example for telemetry. Implementations of ACP to solely support traditional/SDN style use cases can benefit from ACP at lower performance, especially if the ACP is used only for critical operations, e.g., when the Data-Plane is not available. The design of the ACP as specified in this document is intended to support a wide range of performance options: It is intended to allow software-only implementations at potentially low performance, but can also support high performance options. See [RFC8368] for more details.
In order to be independent of the Data-Plane routing and addressing, the GRASP discovered ACP secure channels use IPv6 link local addresses between adjacent neighbors. Note: Section 8.2 specifies extensions in which secure channels are configured tunnels operating over the Data-Plane, so those secure channels cannot be independent of the Data-Plane.
To avoid that Data-Plane configuration can impact the operations of the IPv6 (link-local) interface/address used for ACP channels, appropriate implementation considerations are required. If the IPv6 interface/link-local address is shared with the Data-Plane it needs to be impossible to unconfigure/disable it through configuration. Instead of sharing the IPv6 interface/link-local address, a separate (virtual) interface with a separate IPv6 link-local address can be used. For example, the ACP interface could be run over a separate MAC address of an underlying L2 (Ethernet) interface. For more details and options, see Appendix A.10.2.
Note that other (non-ideal) implementation choices may introduce additional undesired dependencies against the Data-Plane. For example shared code and configuration of the secure channel protocols (IPsec / DTLS).
The MTU for ACP secure channels MUST be derived locally from the underlying link MTU minus the secure channel encapsulation overhead.
ACP secure Channel protocols do not need to perform MTU discovery because they are built across L2 adjacencies - the MTU on both sides connecting to the L2 connection are assumed to be consistent. Extensions to ACP where the ACP is for example tunneled need to consider how to guarantee MTU consistency. This is an issue of tunnels, not an issue of running the ACP across a tunnel. Transport stacks running across ACP can perform normal PMTUD (Path MTU Discovery). Because the ACP is meant to be prioritize reliability over performance, they MAY opt to only expect IPv6 minimum MTU (1280) to avoid running into PMTUD implementation bugs or underlying link MTU mismatch problems.
If two nodes are connected via several links, the ACP SHOULD be established across every link, but it is possible to establish the ACP only on a sub-set of links. Having an ACP channel on every link has a number of advantages, for example it allows for a faster failover in case of link failure, and it reflects the physical topology more closely. Using a subset of links (for example, a single link), reduces resource consumption on the node, because state needs to be kept per ACP channel. The negotiation scheme explained in Section 6.5 allows Alice (the node with the higher ACP address) to drop all but the desired ACP channels to Bob - and Bob will not re-try to build these secure channels from his side unless Alice shows up with a previously unknown GRASP announcement (e.g., on a different link or with a different address announced in GRASP).
The ACP VRF has conceptually two type of interfaces: The "ACP Loopback interface(s)" to which the ACP ULA address(es) are assigned and the "ACP virtual interfaces" that are mapped to the ACP secure channels.
The term "Loopback interface" was introduced initially to refer to an internal interface on a node that would allow IP traffic between transport endpoints on the node in the absence or failure of any or all external interfaces, see [RFC4291] section 2.5.3.
Even though Loopback interfaces were originally designed to hold only Loopback addresses not reachable from outside the node, these interfaces are also commonly used today to hold addresses reachable from the outside. They are meant to be reachable independent of any external interface being operational, and therefore to be more resilient. These addresses on Loopback interfaces can be thought of as "node addresses" instead of "interface addresses", and that is what ACP address(es) are. This construct makes it therefore possible to address ACP nodes with a well-defined set of addresses independent of the number of external interfaces.
For these reason, the ACP ULA address(es) are assigned to Loopback interface(s).
Any ACP secure channel to another ACP node is mapped to ACP virtual interfaces in one of the following ways. This is independent of the chosen secure channel protocol (IPsec, DTLS or other future protocol - standards or non-standards).
Note that all the considerations described here are assuming point-to-point secure channel associations. Mapping multi-party secure channel associations such as [RFC6407] is out of scope (but would be easy to add).
In this option, each ACP secure channel is mapped into a separate point-to-point ACP virtual interface. If a physical subnet has more than two ACP capable nodes (in the same domain), this implementation approach will lead to a full mesh of ACP virtual interfaces between them.
In a more advanced implementation approach, the ACP will construct a single multi-access ACP virtual interface for all ACP secure channels to ACP capable nodes reachable across the same underlying (physical) subnet. IPv6 link-local multicast packets sent into an ACP multi-access virtual interface are replicated to every ACP secure channel mapped into the ACP multicast-access virtual interface. IPv6 unicast packets sent into an ACP multi-access virtual interface are sent to the ACP secure channel that belongs to the ACP neighbor that is the next-hop in the ACP forwarding table entry used to reach the packets destination address.
There is no requirement for all ACP nodes on the same multi-access subnet to use the same type of ACP virtual interface. This is purely a node local decision.
ACP nodes MUST perform standard IPv6 operations across ACP virtual interfaces including SLAAC (Stateless Address Auto-Configuration) - [RFC4862]) to assign their IPv6 link local address on the ACP virtual interface and ND (Neighbor Discovery - [RFC4861]) to discover which IPv6 link-local neighbor address belongs to which ACP secure channel mapped to the ACP virtual interface. This is independent of whether the ACP virtual interface is point-to-point or multi-access.
"Optimistic Duplicate Address Detection (DAD)" according to [RFC4429] is RECOMMENDED because the likelihood for duplicates between ACP nodes is highly improbable as long as the address can be formed from a globally unique local assigned identifier (e.g., EUI-48/EUI-64, see below).
ACP nodes MAY reduce the amount of link-local IPv6 multicast packets from ND by learning the IPv6 link-local neighbor address to ACP secure channel mapping from other messages such as the source address of IPv6 link-local multicast RPL messages - and therefore forego the need to send Neighbor Solicitation messages.
The ACP virtual interface IPv6 link local address can be derived from any appropriate local mechanism such as node local EUI-48 or EUI-64 ("EUI" stands for "Extended Unique Identifier"). It MUST NOT depend on something that is attackable from the Data-Plane such as the IPv6 link-local address of the underlying physical interface, which can be attacked by SLAAC, or parameters of the secure channel encapsulation header that may not be protected by the secure channel mechanism.
The link-layer address of an ACP virtual interface is the address used for the underlying interface across which the secure tunnels are built, typically Ethernet addresses. Because unicast IPv6 packets sent to an ACP virtual interface are not sent to a link-layer destination address but rather an ACP secure channel, the link-layer address fields SHOULD be ignored on reception and instead the ACP secure channel from which the message was received should be remembered.
Multi-access ACP virtual interfaces are preferable implementations when the underlying interface is a (broadcast) multi-access subnet because they do reflect the presence of the underlying multi-access subnet into the virtual interfaces of the ACP. This makes it for example simpler to build services with topology awareness inside the ACP VRF in the same way as they could have been built running natively on the multi-access interfaces.
Consider also the impact of point-to-point vs. multi-access virtual interface on the efficiency of flooding via link local multicasted messages:
Assume a LAN with three ACP neighbors, Alice, Bob and Carol. Alice's ACP GRASP wants to send a link-local GRASP multicast message to Bob and Carol. If Alice's ACP emulates the LAN as one point-to-point virtual interface to Bob and one to Carol, The sending applications itself will send two copies, if Alice's ACP emulates a LAN, GRASP will send one packet and the ACP will replicate it. The result is the same. The difference happens when Bob and Carol receive their packet. If they use ACP point-to-point virtual interfaces, their GRASP instance would forward the packet from Alice to each other as part of the GRASP flooding procedure. These packets are unnecessary and would be discarded by GRASP on receipt as duplicates (by use of the GRASP Session ID). If Bob and Carol's ACP would emulate a multi-access virtual interface, then this would not happen, because GRASPs flooding procedure does not replicate back packets to the interface that they were received from.
Note that link-local GRASP multicast messages are not sent directly as IPv6 link-local multicast UDP messages into ACP virtual interfaces, but instead into ACP GRASP virtual interfaces, that are layered on top of ACP virtual interfaces to add TCP reliability to link-local multicast GRASP messages. Nevertheless, these ACP GRASP virtual interfaces perform the same replication of message and, therefore, result in the same impact on flooding. See Section 6.8.2 for more details.
RPL does support operations and correct routing table construction across non-broadcast multi-access (NBMA) subnets. This is common when using many radio technologies. When such NBMA subnets are used, they MUST NOT be represented as ACP multi-access virtual interfaces because the replication of IPv6 link-local multicast messages will not reach all NBMA subnet neighbors. In result, GRASP message flooding would fail. Instead, each ACP secure channel across such an interface MUST be represented as a ACP point-to-point virtual interface. See also Appendix A.10.4.
Care needs to be taken when creating multi-access ACP virtual interfaces across ACP secure channels between ACP nodes in different domains or routing subdomains. If for example future inter-domain ACP policies are defined as "peer-to-peer" policies, it is easier to create ACP point-to-point virtual interfaces for these inter-domain secure channels.
ANrtr1 ------ ANswitch1 --- ANswitch2 ------- ANrtr2
.../ \ \ ...
ANrtrM ------ \ ------- ANrtrN
ANswitchM ...
Figure 14: Topology with L2 ACP switches
Consider a large L2 LAN with ANrtr1...ANrtrN connected via some topology of L2 switches. Examples include large enterprise campus networks with an L2 core, IoT networks or broadband aggregation networks which often have even a multi-level L2 switched topology.
If the discovery protocol used for the ACP is operating at the subnet level, every ACP router will see all other ACP routers on the LAN as neighbors and a full mesh of ACP channels will be built. If some or all of the AN switches are autonomic with the same discovery protocol, then the full mesh would include those switches as well.
A full mesh of ACP connections can create fundamental scale challenges. The number of security associations of the secure channel protocols will likely not scale arbitrarily, especially when they leverage platform accelerated encryption/decryption. Likewise, any other ACP operations (such as routing) needs to scale to the number of direct ACP neighbors. An ACP router with just 4 physical interfaces might be deployed into a LAN with hundreds of neighbors connected via switches. Introducing such a new unpredictable scaling factor requirement makes it harder to support the ACP on arbitrary platforms and in arbitrary deployments.
Predictable scaling requirements for ACP neighbors can most easily be achieved if in topologies such as these, ACP capable L2 switches can ensure that discovery messages terminate on them so that neighboring ACP routers and switches will only find the physically connected ACP L2 switches as their candidate ACP neighbors. With such a discovery mechanism in place, the ACP and its security associations will only need to scale to the number of physical interfaces instead of a potentially much larger number of "LAN-connected" neighbors. And the ACP topology will follow directly the physical topology, something which can then also be leveraged in management operations or by ASAs.
In the example above, consider ANswitch1 and ANswitchM are ACP capable, and ANswitch2 is not ACP capable. The desired ACP topology is that ANrtr1 and ANrtrM only have an ACP connection to ANswitch1, and that ANswitch1, ANrtr2, ANrtrN have a full mesh of ACP connection amongst each other. ANswitch1 also has an ACP connection with ANswitchM and ANswitchM has ACP connections to anything else behind it.
To support ACP on L2 switches or L2 switched ports of an L3 device, it is necessary to make those L2 ports look like L3 interfaces for the ACP implementation. This primarily involves the creation of a separate DULL GRASP instance/domain on every such L2 port. Because GRASP has a dedicated link-local IPv6 multicast address (ALL_GRASP_NEIGHBORS), it is sufficient that all packets for this address are being extracted at the port level and passed to that DULL GRASP instance. Likewise the IPv6 link-local multicast packets sent by that DULL GRASP instance need to be sent only towards the L2 port for this DULL GRASP instance (instead of being flooded across all ports of the VLAN to which the port belongs).
When Ports/Interfaces across which the ACP is expected to operate in an ACP-aware L2-switch or L2/L3-switch/router are L2-bridged, packets for the ALL_GRASP_NEIGHBORS multicast address MUST never be forward between these ports. If MLD snooping is used, it MUST be prohibited from bridging packets for the ALL_GRASP_NEIGHBORS IPv6 multicast address.
On hybrid L2/L3 switches, multiple L2 ports are assigned to a single L3 VLAN interface. With the aforementioned changes for DULL GRASP, ACP can simply operate on the L3 VLAN interfaces, so no further (hardware) forwarding changes are required to make ACP operate on L2 ports. This is possible because the ACP secure channel protocols only use link-local IPv6 unicast packets, and these packets will be sent to the correct L2 port towards the peer by the VLAN logic of the device.
This is sufficient when p2p ACP virtual interfaces are established to every ACP peer. When it is desired to create multi-access ACP virtual interfaces (see Section 6.12.5.2.2), it is REQIURED not to coalesce all the ACP secure channels on the same L3 VLAN interface, but only all those on the same L2 port.
