| Dynamic Host Configuration (DHC) | B. Volz |
| Internet-Draft | Cisco |
| Intended status: Standards Track | T. Mrugalski |
| Expires: December 4, 2020 | ISC |
| CJ. Bernardos | |
| UC3M | |
| June 2, 2020 |
Link-Layer Addresses Assignment Mechanism for DHCPv6
draft-ietf-dhc-mac-assign-07
In certain environments, e.g. large scale virtualization deployments, new devices are created in an automated manner. Such devices typically have their link-layer (MAC) addresses randomized. With sufficient scale, the likelihood of collision is not acceptable. Therefore an allocation mechanism is required. This draft proposes an extension to DHCPv6 that allows a scalable approach to link-layer address assignments.
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 December 4, 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.
There are several new deployment types that deal with a large number of devices that need to be initialized. One of them is a scenario where virtual machines (VMs) are created on a massive scale. Typically the new VM instances are assigned a random link-layer (MAC) address, but that does not scale well due to the birthday paradox. Another use case is IoT (Internet of Things) devices (see [RFC7228]). The huge number of such devices would likely exhaust a vendor's OUI (Organizationally Unique Identifier) global address space, and while there is typically no need to provide global uniqueness for such devices, a link-layer assignment mechanism allows for conflicts to be avoided inside an administrative domain. For those reasons, it is desired to have some form of authority that would be able to assign locally unique MAC addresses.
This document proposes a new mechanism that extends DHCPv6 operation to handle link-layer address assignments.
Since DHCPv6 ([RFC8415]) is a protocol that can allocate various types of resources (non-temporary addresses, temporary addresses, prefixes, as well as many options) and has the necessary infrastructure to maintain such allocations (numerous server and client implementations, large deployed relay infrastructure, and supportive solutions such as leasequery and failover), it is a good candidate to address the desired functionality.
While this document presents a design that should be usable for any link-layer address type, some of the details are specific to Ethernet / IEEE 802 48-bit MAC addresses. Future documents may provide specifics for other link-layer address types.
The IEEE originally set aside half of the 48-bit MAC Address space for local use (where the U/L bit is set to 1). In 2017, the IEEE published an optional specification ([IEEEStd802c-2017]) that divides this space into quadrants (Standards Assigned Identifier, Extended Local Identifier, Administratively Assigned Identifier, and a Reserved quadrant) - more details are in Appendix A.
The IEEE is also working to specify protocols and procedures for assignment of locally unique addresses (IEEE 802.1cq). This work may serve as one such protocol for assignment. For additional background, see [IEEE-802-Tutorial].
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.
The DHCP terminology relevant to this specification from [RFC8415] applies here.
This mechanism is designed to be generic and usable in many deployments, but there are two scenarios it attempts to address in particular: (i) proxy client mode, and (ii) direct client mode.
This mode is used when an entity acts as a DHCP client and requests available DHCP servers to assign one or more addresses (an address block), to be then assigned for use by the final end-devices. Large-scale virtualization is one application scenario for proxy client mode. In such environments the governing entity is often called a hypervisor and is frequently required to spawn new VMs. The hypervisor needs to assign new addresses to those machines. The hypervisor does not use those addresses for itself, but rather uses them to create new VMs with appropriate addresses. It is worth pointing out the cumulative nature of this scenario. Over time, the hypervisor is likely to increase its address usage. Some obsolete VMs will be deleted and their addresses will be eligible for reuse for new VMs.
This mode can be used when an entity acts as a DHCP client and requests available DHCP servers to assign one or more addresses (an address block) for its use. This usage scenario is related to IoT, as described earlier (see Section 1). Upon first boot, the device uses a temporary address, as described in [IEEE-802.11-02-109r0], to send initial DHCP packets to available DHCP servers. Then, such devices would typically request a single address for each available network interface, which typically means one address per device. Once the server assigns an address, the device abandons its temporary address and uses the assigned (leased) address.
Note that a client that operates as above that does not have a globally unique link-layer address on any of its interfaces MUST NOT use a link-layer based DUID (DHCP Unique Identifier), i.e., DUID-LLT or DUID-LL, for its client identifier. As of this writing, this means such a device MUST use a DUID-EN or DUID-UUID (though new DUID types may be defined in the future). For more details, refer to Section 11 of [RFC8415].
Also, a client that operates as above may run into issues if the switch it is connected to prohibits or restricts link-layer address changes. This may limit where this capability can be used, or may require the administrator to adjust the configuration of the switch(es) to allow a change in address.
In all scenarios the protocol operates in fundamentally the same way. The device requesting an address, acting as a DHCP client, will send a Solicit message with a IA_LL option to all available DHCP servers. That IA_LL option MUST include a LLADDR option specifying the link-layer-type and link-layer-len and may specify a specific address or address block as a hint for the server. Each available server responds with an Advertise message with offered address or addresses. The client then picks the best server, as governed by [RFC8415], and will send a Request message. The target server will then assign the addresses and send a Reply message. Upon reception, the client can start using those addresses.
Normal DHCP mechanisms are in use. The client is expected to periodically renew the addresses as governed by T1 and T2 timers. This mechanism can be administratively disabled by the server sending "infinity" as the T1 and T2 values (see Section 7.7 of [RFC8415]).
The client can release addresses when they are no longer needed by sending a Release message (see Section 18.2.7 of [RFC8415]).
Figure 1, taken from [RFC8415], shows a timeline diagram of the messages exchanged between a client and two servers for the typical lifecycle of one or more leases
Server Server
(not selected) Client (selected)
v v v
| | |
| Begins initialization |
| | |
start of | _____________/|\_____________ |
4-message |/ Solicit | Solicit \|
exchange | | |
Determines | Determines
configuration | configuration
| | |
|\ | ____________/|
| \________ | /Advertise |
| Advertise\ |/ |
| \ | |
| Collects Advertises |
| \ | |
| Selects configuration |
| | |
| _____________/|\_____________ |
|/ Request | Request \|
| | |
| | Commits configuration
| | |
end of | | _____________/|
4-message | |/ Reply |
exchange | | |
| Initialization complete |
| | |
. . .
. . .
| T1 (Renewal) Timer Expires |
| | |
2-message | _____________/|\_____________ |
exchange |/ Renew | Renew \|
| | |
| | Commits extended lease(s)
| | |
| | _____________/|
| |/ Reply |
. . .
. . .
| | |
| Graceful shutdown |
| | |
2-message | _____________/|\_____________ |
exchange |/ Release | Release \|
| | |
| | Discards lease(s)
| | |
| | _____________/|
| |/ Reply |
| | |
v v v
Figure 1: Timeline diagram of the messages exchanged between a client and two servers for the typical lifecycle of one or more leases
Confirm, Decline, and Information-Request messages are not used in link-layer address assignment.
Clients implementing this mechanism SHOULD use the Rapid Commit option as specified in Section 5.1 and 18.2.1 of [RFC8415].
An administrator may make the address assignment permanent by specifying use of infinite lifetimes, as defined in Section 7.7 of [RFC8415].
Devices supporting this proposal MAY support the reconfigure mechanism, as defined in Section 18.2.11 of [RFC8415]. If supported by both server and client, this mechanism allows the administrator to immediately notify clients that the configuration has changed and triggers retrieval of relevant changes immediately, rather than after the T1 timer elapses. Since this mechanism requires implementation of Reconfigure Key Authentication Protocol (See Section 20.4 of [RFC8415]), small-footprint devices may choose to not support it.
One of the essential aspects of this mechanism is its cumulative nature, especially in the hypervisor scenario. The server-client relationship does not look like other DHCP transactions. This is especially true in the hypervisor scenario. In a typical environment, there would be one server and a rather small number of hypervisors, possibly even only one. However, over time the number of addresses requested by the hypervisor(s) will likely increase as new VMs are spawned.
Another aspect crucial for efficient design is the observation that a single client acting as hypervisor will likely use thousands of addresses. Therefore an approach similar to what is used for IPv6 address or prefix assignment (IA container with all assigned addresses listed, one option for each address) would not work well. Therefore the mechanism should operate on address blocks, rather than single values. A single address can be treated as an address block with just one address.
The DHCP mechanisms are reused to large degree, including message and option formats, transmission mechanisms, relay infrastructure and others. However, a device wishing to support only link-layer address assignment is not required to support full DHCP. In other words, the device may support only assignment of link-layer addresses, but not IPv6 addresses or prefixes.