If VLAN tagging is used, then all the above described logic only applies to untagged GRASP packets. For the purpose of ACP neighbor discovery via GRASP, no VLAN tagged packets SHOULD be sent or received. In a hybrid L2/L3 switch, each VLAN would therefore only create ACP adjacencies across those ports where the VLAN is carried untagged.
In result, the simple logic is that ACP secure channels would operate over the same L3 interfaces that present a single flat bridged network across all routers, but because DULL GRASP is separated on a per-port basis, no full mesh of ACP secure channels is created, but only per-port ACP secure channels to per-port L2-adjacent ACP node neighbors.
For example, in the above picture, ANswitch1 would run separate DULL GRASP instances on its ports to ANrtr1, ANswitch2 and ANswitchI, even though all those three ports may be in the data plane in the same (V)LAN and perform L2 switching between these ports, ANswitch1 would perform ACP L3 routing between them.
The description in the previous paragraph was specifically meant to illustrate that on hybrid L3/L2 devices that are common in enterprise, IoT and broadband aggregation, there is only the GRASP packet extraction (by Ethernet address) and GRASP link-local multicast per L2-port packet injection that has to consider L2 ports at the hardware forwarding level. The remaining operations are purely ACP control plane and setup of secure channels across the L3 interface. This hopefully makes support for per-L2 port ACP on those hybrid devices easy.
In devices without such a mix of L2 port/interfaces and L3 interfaces (to terminate any transport layer connections), implementation details will differ. Logically most simply every L2 port is considered and used as a separate L3 subnet for all ACP operations. The fact that the ACP only requires IPv6 link-local unicast and multicast should make support for it on any type of L2 devices as simple as possible.
A generic issue with ACP in L2 switched networks is the interaction with the Spanning Tree Protocol. Without further L2 enhancements, the ACP would run only across the active STP topology and the ACP would be interrupted and re-converge with STP changes. Ideally, ACP peering SHOULD be built also across ports that are blocked in STP so that the ACP does not depend on STP and can continue to run unaffected across STP topology changes, where re-convergence can be quite slow. The above described simple implementation options are not sufficient to achieve this.
The Autonomic Control Plane can be used by management systems, such as controllers or network management system (NMS) hosts (henceforth called simply "NMS hosts"), to connect to devices (or other type of nodes) through it. For this, an NMS host needs to have access to the ACP. The ACP is a self-protecting overlay network, which allows by default access only to trusted, autonomic systems. Therefore, a traditional, non-ACP NMS system does not have access to the ACP by default, such as any other external node.
If the NMS host is not autonomic, i.e., it does not support autonomic negotiation of the ACP, then it can be brought into the ACP by explicit configuration. To support connections to adjacent non-ACP nodes, an ACP node SHOULD support "ACP connect" (sometimes also called "autonomic connect"):
"ACP connect" is an interface level configured workaround for connection of trusted non-ACP nodes to the ACP. The ACP node on which ACP connect is configured is called an "ACP edge node". With ACP connect, the ACP is accessible from those non-ACP nodes (such as NOC systems) on such an interface without those non-ACP nodes having to support any ACP discovery or ACP channel setup. This is also called "native" access to the ACP because to those NOC systems the interface looks like a normal network interface (without any encryption/novel-signaling).
Data-Plane "native" (no ACP)
.
+--------+ +----------------+ . +-------------+
| ACP | |ACP Edge Node | . | |
| Node | | | v | |
| |-------|...[ACP VRF]....+-----------------| |+
| | ^ |. | | NOC Device ||
| | . | .[Data-Plane]..+-----------------| "NMS hosts" ||
| | . | [ ] | . ^ | ||
+--------+ . +----------------+ . . +-------------+|
. . . +-------------+
. . .
Data-Plane "native" . ACP "native" (unencrypted)
+ ACP auto-negotiated . "ACP connect subnet"
and encrypted .
ACP connect interface
e.g., "VRF ACP native" (config)
Figure 15: ACP connect
ACP connect has security consequences: All systems and processes connected via ACP connect have access to all ACP nodes on the entire ACP, without further authentication. Thus, the ACP connect interface and NOC systems connected to it needs to be physically controlled/secured. For this reason the mechanisms described here do explicitly not include options to allow for a non-ACP router to be connected across an ACP connect interface and addresses behind such a router routed inside the ACP.
An ACP connect interface provides exclusively access to only the ACP. This is likely insufficient for many NMS hosts. Instead, they would require a second "Data-Plane" interface outside the ACP for connections between the NMS host and administrators, or Internet based services, or for direct access to the Data-Plane. The document "Using Autonomic Control Plane for Stable Connectivity of Network OAM" explains in more detail how the ACP can be integrated in a mixed NOC environment.
An ACP connect interface SHOULD use an IPv6 address/prefix from the ACP Manual Addressing Sub-Scheme (Section 6.10.4), letting the operator configure for example only the Subnet-ID and having the node automatically assign the remaining part of the prefix/address. It SHOULD NOT use a prefix that is also routed outside the ACP so that the addresses clearly indicate whether it is used inside the ACP or not.
The prefix of ACP connect subnets MUST be distributed by the ACP edge node into the ACP routing protocol RPL. The NMS hosts MUST connect to prefixes in the ACP routing table via its ACP connect interface. In the simple case where the ACP uses only one ULA prefix and all ACP connect subnets have prefixes covered by that ULA prefix, NMS hosts can rely on [RFC6724] to determine longest match prefix routes towards its different interfaces, ACP and data-plane. With RFC6724, The NMS host will select the ACP connect interface for all addresses in the ACP because any ACP destination address is longest matched by the address on the ACP connect interface. If the NMS hosts ACP connect interface uses another prefix or if the ACP uses multiple ULA prefixes, then the NMS hosts require (static) routes towards the ACP interface for these prefixes.
When an ACP Edge node receives a packet from an ACP connect interface, the ACP Edge node MUST only forward the packet into the ACP if the packet has an IPv6 source address from that interface. This is sometimes called "RPF filtering". This MAY be changed through administrative measures.
To limit the security impact of ACP connect, nodes supporting it SHOULD implement a security mechanism to allow configuration/use of ACP connect interfaces only on nodes explicitly targeted to be deployed with it (those in physically secure locations such as a NOC). For example, the registrar could disable the ability to enable ACP connect on devices during enrollment and that property could only be changed through re-enrollment. See also Appendix A.10.5.
ACP Edge nodes SHOULD have a configurable option to filter packets with RPI headers (xsee Section 6.11.1.13 across an ACP connect interface. These headers are outside the scope of the RPL profile in this specification but may be used in future extensions of this specification.
The previous section assumed that ACP Edge node and NOC devices are separate physical devices and the ACP connect interface is a physical network connection. This section discusses the implication when these components are instead software components running on a single physical device.
The ACP connect mechanism can not only be used to connect physically external systems (NMS hosts) to the ACP but also other applications, containers or virtual machines. In fact, one possible way to eliminate the security issue of the external ACP connect interface is to collocate an ACP edge node and an NMS host by making one a virtual machine or container inside the other; and therefore converting the unprotected external ACP subnet into an internal virtual subnet in a single device. This would ultimately result in a fully ACP enabled NMS host with minimum impact to the NMS hosts software architecture. This approach is not limited to NMS hosts but could equally be applied to devices consisting of one or more VNF (virtual network functions): An internal virtual subnet connecting out-of-band management interfaces of the VNFs to an ACP edge router VNF.
The core requirement is that the software components need to have a network stack that permits access to the ACP and optionally also the Data-Plane. Like in the physical setup for NMS hosts this can be realized via two internal virtual subnets. One that is connecting to the ACP (which could be a container or virtual machine by itself), and one (or more) connecting into the Data-Plane.
This "internal" use of ACP connect approach should not considered to be a "workaround" because in this case it is possible to build a correct security model: It is not necessary to rely on unprovable external physical security mechanisms as in the case of external NMS hosts. Instead, the orchestration of the ACP, the virtual subnets and the software components can be done by trusted software that could be considered to be part of the ANI (or even an extended ACP). This software component is responsible for ensuring that only trusted software components will get access to that virtual subnet and that only even more trusted software components will get access to both the ACP virtual subnet and the Data-Plane (because those ACP users could leak traffic between ACP and Data-Plane). This trust could be established for example through cryptographic means such as signed software packages.
ACP edge nodes, NMS hosts and software components that as described in the previous section are meant to be composed via virtual interfaces SHOULD support on the ACP connect subnet StateLess Address Autoconfiguration (SLAAC - [RFC4862]) and route auto configuration according to [RFC4191].
The ACP edge node acts as the router on the ACP connect subnet, providing the (auto-)configured prefix for the ACP connect subnet to NMS hosts and/or software components. The ACP edge node uses route prefix option of RFC4191 to announce the default route (::/) with a lifetime of 0 and aggregated prefixes for routes in the ACP routing table with normal lifetimes. This will ensure that the ACP edge node does not become a default router, but that the NMS hosts and software components will route the prefixes used in the ACP to the ACP edge node.
Aggregated prefix means that the ACP edge node needs to only announce the /48 ULA prefixes used in the ACP but none of the actual /64 (Manual Addressing Sub-Scheme), /127 (ACP Zone Addressing Sub-Scheme), /112 or /120 (Vlong Addressing Sub-Scheme) routes of actual ACP nodes. If ACP interfaces are configured with non ULA prefixes, then those prefixes cannot be aggregated without further configured policy on the ACP edge node. This explains the above recommendation to use ACP ULA prefix covered prefixes for ACP connect interfaces: They allow for a shorter list of prefixes to be signaled via RFC4191 to NMS hosts and software components.
The ACP edge nodes that have a Vlong ACP address MAY allocate a subset of their /112 or /120 address prefix to ACP connect interface(s) to eliminate the need to non-autonomically configure/provision the address prefixes for such ACP connect interfaces.
Combined ACP and Data-Plane interface
.
+--------+ +--------------------+ . +--------------+
| ACP | |ACP Edge No | . | NMS Host(s) |
| Node | | | . | / Software |
| | | [ACP ]. | . | |+
| | | .[VRF ] .[VRF ] | v | "ACP address"||
| +-------+. .[Select].+--------+ "Date Plane ||
| | ^ | .[Data ]. | | Address(es)"||
| | . | [Plane] | | ||
| | . | [ ] | +--------------+|
+--------+ . +--------------------+ +--------------+
.
Data-Plane "native" and + ACP auto-negotiated/encrypted
Figure 16: VRF select
Using two physical and/or virtual subnets (and therefore interfaces) into NMS Hosts (as per Section 8.1.1) or Software (as per Section 8.1.2) may be seen as additional complexity, for example with legacy NMS Hosts that support only one IP interface.
To provide a single subnet into both ACP and Data-Plane, the ACP Edge node needs to de-multiplex packets from NMS hosts into ACP VRF and Data-Plane. This is sometimes called "VRF select". If the ACP VRF has no overlapping IPv6 addresses with the Data-Plane (it should have no overlapping addresses), then this function can use the IPv6 Destination address. The problem is Source Address Selection on the NMS Host(s) according to RFC6724.
Consider the simple case: The ACP uses only one ULA prefix, the ACP IPv6 prefix for the Combined ACP and Data-Plane interface is covered by that ULA prefix. The ACP edge node announces both the ACP IPv6 prefix and one (or more) prefixes for the Data-Plane. Without further policy configurations on the NMS Host(s), it may select its ACP address as a source address for Data-Plane ULA destinations because of Rule 8 of RFC6724. The ACP edge node can pass on the packet to the Data-Plane, but the ACP source address should not be used for Data-Plane traffic, and return traffic may fail.
If the ACP carries multiple ULA prefixes or non-ULA ACP connect prefixes, then the correct source address selection becomes even more problematic.
With separate ACP connect and Data-Plane subnets and RFC4191 prefix announcements that are to be routed across the ACP connect interface, RFC6724 source address selection Rule 5 (use address of outgoing interface) will be used, so that above problems do not occur, even in more complex cases of multiple ULA and non-ULA prefixes in the ACP routing table.
To achieve the same behavior with a Combined ACP and Data-Plane interface, the ACP Edge Node needs to behave as two separate routers on the interface: One link-local IPv6 address/router for its ACP reachability, and one link-local IPv6 address/router for its Data-Plane reachability. The Router Advertisements for both are as described above (Section 8.1.3): For the ACP, the ACP prefix is announced together with RFC4191 option for the prefixes routed across the ACP and lifetime=0 to disqualify this next-hop as a default router. For the Data-Plane, the Data-Plane prefix(es) are announced together with whatever dafault router parameters are used for the Data-Plane.