A client MUST send a LLADDR option encapsulated in an IA_LL option to specify the link-layer-type and link-layer-len values. For link-layer-type 1 (Ethernet / IEEE 802 48-bit MAC addresses), a client sets the link-layer-address field to:
A client sets the extra-addresses field to either 0 for a single address or the size of the requested address block minus 1.
A client SHOULD set the valid-lifetime field to 0 (this field MUST be ignored by the server).
The addresses are assigned in blocks. The smallest block is a single address. To request an assignment, the client sends a Solicit message with an IA_LL option in the message. The IA_LL option MUST contain a LLADDR option as specified in Section 6.
The server, upon receiving an IA_LL option, inspects its content and may offer an address or addresses for each LLADDR option according to its policy. The server MAY take into consideration the address block requested by the client in the LLADDR option. However, the server MAY chose to ignore some or all parameters of the requested address block. In particular, the server may send a different starting address than requested, or grant a smaller number of addresses than requested. The server sends back an Advertise message an IA_LL option containing an LLADDR option that specifies the addresses being offered. If the server is unable to provide any addresses it MUST return the IA_LL option containing a Status Code option (see Section 21.13 of [RFC8415]) with status set to NoAddrsAvail.
The client MUST be able to handle an Advertise message containing a smaller number of addresses, or an address or addresses different from those requested.
The client waits for available servers to send Advertise responses and picks one server as defined in Section 18.2.9 of [RFC8415]. The client then sends a Request message that includes the IA_LL container option with the LLADDR option copied from the Advertise message sent by the chosen server.
Upon reception of a Request message with IA_LL container option, the server assigns requested addresses. The server allocates block of addresses according to its configured policy. The server MAY assign a different block than requested in the Request message. It then generates and sends a Reply message back to the client.
Upon receiving a Reply message, the client parses the IA_LL container option and may start using all provided addresses. It MUST restart its T1 and T2 timers using the values specified in the IA_LL option.
The client MUST be able to handle a Reply message containing a smaller number of addresses, or an address or addresses different from those requested.
A client that has included a Rapid Commit option in the Solicit, may receive a Reply in response to the Solicit and skip the Advertise and Request steps above (see Section 18.2.1 of [RFC8415]).
A client that changes its link-layer address on an interface SHOULD follow the recommendations in Section 7.2.6 of [RFC4861] to inform its neighbors of the new link-layer address quickly.
Address renewals follow the normal DHCP renewals processing described in Section 18.2.4 of [RFC8415]. Once the T1 timer elapses, the client starts sending Renew messages with the IA_LL option containing a LLADDR option for the address block being renewed. The server responds with a Reply message that contains the renewed address block. The server MUST NOT shrink or expand the address block - once a block is assigned and has a non-zero valid lifetime, its size, starting address, and ending address MUST NOT change.
If the requesting client needs additional addresses -- e.g., in the hypervisor scenario because addresses need to be assigned to new VMs -- the simpler approach is for the requesting device to keep the address blocks as atomic once "leased". Therefore, if a client wants more addresses at a later stage, it SHOULD send an IA_LL option with a different IAID to create another "container" for more addresses.
If the client is unable to Renew before the T2 timer elapses, it starts sending Rebind messages as described in 18.2.5 of [RFC8415].
The client may decide to release a leased address block. A client MUST release the whole block in its entirety. A client releases an address block by sending a Release message that includes the IA_LL option containing the LLADDR option for the address block to release. The Release transmission mechanism is described in Section 18.2.7 of [RFC8415].
This mechanism uses an approach similar to the existing mechanisms in DHCP. There is one container option (the IA_LL option) that contains the actual address or addresses, represented by an LLADDR option. Each such option represents an address block, which is expressed as a first address with a number that specifies how many additional addresses are included.
The Identity Association for Link-Layer Addresses option (IA_LL option) is used to carry one or more IA_LL options, the parameters associated with the IA_LL, and the address blocks associated with the IA_LL.
The format of the IA_LL option is:
0 1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| OPTION_IA_LL | option-len |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| IAID (4 octets) |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| T1 (4 octets) |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| T2 (4 octets) |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
. .
. IA_LL-options .
. .
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
Figure 2: IA_LL Option Format
An IA_LL option may only appear in the options area of a DHCP message. A DHCP message may contain multiple IA_LL options (though each must have a unique IAID).