In result, RFC6724 source address selection Rule 5.5 may result in the same correct source address selection behavior of NMS hosts without further configuration on it as the separate ACP connect and Data-Plane interfaces. As described in the text for Rule 5.5, this is only a MAY, because IPv6 hosts are not required to track next-hop information. If an NMS Host does not do this, then separate ACP connect and Data-Plane interfaces are the preferable method of attachment. Hosts implementing [RFC8028] should (instead of may) implement [RFC6724] Rule 5.5, so it is preferred for hosts to support [RFC8028].
ACP edge nodes MAY support the Combined ACP and Data-Plane interface.
GRASP can and should be possible to use across ACP connect interfaces, especially in the architectural correct solution when it is used as a mechanism to connect Software (e.g., ASA or legacy NMS applications) to the ACP.
Given how the ACP is the security and transport substrate for GRASP, the requirements for devices connected via ACP connect is that those are equivalently (if not better) secured against attacks and run only software that is equally (if not better) protected, known (or trusted) not to be malicious and accordingly designed to isolate access to the ACP against external equipment.
The difference in security is that cryptographic security of the ACP secure channel is replaced by required physical security of the network connection between an ACP edge node and the NMS or other host reachable via the ACP connect interface. Node integrity too is expected to be easier because the ACP connect node, the ACP connect link and the nodes connecting to it must be in a contiguous secure environment, hence assuming there can be no physical attack against the devices.
When using "Combined ACP and Data-Plane Interfaces", care hasa to be taken that only GRASP messages intended for the ACP GRASP domain received from Software or NMS Hosts are forwarded by ACP edge nodes. Currently there is no definition for a GRASP security and transport substrate beside the ACP, so there is no definition how such Software/NMS Host could participate in two separate GRASP Domains across the same subnet (ACP and Data-Plane domains). At current it is assumed that all GRASP packets on a Combined ACP and Data-Plane interface belong to the GRASP ACP Domain. They SHOULD all use the ACP IPv6 addresses of the Software/NMS Hosts. The link-local IPv6 addresses of Software/NMS Hosts (used for GRASP M_DISCOVERY and M_FLOOD messages) are also assumed to belong to the ACP address space.
Not all nodes in a network may support the ACP. If non-ACP Layer-2 devices are between ACP nodes, the ACP will work across it since it is IP based. However, the autonomic discovery of ACP neighbors via DULL GRASP is only intended to work across L2 connections, so it is not sufficient to autonomically create ACP connections across non-ACP Layer-3 devices.
On the ACP node, remote ACP neighbors are configured explicitly. The parameters of such a "connection" are described in the following ABNF.
connection = [ method , local-addr, remote-addr, ?pmtu ]
method = [ "IKEv2" , ?port ]
method //= [ "DTLS", port ]
local-addr = [ address , ?vrf ]
remote-addr = [ address ]
address = ("any" | ipv4-address | ipv6-address )
vrf = tstr ; Name of a VRF on this node with local-address
Figure 17: Parameters for remote ACP neighbors
Explicit configuration of a remote-peer according to this ABNF provides all the information to build a secure channel without requiring a tunnel to that peer and running DULL GRASP inside of it.
The configuration includes the parameters otherwise signaled via DULL GRASP: local address, remote (peer) locator and method. The differences over DULL GRASP local neighbor discovery and secure channel creation are as follows:
An IPinIP, GRE or other form of pre-existing tunnel is configured between two remote ACP peers and the virtual interfaces representing the tunnel are configured for "ACP enable". This will enable IPv6 link local addresses and DULL on this tunnel. In result, the tunnel is used for normal "L2 adjacent" candidate ACP neighbor discovery with DULL and secure channel setup procedures described in this document.
Tunneled Remote ACP Neighbor requires two encapsulations: the configured tunnel and the secure channel inside of that tunnel. This makes it in general less desirable than Configured Remote ACP Neighbor. Benefits of tunnels are that it may be easier to implement because there is no change to the ACP functionality - just running it over a virtual (tunnel) interface instead of only native interfaces. The tunnel itself may also provide PMTUD while the secure channel method may not. Or the tunnel mechanism is permitted/possible through some firewall while the secure channel method may not.
Configured/Tunneled Remote ACP neighbors are less "indestructible" than L2 adjacent ACP neighbors based on link local addressing, since they depend on more correct Data-Plane operations, such as routing and global addressing.
Nevertheless, these options may be crucial to incrementally deploy the ACP, especially if it is meant to connect islands across the Internet. Implementations SHOULD support at least Tunneled Remote ACP Neighbors via GRE tunnels - which is likely the most common router-to-router tunneling protocol in use today.
The following sections document important operational aspects of the ACP. They are not normative because they do not impact the interoperability between components of the ACP, but they include recommendations/requirements for the internal operational model beneficial or necessary to achieve the desired use-case benefits of the ACP (see Section 3).
The recommendations and suggestions of this chapter were derived from operational experience gained with a commercially available pre-standard ACP implementation.
Even though ACP and ANI in general are taking out many manual configuration mistakes through their automation, it is important to provide good diagnostics for them.
The basic diagnostics is support of (yang) data models representing the complete (auto-)configuration and operational state of all components: BRSKI, GRASP, ACP and the infrastructure used by them: TLS/DTLS, IPsec, certificates, trust anchors, time, VRF and so on. While necessary, this is not sufficient:
Simply representing the state of components does not allow operators to quickly take action - unless they do understand how to interpret the data, and that can mean a requirement for deep understanding of all components and how they interact in the ACP/ANI.
Diagnostic supports should help to quickly answer the questions operators are expected to ask, such as "is the ACP working correctly?", or "why is there no ACP connection to a known neighboring node?"
In current network management approaches, the logic to answer these questions is most often built as centralized diagnostics software that leverages the above mentioned data models. While this approach is feasible for components utilizing the ANI, it is not sufficient to diagnose the ANI itself:
The most simple form of diagnostics answering questions such as the above is to represent the relevant information sequentially in dependency order, so that the first non-expected/non-operational item is the most likely root cause. Or just log/highlight that item. For example:
Q: Is ACP operational to accept neighbor connections:
If all this looks good, the ACP should be running locally "fine" - but we did not check any ACP neighbor relationships.
Question: why does the node not create a working ACP connection to a neighbor on an interface?
Question: Is the ACP operating correctly across its secure channels?
These lists are not necessarily complete, but illustrate the principle and show that there are variety of issues ranging from normal operational causes (a neighbor in another ACP domain) over problems in the credentials management (certificate lifetimes), explicit security actions (revocation) or unexpected connectivity issues (intervening L2 equipment).
The items so far are illustrating how the ANI operations can be diagnosed with passive observation of the operational state of its components including historic/cached/counted events. This is not necessary sufficient to provide good enough diagnostics overall:
The components of ACP and BRSKI are designed with security in mind but they do not attempt to provide diagnostics for building the network itself. Consider two examples:
An "interested adjacent party" can always determine the IDevID of a BRSKI pledge by behaving like a BRSKI proxy/registrar. Therefore the IDevID of a BRSKI pledge is not meant to be protected - it just has to be queried and is not signaled unsolicited (as it would be in LLDP) so that other observers on the same subnet can determine who is an "interested adjacent party".
As described in Section 6.10.7, the ACP addressing mechanism is designed to enable lightweight, distributed and uncoordinated ACP registrars that are providing ACP address prefixes to candidate ACP nodes by enrolling them with an ACP domain certificate into an ACP domain via any appropriate mechanism/protocol, automated or not.
This section discusses informatively more details and options for ACP registrars.
This section summarizes and discusses the interactions with other entities required by an ACP registrar.
In a simple instance of an ACP network, no central NOC component beside a trust anchor (root CA) is required. One or more uncoordinated acting ACP registrar can be set up, performing the following interactions:
To orchestrate enrolling a candidate ACP node autonomically, the ACP registrar can rely on the ACP and use Proxies to reach the candidate ACP node, therefore allowing minimum pre-existing (auto-)configured network services on the candidate ACP node. BRSKI defines the BRSKI proxy, a design that can be adopted for various protocols that Pledges/candidate ACP nodes could want to use, for example BRSKI over CoAP (Constrained Application Protocol), or proxying of Netconf.
To reach a trust anchor unaware of the ACP, the ACP registrar would use the Data-Plane. ACP and Data-Plane in an ACP registrar could (and by default should be) completely isolated from each other at the network level. Only applications such as the ACP registrar would need the ability for their transport stacks to access both.
In non-autonomic enrollment options, the Data-Plane between a ACP registrar and the candidate ACP node needs to be configured first. This includes the ACP registrar and the candidate ACP node. Then any appropriate set of protocols can be used between ACP registrar and candidate ACP node to discover the other side, and then connect and enroll (configure) the candidate ACP node with an ACP domain certificate. Netconf ZeroTouch ([RFC8572]) is an example protocol that could be used for this. BRSKI using optional discovery mechanisms is equally a possibility for candidate ACP nodes attempting to be enrolled across non-ACP networks, such as the Internet.
When candidate ACP nodes have secure bootstrap, such as BRSKI Pledges, they will not trust to be configured/enrolled across the network, unless being presented with a voucher (see [RFC8366]) authorizing the network to take possession of the node. An ACP registrar will then need a method to retrieve such a voucher, either offline, or online from a MASA (Manufacturer Authorized Signing Authority). BRSKI and Netconf ZeroTouch are two protocols that include capabilities to present the voucher to the candidate ACP node.
An ACP registrar could operate EST for ACP certificate renewal and/or act as a CRL Distribution point. A node performing these services does not need to support performing (initial) enrollment, but it does require the same above described connectivity as an ACP registrar: via the ACP to ACP nodes and via the Data-Plane to the trust anchor and other sources of CRL information.
The interactions of an ACP registrar outlined Section 6.10.7 and Section 9.2.1 above depend on the following parameters:
When an ACP node renews/rekeys its certificate, it may end up doing so via a different registrar (e.g., EST server) than the one it originally received its ACP domain certificate from, for example because that original ACP registrar is gone. The ACP registrar through which the renewal/rekeying is performed would by default trust the ACP domain information from the ACP nodes current ACP domain certificate and maintain this information so that the ACP node maintains its ACP address prefix. In EST renewal/rekeying, the ACP nodes current ACP domain certificate is signaled during the TLS handshake.
This simple scenario has two limitations:
Even when such a malicious ACP registrar is detected, resolving the problem may be difficult because it would require identifying all the wrong ACP domain certificates assigned via the ACP registrar after it was compromised. And without additional centralized tracking of assigned certificates there is no way to do this.
In situations, where either of the above two limitations are an issue, ACP registrars could also be sub-CAs. This removes the need for connectivity to a root-CA whenever an ACP node is enrolled, and reduces the need for connectivity of such an ACP registrar to a root-CA to only those times when it needs to renew its own certificate. The ACP registrar would also now use its own (sub-CA) certificate to enroll and sign the ACP nodes certificates, and therefore it is only necessary to revoke a compromised ACP registrars sub-CA certificate. Alternatively one can let it expire and not renew it, when the certificate of the sub-CA is appropriately short-lived.
As the ACP domain membership check verifies a peer ACP node's ACP domain certificate trust chain, it will also verify the signing certificate which is the compromised/revoked sub-CA certificate. Therefore ACP domain membership for an ACP node enrolled from a compromised and discovered ACP registrar will fail.
ACP nodes enrolled by a compromised ACP registrar would automatically fail to establish ACP channels and ACP domain certificate renewal via EST and therefore revert to their role as a candidate ACP members and attempt to get a new ACP domain certificate from an ACP registrar - for example, via BRSKI. In result, ACP registrars that have an associated sub-CA makes isolating and resolving issues with compromised registrars easier.
Note that ACP registrars with sub-CA functionality also can control the lifetime of ACP domain certificates easier and therefore also be used as a tool to introduce short lived certificates and not rely on CRL, whereas the certificates for the sub-CAs themselves could be longer lived and subject to CRL.
When using multiple, uncoordinated ACP registrars, several advanced operations are potentially more complex than with a single, resilient policy control backend, for example including but not limited to:
These and other operations could be introduced more easily by introducing a centralized Policy Management System (PMS) and modifying ACP registrar behavior so that it queries the PMS for any policy decision occurring during the candidate ACP node enrollment process and/or the ACP node certificate renewal process. For example, which ACP address prefix to assign. Likewise the ACP registrar would report any relevant state change information to the PMS as well, for example when a certificate was successfully enrolled onto a candidate ACP node.
Both ACP and BRSKI require interfaces to be operational enough to support sending/receiving their packets. In node types where interfaces are by default (e.g., without operator configuration) enabled, such as most L2 switches, this would be less of a change in behavior than in most L3 devices (e.g.: routers), where interfaces are by default disabled. In almost all network devices it is common though for configuration to change interfaces to a physically disabled state and that would break the ACP.
In this section, we discuss a suggested operational model to enable/disable interfaces and nodes for ACP/ANI in a way that minimizes the risk of operator action to break the ACP in this way, and that also minimizes operator surprise when ACP/ANI becomes supported in node software.
Whenever this document refers to enabling an interface for ACP (or BRSKI), it only requires to permit the interface to send/receive packets necessary to operate ACP (or BRSKI) - but not any other Data-Plane packets. Unless the Data-Plane is explicitly configured/enabled, all packets not required for ACP/BRSKI should be filtered on input and output:
Both BRSKI and ACP require link-local only IPv6 operations on interfaces and DULL GRASP. IPv6 link-local operations means the minimum signaling to auto-assign an IPv6 link-local address and talk to neighbors via their link-local address: SLAAC (Stateless Address Auto-Configuration - [RFC4862]) and ND (Neighbor Discovery - [RFC4861]). When the device is a BRSKI pledge, it may also require TCP/TLS connections to BRSKI proxies on the interface. When the device has keying material, and the ACP is running, it requires DULL GRASP packets and packets necessary for the secure-channel mechanism it supports, e.g., IKEv2 and IPsec ESP packets or DTLS packets to the IPv6 link-local address of an ACP neighbor on the interface. It also requires TCP/TLS packets for its BRSKI proxy functionality, if it does support BRSKI.
Interfaces on most network equipment have at least two states: "up" and "down". These may have product specific names. "down" for example could be called "shutdown" and "up" could be called "no shutdown". The "down" state disables all interface operations down to the physical level. The "up" state enables the interface enough for all possible L2/L3 services to operate on top of it and it may also auto-enable some subset of them. More commonly, the operations of various L2/L3 services is controlled via additional node-wide or interface level options, but they all become only active when the interface is not "down". Therefore an easy way to ensure that all L2/L3 operations on an interface are inactive is to put the interface into "down" state. The fact that this also physically shuts down the interface is in many cases just a side effect, but it may be important in other cases (see below, Section 9.3.2.2).
To provide ACP/ANI resilience against operators configuring interfaces to "down" state, this document recommends to separate the "down" state of interfaces into an "admin down" state where the physical layer is kept running and ACP/ANI can use the interface and a "physical down" state. Any existing "down" configurations would map to "admin down". In "admin down", any existing L2/L3 services of the Data-Plane should see no difference to "physical down" state. To ensure that no Data-Plane packets could be sent/received, packet filtering could be established automatically as described above in Section 9.3.1.
As necessary (see discussion below) new configuration options could be introduced to issue "physical down". The options should be provided with additional checks to minimize the risk of issuing them in a way that breaks the ACP without automatic restoration. For example they could be denied to be issued from a control connection (netconf/ssh) that goes across the interface itself ("do not disconnect yourself"). Or they could be performed only temporary and only be made permanent with additional later reconfirmation.
In the following sub-sections important aspects to the introduction of "admin down" state are discussed.
Interfaces are physically brought down (or left in default down state) as a form of security. "Admin down" state as described above provides also a high level of security because it only permits ACP/ANI operations which are both well secured. Ultimately, it is subject to security review for the deployment whether "admin down" is a feasible replacement for "physical down".
The need to trust the security of ACP/ANI operations needs to be weighed against the operational benefits of permitting this: Consider the typical example of a CPE (customer premises equipment) with no on-site network expert. User ports are in physical down state unless explicitly configured not to be. In a misconfiguration situation, the uplink connection is incorrectly plugged into such as user port. The device is disconnected from the network and therefore no diagnostics from the network side is possible anymore. Alternatively, all ports default to "admin down". The ACP (but not the Data-Plane) would still automatically form. Diagnostics from the network side is possible and operator reaction could include to either make this port the operational uplink port or to instruct re-cabling. Security wise, only ACP/ANI could be attacked, all other functions are filtered on interfaces in "admin down" state.
"Physical down" state propagates on many interface types (e.g., Ethernet) to the other side. This can trigger fast L2/L3 protocol reaction on the other side and "admin down" would not have the same (fast) result.
Bringing interfaces to "physical down" state is to the best of our knowledge always a result of operator action, but today, never the result of autonomic L2/L3 services running on the nodes. Therefore one option is to change the operator action to not rely on link-state propagation anymore. This may not be possible when both sides are under different operator control, but in that case it is unlikely that the ACP is running across the link and actually putting the interface into "physical down" state may still be a good option.
Ideally, fast physical state propagation is replaced by fast software driven state propagation. For example a DULL GRASP "admin-state" objective could be used to auto configure a Bidirectional Forwarding Protocol (BFD, [RFC5880]) session between the two sides of the link that would be used to propagate the "up" vs. admin down state.
Triggering physical down state may also be used as a mean of diagnosing cabling in the absence of easier methods. It is more complex than automated neighbor diagnostics because it requires coordinated remote access to both (likely) sides of a link to determine whether up/down toggling will cause the same reaction on the remote side.
See Section 9.1 for a discussion about how LLDP and/or diagnostics via GRASP could be used to provide neighbor diagnostics, and therefore hopefully eliminating the need for "physical down" for neighbor diagnostics - as long as both neighbors support ACP/ANI.
"Physical down" is performed to diagnose low-level interface behavior when higher layer services (e.g., IPv6) are not working. Especially Ethernet links are subject to a wide variety of possible wrong configuration/cablings if they do not support automatic selection of variable parameters such as speed (10/100/1000 Mbps), crossover (Auto-MDIX) and connector (fiber, copper - when interfaces have multiple but can only enable one at a time). The need for low level link diagnostic can therefore be minimized by using fully auto configuring links.
In addition to "Physical down", low level diagnostics of Ethernet or other interfaces also involve the creation of other states on interfaces, such as physical Loopback (internal and/or external) or bringing down all packet transmissions for reflection/cable-length measurements. Any of these options would disrupt ACP as well.
In cases where such low-level diagnostics of an operational link is desired but where the link could be a single point of failure for the ACP, ASA on both nodes of the link could perform a negotiated diagnostics that automatically terminates in a predetermined manner without dependence on external input ensuring the link will become operational again.
Power consumption of "physical down" interfaces, may be significantly lower than those in "admin down" state, for example on long-range fiber interfaces. Bringing up interfaces, for example to probe reachability, may also consume additional power. This can make these type of interfaces inappropriate to operate purely for the ACP when they are not currently needed for the Data-Plane.
The interface level configuration option "ACP enable" enables ACP operations on an interface, starting with ACP neighbor discovery via DULL GRAP. The interface level configuration option "ANI enable" on nodes supporting BRSKI and ACP starts with BRSKI pledge operations when there is no domain certificate on the node. On ACP/BRSKI nodes, "ACP enable" may not need to be supported, but only "ANI enable". Unless overridden by global configuration options (see later), "ACP/ANI enable" will result in "down" state on an interface to behave as "admin down".
(Section 6.3) requires that "ACP enable" is automatically set on native interfaces, but not on non-native interfaces (reminder: a native interface is one that exists without operator configuration action such as physical interfaces in physical devices).
Ideally, ACP enable is set automatically on all interfaces that provide access to additional connectivity that allows to reach more nodes of the ACP domain. The best set of interfaces necessary to achieve this is not possible to determine automatically. Native interfaces are the best automatic approximation.
Consider an ACP domain of ACP nodes transitively connected via native interfaces. A Data-Plane tunnel between two of these nodes that are non-adjacent is created and "ACP enable" is set for that tunnel. ACP RPL sees this tunnel as just as a single hop. Routes in the ACP would use this hop as an attractive path element to connect regions adjacent to the tunnel nodes. In result, the actual hop-by-hop paths used by traffic in the ACP can become worse. In addition, correct forwarding in the ACP now depends on correct Data-Plane forwarding config including QoS, filtering and other security on the Data-Plane path across which this tunnel runs. This is the main issue why "ACP/ANI enable" should not be set automatically on non-native interfaces.
If the tunnel would connect two previously disjoint ACP regions, then it likely would be useful for the ACP. A Data-Plane tunnel could also run across nodes without ACP and provide additional connectivity for an already connected ACP network. The benefit of this additional ACP redundancy has to be weighed against the problems of relying on the Data-Plane. If a tunnel connects two separate ACP regions: how many tunnels should be created to connect these ACP regions reliably enough? Between which nodes? These are all standard tunneled network design questions not specific to the ACP, and there are no generic fully automated answers.
Instead of automatically setting "ACP enable" on these type of interfaces, the decision needs to be based on the use purpose of the non-native interface and "ACP enable" needs to be set in conjunction with the mechanism through which the non-native interface is created/configured.
In addition to explicit setting of "ACP/ANI enable", non-native interfaces also need to support configuration of the ACP RPL cost of the link - to avoid the problems of attracting too much traffic to the link as described above.
Even native interfaces may not be able to automatically perform BRSKI or ACP because they may require additional operator input to become operational. Example include DSL interfaces requiring PPPoE credentials or mobile interfaces requiring credentials from a SIM card. Whatever mechanism is used to provide the necessary config to the device to enable the interface can also be expanded to decide on whether or not to set "ACP/ANI enable".
The goal of automatically setting "ACP/ANI enable" on interfaces (native or not) is to eliminate unnecessary "touches" to the node to make its operation as much as possible "zero-touch" with respect to ACP/ANI. If there are "unavoidable touches" such a creating/configuring a non-native interface or provisioning credentials for a native interface, then "ACP/ANI enable" should be added as an option to that "touch". If a wrong "touch" is easily fixed (not creating another high-cost touch), then the default should be not to enable ANI/ACP, and if it is potentially expensive or slow to fix (e.g., parameters on SIM card shipped to remote location), then the default should be to enable ACP/ANI.
A node level command "ACP/ANI enable [up-if-only]" enables ACP or ANI on the node (ANI = ACP + BRSKI). Without this command set, any interface level "ACP/ANI enable" is ignored. Once set, ACP/ANI will operate an interface where "ACP/ANI enable" is set. Setting of interface level "ACP/ANI enable" is either automatic (default) or explicit through operator action as described in the previous section.
If the option "up-if-only" is selected, the behavior of "down" interfaces is unchanged, and ACP/ANI will only operate on interfaces where "ACP/ANI enable" is set and that are "up". When it is not set, then "down" state of interfaces with "ACP/ANI enable" is modified to behave as "admin down".
A "brownfield" node is one that already has a configured Data-Plane.
Executing global "ACP/ANI enable [up-if-only]" on each node is the only command necessary to create an ACP across a network of brownfield nodes once all the nodes have a domain certificate. When BRSKI is used ("ANI enable"), provisioning of the certificates only requires set-up of a single BRSKI registrar node which could also implement a CA for the network. This is the most simple way to introduce ACP/ANI into existing (== brownfield) networks.
The need to explicitly enable ACP/ANI is especially important in brownfield nodes because otherwise software updates may introduce support for ACP/ANI: Automatic enablement of ACP/ANI in networks where the operator does not only not want ACP/ANI but where the operator likely never even heard of it could be quite irritating to the operator. Especially when "down" behavior is changed to "admin down".
Automatically setting "ANI enable" on brownfield nodes where the operator is unaware of BRSKI and MASA operations could also be an unlikely but then critical security issue. If an attacker could impersonate the operator and register as the operator at the MASA or otherwise get hold of vouchers and can get enough physical access to the network so pledges would register to an attacking registrar, then the attacker could gain access to the network through the ACP that the attacker then has access to.
In networks where the operator explicitly wants to enable the ANI this could not happen, because the operator would create a BRSKI registrar that would discover attack attempts, and the operator would be setting up his registrar with the MASA. Nodes requiring "ownership vouchers" would not be subject to that attack. See [I-D.ietf-anima-bootstrapping-keyinfra] for more details. Note that a global "ACP enable" alone is not subject to these type of attacks, because it always depends on some other mechanism first to provision domain certificates into the device.
A "greenfield" node is one that did not have any prior configuration.
For greenfield nodes, only "ANI enable" is relevant. If another mechanism than BRSKI is used to (zero-touch) bootstrap a node, then it is up to that mechanism to provision domain certificates and to set global "ACP enable" as desired.
Nodes supporting full ANI functionality set "ANI enable" automatically when they decide that they are greenfield, e.g., that they are powering on from factory condition. They will then put all native interfaces into "admin down" state and start to perform BRSKI pledge functionality - and once a domain certificate is enrolled they automatically enable ACP.
Attempts for BRSKI pledge operations in greenfield state should terminate automatically when another method of configuring the node is used. Methods that indicate some form of physical possession of the device such as configuration via the serial console port could lead to immediate termination of BRSKI, while other parallel auto configuration methods subject to remote attacks might lead to BRSKI termination only after they were successful. Details of this may vary widely over different type of nodes. When BRSKI pledge operation terminates, this will automatically unset "ANI enable" and should terminate any temporarily needed state on the device to perform BRSKI - DULL GRASP, BRSKI pledge and any IPv6 configuration on interfaces.