The status of any operations involving this IA_LL is indicated in a Status Code option (see Section 21.13 of [RFC8415]) in the IA_LL-options field.
Note that an IA_LL has no explicit "lifetime" or "lease length" of its own. When the valid lifetimes of all of the addresses in an IA_LL have expired, the IA_LL can be considered as having expired. T1 and T2 are included to give servers explicit control over when a client recontacts the server about a specific IA_LL.
In a message sent by a client to a server, the T1 and T2 fields SHOULD be set to 0. The server MUST ignore any values in these fields in messages received from a client.
In a message sent by a server to a client, the client MUST use the values in the T1 and T2 fields for the T1 and T2 times, unless those values in those fields are 0. The values in the T1 and T2 fields are the number of seconds until T1 and T2.
As per Section 7.7 of [RFC8415]), the value 0xffffffff is taken to mean "infinity" and should be used carefully.
The server selects the T1 and T2 times to allow the client to extend the lifetimes of any address block in the IA_LL before the lifetimes expire, even if the server is unavailable for some short period of time. Recommended values for T1 and T2 are .5 and .8 times the shortest valid lifetime of the address blocks in the IA that the server is willing to extend, respectively. If the "shortest" valid lifetime is 0xffffffff ("infinity"), the recommended T1 and T2 values are also 0xffffffff. If the time at which the addresses in an IA_LL are to be renewed is to be left to the discretion of the client, the server sets T1 and T2 to 0. The client MUST follow the rules defined in Section 14.2 in [RFC8415].
If a client receives an IA_LL with T1 greater than T2, and both T1 and T2 are greater than 0, the client discards the IA_LL option and processes the remainder of the message as though the server had not included the invalid IA_LL option.
The Link-Layer Addresses option is used to specify an address block associated with a IA_LL. The option must be encapsulated in the IA_LL-options field of an IA_LL option. The LLaddr-options fields encapsulates those options that are specific to this address block.
The format of the Link-Layer Addresses option is:
0 1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| OPTION_LLADDR | option-len |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| link-layer-type | link-layer-len |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
. .
. link-layer-address .
. .
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| extra-addresses |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| valid-lifetime |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
. .
. LLaddr-options .
. .
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
Figure 3: LLADDR Option Format
In a message sent by a client to a server, the valid lifetime field SHOULD be set to 0. The server MUST ignore any received value.
In a message sent by a server to a client, the client MUST use the value in the valid lifetime field for the valid lifetime for the address block. The value in the valid lifetime field is the number of seconds remaining in the lifetime.
As per Section 7.7 of [RFC8415], the valid lifetime of 0xffffffff is taken to mean "infinity" and should be used carefully.
More than one LLADDR option can appear in an IA_LL option.
A server selects link-layer addresses to be assigned to an IA_LL according to the assignment policies determined by the server administrator.
Link-layer addresses are typically specific to a link and the server SHOULD follow the steps in Section 13.1 of [RFC8415] to determine the client's link.
For Ethernet / IEEE 802 MAC addresses, a server MAY use additional options supplied by a relay agent or client to select the quadrant (see Appendix A) from which addresses are to be assigned. This MAY include new options, such as those specified in [I-D.ietf-dhc-slap-quadrant].
IANA is requested to assign the OPTION_IA_LL (tbd1) option code from the DHCPv6 "Option Codes" registry maintained at http://www.iana.org/assignments/dhcpv6-parameters and use the following data when adding the option to the registry:
Value: tbd1
Description: OPTION_IA_LL
Client ORO: No
Singleton Option: No
Reference: this document
IANA is requested to assign the OPTION_LLADDR (tbd2) option code from the DHCPv6 "Option Codes" registry maintained at http://www.iana.org/assignments/dhcpv6-parameters and use the following data when adding the option to the registry:
Value: tbd2
Description: OPTION_LLADDR
Client ORO: No
Singleton Option: No
Reference: this document
See [RFC8415] and [RFC7227] for the DHCP security considerations. See [RFC8200] for the IPv6 security considerations.
There is a possibility of the same link-layer address being used by more than one device if not all parties on a link use this mechanism to obtain an address from the space assigned to the DHCP server. Note that this issue would exist on these networks even if DHCP were not used to obtain the address. See [IEEE-802.11-02-109r0] for techniques that can be used on 802.11 networks to probe for and avoid collisions.