Disabling ANI/ACP by undoing "ACP/ANI enable" is a risk for the reliable operations of the ACP if it can be executed by mistake or unauthorized. This behavior could be influenced through some additional (future) property in the certificate (e.g., in the domain information extension field): In an ANI deployment intended for convenience, disabling it could be allowed without further constraints. In an ANI deployment considered to be critical more checks would be required. One very controlled option would be to not permit these commands unless the domain certificate has been revoked or is denied renewal. Configuring this option would be a parameter on the BRSKI registrar(s). As long as the node did not receive a domain certificate, undoing "ANI/ACP enable" should not have any additional constraints.
Node-wide "ACP/ANI enable [up-if-only]" commands enable the operation of ACP/ANI. This is only auto-enabled on ANI greenfield devices, otherwise it must be configured explicitly.
If the option "up-if-only" is not selected, interfaces enabled for ACP/ANI interpret "down" state as "admin down" and not "physical down". In "admin-down" all non-ACP/ANI packets are filtered, but the physical layer is kept running to permit ACP/ANI to operate.
(New) commands that result in physical interruption ("physical down", "loopback") of ACP/ANI enabled interfaces should be built to protect continuance or reestablishment of ACP as much as possible.
Interface level "ACP/ANI enable" control per-interface operations. It is enabled by default on native interfaces and has to be configured explicitly on other interfaces.
Disabling "ACP/ANI enable" global and per-interface should have additional checks to minimize undesired breakage of ACP. The degree of control could be a domain wide parameter in the domain certificates.
There is no desirable configuration for the ACP. Instead, all parameters that need to be configured in support of the ACP are limitations of the solution, but they are only needed in cases where not all components are made autonomic. Whereever this is necessary, it relies on pre-existing mechanisms for configuration such as CLI or YANG ([RFC7950]) data models.
The most important examples of such configuration include:
Once the ACP is operating, any further configuration for the data-plane can be configured more reliably across the ACP itself because the ACP provides addressing and connectivity (routing) independent of the data-plane itself. For this, the configuration methods simply need to also allow to operate across the ACP VRF - netconf, ssh or any other method.
The ACP also provides additional security through its hop-by-hop encryption for any such configuration operations: Some legacy configuration methods (SNMP, TFTP, HTTP) may not use end-to-end encryption, and most of the end-to-end secured configuration methods still allow for easy passive observation along the path about configuration taking place (transport flows, port numbers, IP addresses).
The ACP can and should equally be used as the transport to configure any of the aforemention non-automic components of the ACP, but in that case, the same caution needs to be exercised as with data-plane configuration without ACP: Misconfiguration may cause the configuring entity to be disconnected from the node it configures - for example when incorrectly unconfiguring a remote ACP neighbor through which the configured ACP node is reached.
The ACP is self-healing:
The ACP can also sustain network partitions and mergers. Practically all ACP operations are link local, where a network partition has no impact. Nodes authenticate each other using the domain certificates to establish the ACP locally. Addressing inside the ACP remains unchanged, and the routing protocol inside both parts of the ACP will lead to two working (although partitioned) ACPs.
There are few central dependencies: A CRL may not be available during a network partition; a suitable policy to not immediately disconnect neighbors when no CRL is available can address this issue. Also, an ACP registrar or Certificate Authority might not be available during a partition. This may delay renewal of certificates that are to expire in the future, and it may prevent the enrollment of new nodes during the partition.
Highly resilient ACP designs can be built by using ACP registrars with embedded sub-CA, as outlined in Section 9.2.4. As long as a partition is left with one or more of such ACP registrars, it can continue to enroll new candidate ACP nodes as long as the ACP registrar's sub-CA certificate does not expire. Because the ACP addressing relies on unique Registrar-IDs, a later re-merge of partitions will also not cause problems with ACP addresses assigned during partitioning.
After a network partition, a re-merge will just establish the previous status, certificates can be renewed, the CRL is available, and new nodes can be enrolled everywhere. Since all nodes use the same trust anchor(s), a re-merge will be smooth.
Merging two networks with different trust anchors requires the ACP nodes to trust the union of Trust Anchors. As long as the routing-subdomain hashes are different, the addressing will not overlap, which only happens in the unlikely event of a 40-bit hash collision in SHA256 (see Section 6.10). Note that the complete mechanisms to merge networks is out of scope of this specification.
It is also highly desirable for implementation of the ACP to be able to run it over interfaces that are administratively down. If this is not feasible, then it might instead be possible to request explicit operator override upon administrative actions that would administratively bring down an interface across which the ACP is running. Especially if bringing down the ACP is known to disconnect the operator from the node. For example any such down administrative action could perform a dependency check to see if the transport connection across which this action is performed is affected by the down action (with default RPL routing used, packet forwarding will be symmetric, so this is actually possible to check).
As explained in Section 6, the ACP is based on secure channels built between nodes that have mutually authenticated each other with their domain certificates. The channels themselves are protected using standard encryption technologies such as DTLS or IPsec which provide additional authentication during channel establishment, data integrity and data confidentiality protection of data inside the ACP and in addition, provide replay protection.
An attacker will not be able to join the ACP unless having a valid domain certificate, also packet injection and sniffing traffic will not be possible due to the security provided by the encryption protocol.
The ACP also serves as protection (through authentication and encryption) for protocols relevant to OAM that may not have secured protocol stack options or where implementation or deployment of those options fail on some vendor/product/customer limitations. This includes protocols such as SNMP ([RFC3411]), NTP ([RFC5905]), PTP ([IEEE-1588-2008]), DNS ([RFC3596]), DHCPv6 ([RFC3315]), syslog ([RFC3164]), Radius ([RFC2865]), Diameter ([RFC6733]), TACACS ([RFC1492]), IPFIX ([RFC7011]), Netflow ([RFC3954]) - just to name a few. Protection via the ACP secure hop-by-hop channels for these protocols is meant to be only a stopgap though: The ultimate goal is for these and other protocols to use end-to-end encryption utilizing the domain certificate and rely on the ACP secure channels primarily for zero-touch reliable connectivity, but not primarily for security.
The remaining attack vector would be to attack the underlying ACP protocols themselves, either via directed attacks or by denial-of-service attacks. However, as the ACP is built using link-local IPv6 addresses, remote attacks from the data-plane are impossible as long as the data-plane has no facilities to remotely sent IPv6 link-local packets. The only exception are ACP connected interfaces which require higher physical protection. The ULA addresses are only reachable inside the ACP context, therefore, unreachable from the Data-Plane. Also, the ACP protocols should be implemented to be attack resistant and not consume unnecessary resources even while under attack.
The security model of the ACP is based on trusting all members of the group of nodes that receive an ACP domain certificate for the same domain. Attacks from the inside by a compromised group member are therefore the biggest challenge.
Group members must be protected against attackers so that there is no easy way to compromise them, or use them as a proxy for attacking other devices across the ACP. For example, management plane functions (transport ports) should only be reachable from the ACP but not the Data-Plane. Especially for those management plane functions that have no good protection by themselves because they do not have secure end-to-end transport and to whom ACP not only provides automatic reliable connectivity but also protection against attacks. Protection across all potential attack vectors is typically easier to do in devices whose software is designed from the ground up with security in mind than with legacy software based systems where the ACP is added on as another feature.
As explained above, traffic across the ACP SHOULD still be end-to-end encrypted whenever possible. This includes traffic such as GRASP, EST and BRSKI inside the ACP. This minimizes man in the middle attacks by compromised ACP group members. Such attackers cannot eavesdrop or modify communications, they can just filter them (which is unavoidable by any means).
See Appendix A.10.8 for further considerations how to avoid and deal with compromised nodes.
An ACP is self-forming, self-managing and self-protecting, therefore has minimal dependencies on the administrator of the network. Specifically, since it is (intended to be) independent of configuration, there is only limited scope for configuration errors on the ACP itself. The administrator may have the option to enable or disable the entire approach, but detailed configuration is not possible. This means that the ACP must not be reflected in the running configuration of nodes, except a possible on/off switch (and even that is undesirable).
While configuration (except for Section 8 and Section 9.2) is not possible, an administrator must have full visibility of the ACP and all its parameters, to be able to do trouble-shooting. Therefore, an ACP must support all show and debug options, as for any other network function. Specifically, a network management system or controller must be able to discover the ACP, and monitor its health. This visibility of ACP operations must clearly be separated from visibility of Data-Plane so automated systems will never have to deal with ACP aspects unless they explicitly desire to do so.
Since an ACP is self-protecting, a node not supporting the ACP, or without a valid domain certificate cannot connect to it. This means that by default a traditional controller or network management system cannot connect to an ACP. See Section 8.1.1 for more details on how to connect an NMS host into the ACP.
After seeding an ACP by configuring at least one ACP registrar with routing-subdomain and a CA, an ACP is self-protecting and there is no need to apply configuration to make it secure (typically the ACP Registrar doubles as EST server for certificate renewal). Its security therefore does not depend on configuration. This does not include workarounds for non-autonomic components as explained in Section 8. See Section 10.2 for details of how the ACP protects itself against attacks from the outside and to a more limited degree from the inside as well.
However, the security of the ACP depends on a number of other factors:
There is no prevention of source-address spoofing inside the ACP. This implies that if an attacker gains access to the ACP, it can spoof all addresses inside the ACP and fake messages from any other node.
The ACP is designed to enable automation of current network management and future autonomic peer-to-peer/distributed network automation. Any ACP member can send ACP IPv6 packet to other ACP members and announce via ACP GRASP services to all ACP members without depenency against centralized components.
The ACP relies on peer-to-peer authentication and authorization using ACP certificates. This security model is necessary to enable the autonomic ad-hoc any-to-any connectivity between ACP nodes. It provides infrastructure protection through hop by hop authentication and encryption - without relying on third parties. For any services where this complete autonomic peer-to-peer group security model is appropriate, the ACP domain certificate can also be used unchanged. For example for any type of data-plane routing protocol security.
This ACP security model is designed primarily to protect against attack from the outside, but not against attacks from the inside. To protect against spoofing attacks from compromised on-path ACP nodes, end-to-end encryption inside the ACP is used by new ACP signaling: GRASP across the ACP using TLS. The same is expected from any non-legacy services/protocols using the ACP. Because no group-keys are used, there is no risk for impacted nodes to access end-to-end encrypted traffic from other ACP nodes.
Attacks from impacted ACP nodes against the ACP are more difficult than against the data-plane because of the autoconfiguration of the ACP and the absence of configuration options that could be abused that allow to change/break ACP behavior. This is excluding configuration for workaround in support of non-autonomic components.
Mitigation against compromised ACP members is possible through standard automated certificate management mechanisms including revocation and non-renewal of short-lived certificates. In this version of the specification, there are no further optimization of these mechanisms defined for the ACP (but see Appendix A.10.8).
Higher layer service built using ACP domain certificates should not solely rely on undifferentiated group security when another model is more appropriate/more secure. For example central network configuration relies on a security model where only few especially trusted nodes are allowed to configure the data-plane of network nodes (CLIL, Netconf). This can be done through ACP domain certificates by differentiating them and introduce roles. See Appendix A.10.5.
Fundamentally, security depends on avoiding operator and network operations automation mistakes, implementation and architecture. Autonomic approaches such as the ACP largely eliminate operator mistakes and make it easier to recover from network operations mistakes. Implementation and architectural mistakes are still possible, as in all networking technologies.
Many details of ACP are designed with security in mind and discussed elsewhere in the document:
IPv6 addresses used by nodes in the ACP are covered as part of the node's domain certificate as described in Section 6.1.2. This allows even verification of ownership of a peer's IPv6 address when using a connection authenticated with the domain certificate.
The ACP acts as a security (and transport) substrate for GRASP inside the ACP such that GRASP is not only protected by attacks from the outside, but also by attacks from compromised inside attackers - by relying not only on hop-by-hop security of ACP secure channels, but adding end-to-end security for those GRASP messages. See Section 6.8.2.
ACP provides for secure, resilient zero-touch discovery of EST servers for certificate renewal. See Section 6.1.5.
ACP provides extensible, auto-configuring hop-by-hop protection of the ACP infrastructure via the negotiation of hop-by-hop secure channel protocols. See Section 6.5.
The ACP is designed to minimize attacks from the outside by minimizing its dependency against any non-ACP (Data-Plane) operations/configuration on a node. See also Section 6.12.2.
In combination with BRSKI, ACP enables a resilient, fully zero-touch network solution for short-lived certificates that can be renewed or re-enrolled even after unintentional expiry (e.g., because of interrupted connectivity). See Appendix A.2.
Because ACP secure channels can be long lived, but certificates used may be short lived, secure channels, for example built via IPsec need to be terminated when peer certificates expire. See Section 6.7.5.
The ACP is designed to minimize attacks from the outside by minimizing its dependency against any non-ACP (Data-Plane) operations/configuration on a node. See also Section 6.12.2.
Section 7.2 describes how to implement a routed ACP topology operating on what effectively is a large bridge-domain when using L3/L2 routers that operate at L2 in the data-plane. In this case, the ACP is subject to much higher likelyhood of attacks by other nodes "stealing" L2 addresses than in the actual routed case. Especially when the bridged network includes non-trusted devices such as hosts. This is a generic issue in L2 LANs. L2/L3 devices often already have some form of "port security" to prohibit this. They rely on NDP or DHCP learning of which port/MAC-address and IPv6 address belong together and block MAC/IPv6 source addresses from wrong ports. This type of function needs to be enabled to prohibit DoS attacks and specifically to protect the ACP. Likewise the GRASP DULL instance needs to ensure that the IPv6 address in the locator-option matches the source IPv6 address of the DULL GRASP packet.
This document defines the "Autonomic Control Plane".
The IANA is requested to register the value "AN_ACP" (without quotes) to the GRASP Objectives Names Table in the GRASP Parameter Registry. The specification for this value is this document, Section 6.3.
The IANA is requested to register the value "SRV.est" (without quotes) to the GRASP Objectives Names Table in the GRASP Parameter Registry. The specification for this value is this document, Section 6.1.5.
Explanation: This document chooses the initially strange looking format "SRV.<service-name>" because these objective names would be in line with potential future simplification of the GRASP objective registry. Today, every name in the GRASP objective registry needs to be explicitly allocated with IANA. In the future, this type of objective names could considered to be automatically registered in that registry for the same service for which <service-name> is registered according to [RFC6335]. This explanation is solely informational and has no impact on the requested registration.
The IANA is requested to create an ACP Parameter Registry with currently one registry table - the "ACP Address Type" table.
"ACP Address Type" Table. The value in this table are numeric values 0...3 paired with a name (string). Future values MUST be assigned using the Standards Action policy defined by [RFC8126]. The following initial values are assigned by this document:
0: ACP Zone Addressing Sub-Scheme (ACP RFC Figure 11) / ACP Manual Addressing Sub-Scheme (ACP RFC Section 6.10.4)
1: ACP Vlong Addressing Sub-Scheme (ACP RFC Section 6.10.5)
This work originated from an Autonomic Networking project at Cisco Systems, which started in early 2010. Many people contributed to this project and the idea of the Autonomic Control Plane, amongst which (in alphabetical order): Ignas Bagdonas, Parag Bhide, Balaji BL, Alex Clemm, Yves Hertoghs, Bruno Klauser, Max Pritikin, Michael Richardson, Ravi Kumar Vadapalli.
Special thanks to Brian Carpenter, Elwyn Davies, Joel Halpern and Sheng Jiang for their thorough reviews and to Pascal Thubert and Michael Richardson to provide the details for the recommendations of the use of RPL in the ACP.
Thanks to Valery Smyslov for review of the IPsec and IKEv2 configuration parameters.
Further input, review or suggestions were received from: Rene Struik, Brian Carpenter, Benoit Claise, William Atwood and Yongkang Zhang.
This text replaces the prior changelog with a summary to provide guidance for further IESG review.
Please see revision -21 for the individual changelogs of prior versions .
This document entered IESG review with version -13. It has since seen the following reviews:
IESG: Original owner/Yes: Terry Manderson (INT).
IESG: No Objection: Deborah Brungard (RTG), Alissa Cooper (GEN), Warren Kumari (OPS), Mirja Kühlewind (TSV), Alexey Melnikov (ART), Adam Roach (ART).
IESG: No Objection, not counted anymore as they have left IESG: Ben Campbell (ART), Spencer Dawkins (TSV).
IESG: Open DISCUSS hopefully resolved by this version: Eric Rescorla (SEC, left IESG), Benjamin Kaduk (SEC).
Other: Michael Richardson (WG), Brian Carpenter (WG), Pascal Thubert (WG), Frank Xialiang (WG), Elwyn Davies (GEN), Joel Halpern (RTGdir), Yongkang Zhang (WG), William Atwood (WG).
Only after ACP entered IESG review did it become clear that the in-progress BRSKI document would not provide all the explanations needed for ACP registrars as expected earlier by ACP authors. Instead, BRSKI will only specify a subset of required ACP behavior related to certificate handling and registrar. There, it became clear that the ACP draft should specify generic ACP registrar behavior independent of BRSKI so ACP could be implemented with or without BRSKI and any manual/proprietary or future standardized BRSKI alternatives (for example via NetConf) would understand the requirements for ACP registrars and its certificate handling.
This lead to additional text about ACP registrars in the ACP document:
1. Defined relationship ACP / ANI (ANI = ACP + BRSKI).
6.1.4 (new) Overview of trust points and trust anchors required for ACP.
6.1.5.5 Added explanations/requirements for Re-enrolment.
6.10.7 Normative requirements for ACP registrars (BRSKI or not).
10.2 Operational expectations against ACP registrars (BRSKI or not).
In addition to above ACP registrar / BRSKI related enhancements there is a range of minor normative (also explanatory) enhancements since the start of IESG review:
6.1.1 Hex digits in ACP domain information field now upper-case (no specific reason except that both options are equally good, but capitalized ones are used in rfc5234).
6.1.5.3 Added explanations about CRLs.
6.1.5.6 Added explanations of behavior under failing certificates.
6.1.2 Allow ACP adddress '0' in ACP domain information field: presence of address indicates permission to build ACP secure channel to node, 0 indicates that the address of the node is assigned by (future) other means than certificate. Non-autonomic nodes have no address at all (that was in -13), and can only connect via ACP connect interfaces to ACP.
6.1.3 Distinction of real ACP nodes (with address) and those with domain certificate without address added as a new rule to ACP domain membership check.
6.6 Added throttling of secure-channel setup attempts.
6.11.1.14 Removed requirement to handle unknown destination ACP traffic in low-end nodes that would never be RPL roots.
6.12.5 Added recommendation to use IPv6 DAD.
6.1.1, 6.7.1.1, 6.7.2, 6.7.3, 6.8.2 Various refined additional certificate, secure channel protocol (IPsec/IKEv2 and DTLS) and ACP GRASP TLS protocol parameter requirements to ensure interoperating implementations (from SEC-AD review).
Beyond the functional enhancements from the previous two sections, the mayority of changes since -13 are additional explanations from review feedback, textual nits and restructuring - with no functional requirement additions/changes.
1.1 Added "applicability and scope" section with summarized explanations.
2.Added in-band vs. out-of-band management definitions.
6.1.2 (was 6.1.1) expanded explanations of reasoning for elements of the ACP domain information field.
6.1.3 refined explanations of ACP domain membership check and justifications for it.
6.5 Elaborated step-by-step secure channel setup.
6.10 Additional explanations for addressing modes, additional table of addressing formats (thanks MichaelR).
6.10.5 introduced 'F' bit position as a better visual representation in the Vlong address space.
6.11.1.1 extensive overhaul to improve readability of use of RPL (from IESG feedback of non-routing/RPL experts).
6.12.2 Added caution about unconfiguring data-plane IPv6 addresses and impact to ACP (limitation of current ACP design, and pointint to more details in 10.2).
10.4 New explanations / summary of configurations for ACP (aka: all config is undesirable and only required for integrating with non-autonomic components, primarily ACP-connect and Registrars).
11. Textually enhanced / better structured security considerations section after IESG security review.
A. (new) Moved all explanations and discussions about futures from section 10 into this new appendix. This text should not be removed because it captures a lot of repeated asked questions in WG and during reviews and from users, and also captures ideas for some likely important followup work. But none of this is relevant to implementing (section 6) and operating (section 10) the ACP.
Leftover from -23 review by Eric Vyncke:
Swapping sections 9 and 10, section 9 was meant to be at end of document and summarize. Its not meant to be misinterpreted as introducing any new information. This did happen because section 10 was added after section 9.
Note: big rfcdiff of TOC is an rfcdiff bug, changes really minimal.
Review of IPsec security with Mcr and ipsec mailing list.
6.7.1 - new section: Moved general considerations for secure channel protocols here, refined them.
6.7.2 - new section: Moved common requirements for secure channel protocols here, refined them.
6.7.3.1.1. - improved requirements text related to RFC8221, better explamations re. HW acceleration issues.
6.7.3.1.2. - improved requirements text related to RFC8247, (some requirements still discussed to be redundant, will be finalized in next weeks.
Eric Vyncke review of -21:
Only noting most important changes, long list of smaller text/readability enhancements.
2. - New explanation of "normative" , "informational" section title tags. alphabetic reordering of terms, refined definitions for CA, CRL. root CA.
6.1.1. - explanation when IDevID parameters may be copied into LDevID.
6.1.2. - Fixed hex digits in ACP domain information to lower case.
6.1.3.1. - New section on Realtime clock and Time Validation.
6.3 - Added explanation that DTLS means >= version 1.2 (not only 1.2).
6.7 - New text in this main section explaing relationship of ACP secure channels and ACP virtual interfaces - with forward references to virtual interface section.
6.8.2 - reordered text and picture, no text change.
6.10.7.2 - describe first how Registrar-ID can be allocted for all type of registrars, then refined text for how to potentially use MAC addresses on physical registrars.
6.11.1.1 - Added text how this profile does not use data-plane artefacts (RPI) because hadware forwarding. This was previously hidden only later in the text.
6.11.1.13. - Rewrote RPL data-plane artefact text. Provide decoder ring for abbreviations and all relevant RFCs.
6.12.5.2. - Added more explicit text that secure channels are mapped into virtual interfaces, moved different type of interfaces used by ACP into seperate subsections to be able to refer to them.
7.2 - Rewrote/refined text for ACP on L2, prior text was confusing and did not well explain why ACP for L2/L3 switches can be implemented without any L2 (HW) changes. Also missing explanation of only running GRASP untagged when VLANs are used.
8.1.1 - Added requirement for ACP Edge nodes to allow configurable filtering of IPv6 RPI headers.
11. - (security section). Moved explanation of address stealing from 7.2 to here.
Ben Kaduk review of -21:
RFC822 encoding of ACP domain information:
6.1.2 rewrote text for explaining / justifying use of rfc822name as identifier for node CP in certificate (was discussed in thread, but badly written in prior versions).
6.1.2 Changed EBNF syntax to use "+" after rfcSELF because that is the known primary name to extensions separator in many email systems ("." was wrong in prior versions).
6.1.2 Rewrote/improved explanations for use of rfc822name field to explain better why it is PKIX compliant and the right thing to do.
Crypto parameters for IPsec:
6.1 - Added explanation of why manual keying for ACP is not feasible for ACP. Surprisingly, that text did not exist. Referred to by IPsec text (6.7.1), but here is the right place to describe the reasoning.
6.1.2 - Small textual refinement referring to requirements to authenticate peers (for the special cases of empty or '0' ACP address in ACP domain information field.
6.3 - To better justify Bens proposed change of secure channel protocol being IPsec vs. GRASP objective being IKEv2, better explained how protocol indicated in GRASP objective-value is name of protocol used to negotiate secure channel, use example of IKEv2 to negotiate IPsec.
6.7.1 - refinemenet similar to 6.3.
- moved new paragraph from Bens pull request up from 6.7.1.1 to 6.7.1 as it equally applies to GRE encapped IPsec (looks nicer one level up).
- created subsections 6.7.1.1 (IPsec/ESP) / 6.7.1.2 (IKEv2) to clearer distinguish between these two requirements blocks.
- Refined the text in these two sections to hopefully be a good answer to Valery's concern of not randomnly mocking with existing requirements docs (rfc8247 / rfc8221).
6.7.1.1.1 - IPsec/ESP requirements section:
- MUST support rfc8221 mandatory EXCEPT for the superceeding requirements in this section. Previously, this was not quite clear from the text.
- Hopefully persuasive explanations about the requirements levels for ENCR_AES_GCM_16, ENCR_AES_CBC, ENCR_AES_CCM_8 and ENCR_CHACHA20_POLY1305: Restructured text for why not ENCR_AES_CBC (was in prior version, just not well structured), added new expanations for ENCR_AES_CCM_8 and ENCR_CHACHA20_POLY130.
- In simple terms, requirements for ENCR_AES_CBC, ENCR_AES_CCM_8, ENCR_CHACHACHA are SHOULD when they are implementable with rqual or faster performancce than ENCR_AES_GCM_16.
- Removed text about "additional rfc8221" reqiurements MAY be used. Now the logic is that all other requirements apply. Hopefully we have written enough so that we prohibited downgrades.
6.7.1.1.2 - RFC8247 requirements:
- Added mandate to support rfc8247, added explanation that there is no "stripping down" requirement, just additional stronger requirements to mandate correct use of ACP certificartes during authentication.
- refined text on identifying ACP by IPv6 address to be clearer: Identifying in the context of IKEv2 and cases for '0' in ACP domain information.
- removed last two paragraphs about relationship to rfc8247, as his is now written in first paragraph of the section.