Server implementations SHOULD consider configuration options to limit the maximum number of addresses to allocate (both in a single request and in total) to a client. However, note that this does not prevent a bad client actor from pretending to be many different clients and consuming all available addresses.
See [RFC8415] for the DHCP privacy considerations.
For a client requesting a link-layer address directly from a server, as the address assigned to a client will likely be used by the client to communicate on the link, the address will be exposed to those able to listen in on this communication. For those peers on the link that are able to listen in on the DHCP exchange, they would also be able to correlate the client's identity (based on the DUID used) with the assigned address. Additional mechanisms, such as the ones described in [RFC7844] can also be used.
Thanks to the DHC Working Group participants that reviewed this document, provided comments, and support. With special thanks to Ian Farrer for his thorough reviews and shepherding of this document through the IETF process. Thanks also to area reviewers Samita Chakrabarti, Roni Even and Tianran Zhou, and IESG members Warren Kumari, Barry Leiba, and Eric Vyncke for their suggestions.
| [RFC2119] | Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, DOI 10.17487/RFC2119, March 1997. |
| [RFC4861] | Narten, T., Nordmark, E., Simpson, W. and H. Soliman, "Neighbor Discovery for IP version 6 (IPv6)", RFC 4861, DOI 10.17487/RFC4861, September 2007. |
| [RFC8174] | Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC 2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174, May 2017. |
| [RFC8415] | Mrugalski, T., Siodelski, M., Volz, B., Yourtchenko, A., Richardson, M., Jiang, S., Lemon, T. and T. Winters, "Dynamic Host Configuration Protocol for IPv6 (DHCPv6)", RFC 8415, DOI 10.17487/RFC8415, November 2018. |
This appendix provides a brief summary of IEEE 802c from [IEEEStd802c-2017].
The original IEEE 802 specifications assigned half of the 48-bit MAC address space to local use -- these addresses have the U/L bit set to 1 and are locally administered with no imposed structure.
In 2017, the IEEE issued the 802c specification which defines a new "optional Structured Local Address Plan (SLAP) that specifies different assignment approaches in four specified regions of the local MAC address space." Under this plan, there are 4 SLAP quadrants that use different assignment policies.
The first octet of the MAC address Z and Y bits define the quadrant for locally assigned addresses (X-bit is 1). In IEEE representation, these bits are as follows:
LSB MSB
M X Y Z - - - -
| | | |
| | | +------------ SLAP Z-bit
| | +--------------- SLAP Y-bit
| +------------------ X-bit (U/L) = 1 for locally assigned
+--------------------- M-bit (I/G) (unicast/group)
Figure 4: SLAP Bits
The SLAP quadrants are:
| Quadrant | Y-bit | Z-bit | Local Identifier Type | Local Identifier |
|---|---|---|---|---|
| 01 | 0 | 1 | Extended Local | ELI |
| 11 | 1 | 1 | Standard Assigned | SAI |
| 00 | 0 | 0 | Administratively Assigned | AAI |
| 10 | 1 | 0 | Reserved | Reserved |
Extended Local Identifier (ELI) derived MAC addresses are based on an assigned Company ID (CID), which is 24-bits (including the M, X, Y, and Z bits) for 48-bit MAC addresses. This leaves 24-bits for the locally assigned address for each CID for unicast (M-bit = 0) and also for multicast (M-bit = 1). The CID is assigned by the IEEE RA.
Standard Assigned Identifier (SAI) derived MAC addresses are assigned by a protocol specified in an IEEE 802 standard. For 48-bit MAC addresses, 44 bits are available. Multiple protocols for assigning SAIs may be specified in IEEE standards. Coexistence of multiple protocols may be supported by limiting the subspace available for assignment by each protocol.
Administratively Assigned Identifier (AAI) derived MAC addresses are assigned locally. Administrators manage the space as needed. Note that multicast IPv6 packets ([RFC2464]) use a destination address starting in 33-33 and this falls within this space and therefore should not be used to avoid conflict with IPv6 multicast addresses. For 48-bit MAC addresses, 44 bits are available.
The last quadrant is reserved for future use. While this quadrant may also be used for AAI space, administrators should be aware that future specifications may define alternate uses that could be incompatible.