End of Ben Kaduk review related fixes.
Other:
Forgot to update example of ACP domain information to use capitalized hex-digits as required by HEXDIGIT used.
Added reference to RFC8316 (AN use-cases) to beginning of section 3 (ACP use cases).
Small Enhanced IPsec parameters description / requirements fixes (from Michael Richardson).
The following sections discuss additional background information about aspects of the normative parts of this document or associated mechanisms such as BRSKI (such as why specific choices were made by the ACP) and they provide discussion about possible future variations of the ACP.
This document defines the Zone, Vlong and Manual sub address schemes primarily to support address prefix assignment via distributed, potentially uncoordinated ACP registrars as defined in Section 6.10.7. This costs 48/46-bit identifier so that these ACP registrar can assign non-conflicting address prefixes. This design does not leave enough bits to simultaneously support a large number of nodes (Node-ID) plus a large prefix of local addresses for every node plus a large enough set of bits to identify a routing Zone. In result, Zone, Vlong 8/16 attempt to support all features, but in via separate prefixes.
In networks that always expect to rely on a centralized PMS as described above (Section 9.2.5), the 48/46-bits for the Registrar-ID could be saved. Such variations of the ACP addressing mechanisms could be introduced through future work in different ways. If the prefix rfcSELF in the ACP information field was changed, incompatible ACP variations could be created where every design aspect of the ACP could be changed. Including all addressing choices. If instead a new addressing sub-type would be defined, it could be a backward compatible extension of this ACP specification. Information such as the size of a zone-prefix and the length of the prefix assigned to the ACP node itself could be encoded via the extension field of the ACP domain information.
Note that an explicitly defined "Manual" addressing sub-scheme is always beneficial to provide an easy way for ACP nodes to prohibit incorrect manual configuration of any non-"Manual" ACP address spaces and therefore ensure that "Manual" operations will never impact correct routing for any non-"Manual" ACP addresses assigned via ACP domain certificates.
BRSKI describes how nodes with an IDevID certificate can securely and zero-touch enroll with an LDevID to support the ACP. BRSKI also leverages the ACP to enable zero-touch bootstrap of new nodes across networks without any configuration requirements across the transit nodes (e.g., no DHCP/DNS forwarding/server setup). This includes otherwise not configured networks as described in Section 3.2. Therefore BRSKI in conjunction with ACP provides for a secure and zero-touch management solution for complete networks. Nodes supporting such an infrastructure (BRSKI and ACP) are called ANI nodes (Autonomic Networking Infrastructure), see [I-D.ietf-anima-reference-model]. Nodes that do not support an IDevID but only an (insecure) vendor specific Unique Device Identifier (UDI) or nodes whose manufacturer does not support a MASA could use some future security reduced version of BRSKI.
When BRSKI is used to provision a domain certificate (which is called enrollment), the BRSKI registrar (acting as an enhanced EST server) must include the subjectAltName / rfc822Name encoded ACP address and domain name to the enrolling node (called pledge) via its response to the pledges EST CSR Attribute request that is mandatory in BRSKI.
The Certificate Authority in an ACP network must not change the subjectAltName / rfc822Name in the certificate. The ACP nodes can therefore find their ACP address and domain using this field in the domain certificate, both for themselves, as well as for other nodes.
The use of BRSKI in conjunction with the ACP can also help to further simplify maintenance and renewal of domain certificates. Instead of relying on CRL, the lifetime of certificates can be made extremely small, for example in the order of hours. When a node fails to connect to the ACP within its certificate lifetime, it cannot connect to the ACP to renew its certificate across it (using just EST), but it can still renew its certificate as an "enrolled/expired pledge" via the BRSKI bootstrap proxy. This requires only that the BRSKI registrar honors expired domain certificates and that the pledge attempts to perform TLS authentication for BRSKI bootstrap using its expired domain certificate before falling back to attempting to use its IDevID for BRSKI. This mechanism could also render CRLs unnecessary because the BRSKI registrar in conjunction with the CA would not renew revoked certificates - only a "Do-not-renew" list would be necessary on BRSKI registrars/CA.
In the absence of BRSKI or less secure variants thereof, provisioning of certificates may involve one or more touches or non-standardized automation. Node vendors usually support provisioning of certificates into nodes via PKCS#7 (see [RFC2315]) and may support this provisioning through vendor specific models via Netconf ([RFC6241]). If such nodes also support Netconf Zero-Touch ([RFC8572]) then this can be combined to zero-touch provisioning of domain certificates into nodes. Unless there are equivalent integration of Netconf connections across the ACP as there is in BRSKI, this combination would not support zero-touch bootstrap across a not configured network though.
This section discusses why GRASP DULL was chosen as the discovery protocol for L2 adjacent candidate ACP neighbors. The contenders considered where GRASP, mDNS or LLDP.
LLDP and Cisco's earlier Cisco Discovery Protocol (CDP) are example of L2 discovery protocols that terminate their messages on L2 ports. If those protocols would be chosen for ACP neighbor discovery, ACP neighbor discovery would therefore also terminate on L2 ports. This would prevent ACP construction over non-ACP capable but LLDP or CDP enabled L2 switches. LLDP has extensions using different MAC addresses and this could have been an option for ACP discovery as well, but the additional required IEEE standardization and definition of a profile for such a modified instance of LLDP seemed to be more work than the benefit of "reusing the existing protocol" LLDP for this very simple purpose.
Multicast DNNS (mDNS) [RFC6762] with DNS Service Discovery (DNS-SD) Resource Records (RRs) as defined in [RFC6763] is a key contender as an ACP discovery protocol. because it relies on link-local IP multicast, it does operates at the subnet level, and is also found in L2 switches. The authors of this document are not aware of mDNS implementation that terminate their mDNS messages on L2 ports instead of the subnet level. If mDNS was used as the ACP discovery mechanism on an ACP capable (L3)/L2 switch as outlined in Section 7, then this would be necessary to implement. It is likely that termination of mDNS messages could only be applied to all mDNS messages from such a port, which would then make it necessary to software forward any non-ACP related mDNS messages to maintain prior non-ACP mDNS functionality. Adding support for ACP into such L2 switches with mDNS could therefore create regression problems for prior mDNS functionality on those nodes. With low performance of software forwarding in many L2 switches, this could also make the ACP risky to support on such L2 switches.
LLDP was not considered because of the above mentioned issues. mDNS was not selected because of the above L2 mDNS considerations and because of the following additional points:
If mDNS was not already existing in a node, it would be more work to implement than DULL GRASP, and if an existing implementation of mDNS was used, it would likely be more code space than a separate implementation of DULL GRASP or a shared implementation of DULL GRASP and GRASP in the ACP.
This section motivates why RPL - "IPv6 Routing Protocol for Low-Power and Lossy Networks ([RFC6550] was chosen as the default (and in this specification only) routing protocol for the ACP. The choice and above explained profile was derived from a pre-standard implementation of ACP that was successfully deployed in operational networks.
Requirements for routing in the ACP are:
IP multicast is not used by the ACP because the ANI (Autonomic Networking Infrastructure) itself does not require IP multicast but only service announcement/discovery. Using IP multicast for that would have made it necessary to develop a zero-touch auto configuring solution for ASM (Any Source Multicast - the original form of IP multicast defined in [RFC1112]), which would be quite complex and difficult to justify. One aspect of complexity where no attempt at a solution has been described in IETF documents is the automatic-selection of routers that should be PIM Sparse Mode (PIM-SM) Rendezvous Points (RPs) (see [RFC7761]). The other aspects of complexity are the implementation of MLD ([RFC4604]), PIM-SM and Anycast-RP (see [RFC4610]). If those implementations already exist in a product, then they would be very likely tied to accelerated forwarding which consumes hardware resources, and that in return is difficult to justify as a cost of performing only service discovery.
Some future ASA may need high performance in-network data replication. That is the case when the use of IP multicast is justified. Such an ASA can then use service discovery from ACP GRASP, and then they do not need ASM but only SSM (Source Specific Multicast, see [RFC4607]) for the IP multicast replication. SSM itself can simply be enabled in the Data-Plane (or even in an update to the ACP) without any other configuration than just enabling it on all nodes and only requires a simpler version of MLD (see [RFC5790]).
LSP (Link State Protocol) based IGP routing protocols typically have a mechanism to flood information, and such a mechanism could be used to flood GRASP objectives by defining them to be information of that IGP. This would be a possible optimization in future variations of the ACP that do use an LSP routing protocol. Note though that such a mechanism would not work easily for GRASP M_DISCOVERY messages which are intelligently (constrained) flooded not across the whole ACP, but only up to a node where a responder is found. We do expect that many future services in ASA will have only few consuming ASA, and for those cases, M_DISCOVERY is the more efficient method than flooding across the whole domain.
Because the ACP uses RPL, one desirable future extension is to use RPLs existing notion of DODAG, which are loop-free distribution trees, to make GRASP flooding more efficient both for M_FLOOD and M_DISCOVERY. See Section 6.12.5 how this will be specifically beneficial when using NBMA interfaces. This is not currently specified in this document because it is not quite clear yet what exactly the implications are to make GRASP flooding depend on RPL DODAG convergence and how difficult it would be to let GRASP flooding access the DODAG information.
[RFC Editor: This section to be removed before RFC.
[This section kept for informational purposes up until the last draft version as that would be the version that readers interested in the changelog would also go to to revisit it.]
The mechanism described in the normative part of this document to support multiple different ACP secure channel protocols without a single network wide MTI protocol is important to allow extending secure ACP channel protocols beyond what is specified in this document, but it will run into problem if it would be used for multiple protocols:
The need to potentially have multiple of these security associations even temporarily run in parallel to determine which of them works best does not support the most lightweight implementation options.
The simple policy of letting one side (Alice) decide what is best may not lead to the mutual best result.
The two limitations can easier be solved if the solution was more modular and as few as possible initial secure channel negotiation protocols would be used, and these protocols would then take on the responsibility to support more flexible objectives to negotiate the mutually preferred ACP security channel protocol.
IKEv2 is the IETF standard protocol to negotiate network security associations. It is meant to be extensible, but it is unclear whether it would be feasible to extend IKEv2 to support possible future requirements for ACP secure channel negotiation:
Consider the simple case where the use of native IPsec vs. IPsec via GRE is to be negotiated and the objective is the maximum throughput. Both sides would indicate some agreed upon performance metric and the preferred encapsulation is the one with the higher performance of the slower side. IKEv2 does not support negotiation with such objectives.
Consider DTLS and some form of MacSec are to be added as negotiation options - and the performance objective should work across all IPsec, DTLS and MacSec options. In the case of MacSEC, the negotiation would also need to determine a key for the peering. It is unclear if it would be even appropriate to consider extending the scope of negotiation in IKEv2 to those cases. Even if feasible to define, it is unclear if implementations of IKEv2 would be eager to adopt those type of extension given the long cycles of security testing that necessarily goes along with core security protocols such as IKEv2 implementations.
A more modular alternative to extending IKEv2 could be to layer a modular negotiation mechanism on top of the multitude of existing or possible future secure channel protocols. For this, GRASP over TLS could be considered as a first ACP secure channel negotiation protocol. The following are initial considerations for such an approach. A full specification is subject to a separate document:
To explicitly allow negotiation of the ACP channel protocol, GRASP over a TLS connection using the GRASP_LISTEN_PORT and the node's and peer's link-local IPv6 address is used. When Alice and Bob support GRASP negotiation, they do prefer it over any other non-explicitly negotiated security association protocol and should wait trying any non-negotiated ACP channel protocol until after it is clear that GRASP/TLS will not work to the peer.
When Alice and Bob successfully establish the GRASP/TSL session, they will negotiate the channel mechanism to use using objectives such as performance and perceived quality of the security. After agreeing on a channel mechanism, Alice and Bob start the selected Channel protocol. Once the secure channel protocol is successfully running, the GRASP/TLS connection can be kept alive or timed out as long as the selected channel protocol has a secure association between Alice and Bob. When it terminates, it needs to be re-negotiated via GRASP/TLS.
Notes:
There is a wide range of setting up different ACP solution by appropriately using CAs and the domain and rsub elements in the domain information field of the domain certificate. We summarize these options here as they have been explained in different parts of the document in before and discuss possible and desirable extensions:
An ACP domain is the set of all ACP nodes using certificates from one or more Trust Anchors (typically one CA) using the same domain field. GRASP inside the ACP is run across all transitively connected ACP nodes in a domain.
The rsub element in the domain information field permits the use of addresses from different ULA prefixes. One use case is to create multiple physical networks that initially may be separated with one ACP domain but different routing subdomains, so that all nodes can mutual trust their ACP domain certificates (not depending on rsub) and so that they could connect later together into a contiguous ACP network.
One instance of such a use case is an ACP for regions interconnected via a non-ACP enabled core, for example due to the absence of product support for ACP on the core nodes. ACP connect configurations as defined in this document can be used to extend and interconnect those ACP islands to the NOC and merge them into a single ACP when later that product support gap is closed.
Note that RPL scales very well. It is not necessary to use multiple routing subdomains to scale ACP domains in a way it would be possible if other routing protocols where used. They exist only as options for the above mentioned reasons.
If different ACP domains are to be created that should not allow to connect to each other by default, these ACP domains simply need to have different domain elements in the domain information field. These domain elements can be arbitrary, including subdomains of one another: Domains "example.com" and "research.example.com" are separate domains if both are domain elements in the domain information element of certificates.
It is not necessary to have a separate CA for different ACP domains: an operator can use a single CA to sign certificates for multiple ACP domains that are not allowed to connect to each other because the checks for ACP adjacencies includes comparison of the domain part.
If multiple independent networks choose the same domain name but had their own CA, these would not form a single ACP domain because of CA mismatch. Therefore there is no problem in choosing domain names that are potentially also used by others. Nevertheless it is highly recommended to use domain names that one can have high probability to be unique. It is recommended to use domain names that start with a DNS domain names owned by the assigning organization and unique within it. For example "acp.example.com" if you own "example.com".
Intent is the architecture component of autonomic networks according to [I-D.ietf-anima-reference-model] that allows operators to issue policies to the network. Its applicability for use is quite flexible and freeform, with potential applications including policies flooded across ACP GRASP and interpreted on every ACP node.
One concern for future definitions of Intent solutions is the problem of circular dependencies when expressing Intent policies about the ACP itself.
For example, Intent could indicate the desire to build an ACP across all domains that have a common parent domain (without relying on the rsub/routing-subdomain solution defined in this document). For example ACP nodes with domain "example.com", "access.example.com", "core.example.com" and "city.core.example.com" should all establish one single ACP.
If each domain has its own source of Intent, then the Intent would simply have to allow adding the peer domains TA and domain names to the parameters for the ACP domain membership check (Section 6.1.3) so that nodes from those other domains are accepted as ACP peers.
If this Intent was to be originated only from one domain, it could likely not be made to work because the other domains will not build any ACP connection amongst each other, whether they use the same or different CA due to the ACP domain membership check.
If the domains use the same CA one could change the ACP setup to permit for the ACP to be established between two ACP nodes with different acp-domain-names, but only for the purpose of disseminating limited information, such as Intent, but not to set up full ACP connectivity, specifically not RPL routing and passing of arbitrary GRASP information. Unless the Intent policies permit this to happen across domain boundaries.
This type of approach where the ACP first allows Intent to operate and only then sets up the rest of ACP connectivity based on Intent policy could also be used to enable Intent policies that would limit functionality across the ACP inside a domain, as long as no policy would disturb the distribution of Intent. For example to limit reachability across the ACP to certain type of nodes or locations of nodes.
The ACP as specified in this document is very explicit about the choice of options to allow interoperable implementations. The choices made may not be the best for all environments, but the concepts used by the ACP can be used to build derived solutions:
The ACP specifies the use of ULA and deriving its prefix from the domain name so that no address allocation is required to deploy the ACP. The ACP will equally work not using ULA but any other /48 IPv6 prefix. This prefix could simply be a configuration of the ACP registrars (for example when using BRSKI) to enroll the domain certificates - instead of the ACP registrar deriving the /48 ULA prefix from the AN domain name.
Some solutions may already have an auto-addressing scheme, for example derived from existing unique device identifiers (e.g., MAC addresses). In those cases it may not be desirable to assign addresses to devices via the ACP address information field in the way described in this document. The certificate may simply serve to identify the ACP domain, and the address field could be empty/unused. The only fix required in the remaining way the ACP operate is to define another element in the domain certificate for the two peers to decide who is Alice and who is Bob during secure channel building. Note though that future work may leverage the acp address to authenticate "ownership" of the address by the device. If the address used by a device is derived from some pre-existing permanent local ID (such as MAC address), then it would be useful to store that address in the certificate using the format of the access address information field or in a similar way.
The ACP is defined as a separate VRF because it intends to support well managed networks with a wide variety of configurations. Therefore, reliable, configuration-indestructible connectivity cannot be achieved from the Data-Plane itself. In solutions where all transit connectivity impacting functions are fully automated (including security), indestructible and resilient, it would be possible to eliminate the need for the ACP to be a separate VRF. Consider the most simple example system in which there is no separate Data-Plane, but the ACP is the Data-Plane. Add BRSKI, and it becomes a fully autonomic network - except that it does not support automatic addressing for user equipment. This gap can then be closed for example by adding a solution derived from [I-D.ietf-anima-prefix-management].
TCP/TLS as the protocols to provide reliability and security to GRASP in the ACP may not be the preferred choice in constrained networks. For example, CoAP/DTLS (Constrained Application Protocol) may be preferred where they are already used, allowing to reduce the additional code space footprint for the ACP on those devices. Hop-by-hop reliability for ACP GRASP messages could be made to support protocols like DTLS by adding the same type of negotiation as defined in this document for ACP secure channel protocol negotiation. End-to-end GRASP connections can be made to select their transport protocol in future extensions of the ACP meant to better support constrained devices by indicating the supported transport protocols (e.g.: TLS/DTLS) via GRASP parameters of the GRASP objective through which the transport endpoint is discovered.
The routing protocol RPL used for the ACP does explicitly not optimize for shortest paths and fastest convergence. Variations of the ACP may want to use a different routing protocol or introduce more advanced RPL profiles.
Variations such as what routing protocol to use, or whether to instantiate an ACP in a VRF or (as suggested above) as the actual Data-Plane, can be automatically chosen in implementations built to support multiple options by deriving them from future parameters in the certificate. Parameters in certificates should be limited to those that would not need to be changed more often than certificates would need to be updated anyhow; Or by ensuring that these parameters can be provisioned before the variation of an ACP is activated in a node. Using BRSKI, this could be done for example as additional follow-up signaling directly after the certificate enrollment, still leveraging the BRSKI TLS connection and therefore not introducing any additional connectivity requirements.
Last but not least, secure channel protocols including their encapsulations are easily added to ACP solutions. ACP hop-by-hop network layer secure channels could also be replaced by end-to-end security plus other means for infrastructure protection. Any future network OAM should always use end-to-end security anyhow and can leverage the domain certificates and is therefore not dependent on security to be provided for by ACP secure channels.
Routing in the ACP according to this specification only leverages the standard RPL mechanism of route optimization, e.g. keeping only routes that are not towards the RPL root. This is known to scale to networks with 20,000 or more nodes. There is no auto-aggregation of routes for /48 ULA prefixes (when using rsub in the domain information field) and/or Zone-ID based prefixes.
Automatic assignment of Zone-ID and auto-aggregation of routes could be achieved for example by configuring zone-boundaries, announcing via GRASP into the zones the zone parameters (zone-ID and /48 ULA prefix) and auto-aggegating routes on the zone-boundaries. Nodes would assign their Zone-ID and potentially even /48 prefix based on the GRASP announcements.
As described in Section 6.12.2, the ACP depends on the Data-Plane to establish IPv6 link-local addressing on interfaces. Using a separate MAC address for the ACP allows to fully isolate the ACP from the data-plane in a way that is compatible with this specification. It is also an ideal option when using Single-root input/output virtualization (SR-IOV - see https://en.wikipedia.org/wiki/Single-root_input/output_virtualization) in an implementation to isolate the ACP because different SR-IOV interfaces use different MAC addresses.
When additional MAC address(es) are not available, separation of the ACP could be done at different demux points. The same subnet interface could have a separate IPv6 interface for the ACP and Data-Plane and therefore separate link-local addresses for both, where the ACP interface is non-configurable on the Data-Plane. This too would be compatible with this specification and not impact interoperability.
An option that would require additional specification is to use a different Ethertype from 0x86DD (IPv6) to encapsulate IPv6 packets for the ACP. This would be a similar approach as used for IP authentication packets in [IEEE-802.1X] which use the Extensible Authentication Protocol over Local Area Network (EAPoL) ethertype (0x88A2).
Note that in the case of ANI nodes, all the above considerations equally apply to the encapsulation of BRSKI packets including GRASP used for BRSKI.
Future work should define YANG ([RFC7950]) data model and/or node internal APIs to monitor and manage the ACP.
Support for the ACP Adjacency Table (Section 6.2) and ACP GRASP need to be included into such model/API.
..... USA ...... ..... Europe ......
NOC1 NOC2
| |
| metric 100 |
ACP1 --------------------------- ACP2 .
| | . WAN
| metric 10 metric 20 | . Core
| | .
ACP3 --------------------------- ACP4 .
| metric 100 |
| | .
| | . Sites
ACP10 ACP11 .
Figure 18: Dual NOC
The profile for RPL specified in this document builds only one spanning-tree path set to a root, typically a registrar in one NOC. In the presence of multiple NOCs, routing toward the non-root NOCs may be suboptimal. Figure 18 shows an extreme example. Assuming that node ACP1 becomes the RPL root, traffic between ACP11 and NOC2 will pass through ACP4-ACP3-ACP1-ACP2 instead of ACP4-ACP2 because the RPL calculated DODAG/routes are shortest paths towards the RPL root.
To overcome these limitations, extensions/modifications to the RPL profile can provide optimality for multiple NOCs. This requires utilizing Data-Plane artifact including IPinIP encap/decap on ACP routers and processing of IPv6 RPI headers. Alternatively, (Src,Dst) routing table entries could be used.
Flooding of ACP GRASP messages can be further constrained and therefore optimized by flooding only via links that are part of the RPL DODAG.
ACP connect is an explicit mechanism to "leak" ACP traffic explicitly (for example in a NOC). It is therefore also a possible security gap when it is easy to enable ACP connect on arbitrary compromised ACP nodes.
One simple solution is to define an extension in the ACP certificates ACP information field indicating the permission for ACP connect to be configured on that ACP node. This could similarly be done to decide whether a node is permitted to be a registrar or not.
Tying the permitted "roles" of an ACP node to the ACP domain certificate provides fairly strong protection against misconfiguration, but is still subject to code modifications.
Another interesting role to assign to certificates is that of a NOC node. This would allow to limit certain type of connections such as OAM TLS connections to only NOC initiator or responders.
In this specification, the ACP can only establish autonomic connectivity across L2 hops and only explicitly configured options to tunnel across L3. Future work should specify mechanisms to automatically tunnel ACP across L3 networks. A hub&spoke option would allow to tunnel across the Internet to a cloud or central instance of the ACP, a peer-to-peer tunneling mechanism could tunnel ACP islands across an L3VPN infrastructure.
Section 9.1 describes diagnostics options that can be done without changing the external, interoperability affecting characteristics of ACP implementations.
Even better diagnostics of ACP operations is possible with additional signaling extensions, such as:
Compromised ACP nodes pose the biggest risk to the operations of the network. The most common type of compromise is leakage of credentials to manage/configure the device and the application of malicious configuration including the change of access credentials, but not the change of software. Most of todays networking equipment should have secure boot/software infrastructure anyhow, so attacks that introduce malicious software should be a lot harder.
The most important aspect of security design against these type of attacks is to eliminate password based configuration access methods and instead rely on certificate based credentials handed out only to nodes where it is clear that the private keys can not leak. This limits unexpected propagation of credentials.
If password based credentials to configure devices still need to be supported, they must not be locally configurable, but only be remotely provisioned or verified (through protocols like Radius or Diameter), and there must be no local configuration permitting to change these authentication mechanisms, but ideally they should be autoconfiguring across the ACP. See [I-D.eckert-anima-noc-autoconfig].
Without physical access to the compromised device, attackers with access to configuration should not be able to break the ACP connectivity, even when they can break or otherwise manipulate (spoof) the data-plane connectivity through configuration. To achieve this, it is necessary to avoid providing configuration options for the ACP, such as enabling/disabling it on interfaces. For example there could be an ACP configuration that locks down the current ACP config unless factory reseet is done.
With such means, the valid administration has the best chances to maintain access to ACP nodes, discover malicious configuration though ongoing configuration tracking from central locations for example, and to react accordingly.
The primary reaction is withdrawal/change of credentials, terminate malicious existing management sessions and fixing the configuration. Ensuring that management sessions using invalidated credentials are terminated automatically without recourse will likely require new work.
Only when these steps are not feasible would it be necessary to revoke or expire the ACP domain certificate credentials and consider the node kicked off the network - until the situation can be further rectified, likely requiring direct physical access to the node.
Without extensions, compromised ACP nodes can only be removed from the ACP at the speed of CRL/OCSP information refresh or expiry (and non-removal) of short lived certificates. Future extensions to the ACP could for example use GRASP flooding distribution of triggered updates of CRL/OCSP or explicit removal indication of the compromised nodes domain certificate.