| SACM Working Group | A. Montville |
| Internet-Draft | B. Munyan |
| Intended status: Standards Track | CIS |
| Expires: August 26, 2019 | February 22, 2019 |
Security Automation and Continuous Monitoring (SACM) Architecture
draft-ietf-sacm-arch-01
This memo defines a Security Automation and Continuous Monitoring (SACM) architecture. This work is built upon [I-D.ietf-mile-xmpp-grid], and is predicated upon information gleaned from SACM Use Cases and Requirements ([RFC7632] and [RFC8248] respectively), and terminology as found in [I-D.ietf-sacm-terminology].
WORKING GROUP: The source for this draft is maintained in GitHub. Suggested changes should be submitted as pull requests at https://github.com/sacmwg/ietf-mandm-sacm-arch/. Instructions are on that page as well.
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 August 26, 2019.
Copyright (c) 2019 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.
The purpose of this draft is to define an architectural solution for a SACM Domain. This draft also defines an implementation of the architecutre, built upon [I-D.ietf-mile-xmpp-grid] and [I-D.ietf-sacm-ecp]. These approaches complement each other to more completely meet the spirit of [RFC7632] and requirements found in [RFC8248].
This solution gains the most advantage by supporting a variety of collection mechanisms. In this sense, the solution ideally intends to enable a cooperative ecosystem of tools from disparate sources with minimal operator configuration. The solution described in this document seeks to accommodate these recognitions by first defining a generic abstract architecture, then making that solution somewhat more concrete.
Keep in mind that, at this point, the draft is tracking ongoing work being performed primarily around and during IETF hackathons. The list of hackathon efforts follows:
[NOTE: This section will eventually be removed.]
The following is a list of open questions we still have about the path forward with this exploration:
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 RFC 2119, BCP 14 [RFC2119].
This draft defers to [I-D.ietf-sacm-terminology] for terms and definitions.
The generic approach proposed herein recognizes the need to obtain information from existing state collection mechanisms, and makes every attempt to respect [RFC7632] and [RFC8248]. At the foundation of any architecture are entities, or components, that need to communicate. They communicate by sharing information, where, in a given flow one or more components are consumers of information and one or more components are providers of information.
+--------------------+
| Feeds/Repositories |
| of External Data |
+--------------------+
|
|
*****************************v**************** Enterprise Boundary ************
* | *
* +--------------+ | +--------------+ *
* | Orchestrator | | | Repositories | *
* +------^-------+ | +----^---------+ *
* | | | +----------------+ *
* A | B | C | | Downstream Uses| *
* | | | | +-----------+ | *
* +------v------------------v----------v------+ | |Evaluations| | *
* | Message Transfer System <-------> +-----------+ | *
* +----------------------^--------------------+ D | +-----------+ | *
* E | | | Analytics | | *
* | | +-----------+ | *
* +-------------v---------+ | +-----------+ | *
* | Collection Subsystems | | | Reporting | | *
* +-----------------------+ | +-----------+ | *
* +----------------+ *
*******************************************************************************
Figure 1: Notional Architecture
As shown in Figure 1, the notional SACM architecture consists of some basic SACM Components using a message transfer system to communicate. While not depicted, the message transfer system is expected to maximally align with the requirements described in [RFC8248], which means that the message transfer system will support brokered (i.e. point-to-point) and proxied data exchange.
Additionally, component-specific interfaces (i.e. such as A, B, C, D, and E in Figure 1) are expected to be specified logically then bound to one or more specific implementations. This SHOULD be done for each capability related to the given SACM Component.
This document suggests a variety of players in a cooperative ecosystem - we call these players SACM Components. SACM Components may be composed of other SACM Components, and each SACM Component plays one, or more, of several roles relevant to the ecosystem. Generally each role is either a consumer of information or a provider of information. The "Components, Capabilities, Interfaces, and Workflows" section provides more details about SACM Components that play these types of roles.
Figure 2 depicts a slightly more detailed view of the architecture (within the enterprise boundary) - one that fosters the development of a pluggable ecosystem of cooperative tools. Existing collection mechanisms can be brought into this architecture by specifying the interface of the collector and creating the XMPP-Grid Connector binding for that interface.
Additionally, while not directly depicted in Figure 2, this architecture does allow point-to-point interfaces. In fact, [I-D.ietf-mile-xmpp-grid] provides brokering capabilities to facilitate such point-to-point data transfers). Additionally, each of the SACM Components depicted in Figure 2 may be a provider, a consumer, or both, depending on the workflow in context.
+--------------+ +--------------+
| Orchestrator | | Repositories |
+------^-------+ +------^-------+
| |
| |
+-------v--------------------------v--------+ +-----------------+
| XMPP-Grid+ <-----> Downstream Uses |
+------------------------^------------------+ +-----------------+
|
|
+-------v------+
| XMPP-Grid |
| Connector(s) |
+------^-------+
|
+------v-------+
| Collector(s) |
+--------------+
Figure 2: XMPP-based Architecture
[I-D.ietf-mile-xmpp-grid] details a number of XMPP extensions (XEPs) that MUST be utilized to meet the needs of [RFC7632] and [RFC8248]:
At this point, [I-D.ietf-mile-xmpp-grid] specifies fewer features than SACM requires, and there are other XMPP extensions (XEPs) we need to consider to meet the needs of [RFC7632] and [RFC8248]. In Figure 2 we therefore use "XMPP-Grid+" to indicate something more than [I-D.ietf-mile-xmpp-grid] alone, even though we are not yet fully confident in the exact set of XMPP-related extensions we will require. The authors propose work to extend (or modify) [I-D.ietf-mile-xmpp-grid] to include additional XEPs - possibly the following:
Figure 3 depicts a further detailed view of the architecture including the Endpoint Posture Collection Protocol as the collection subsystem, illustrating the idea of a pluggable ecosystem of cooperative tools.
+--------------------+
| Feeds/Repositories |
| of External Data |
+--------------------+
|
********************v************************* Enterprise Boundary ************
* | *
* +--------------+ | +-------------------+ +-------------+ *
* | Orchestrator | | | Posture Attr Repo | | Policy Repo | *
* +------^-------+ | +---------^---------+ +---^---------+ *
* | | | | +----------------+ *
* | | | | | Downstream Uses| *
* | | | | | +-----------+ | *
* +------v---------v-----------v---------------v--+ | |Evaluations| | *
* | XMPP-Grid <-------> +-----------+ | *
* +----------------^-------------------^----------+ | +-----------+ | *
* | | | | Analytics | | *
* | | | +-----------+ | *
* | +-----v--------+ | +-----------+ | *
* | | Results Repo | | | Reporting | | *
* | +--------------+ | +-----------+ | *
* | +----------------+ *
* +---------v-----------+ *
* | XMPP-Grid Connector | *
* +---------^-----------+ *
* | *
* +-----------------v-------------------------------------------------------+ *
* | | *
* | +--Posture Collection Manager------------------------------------------+| *
* | |+-----------------------+ +----------------+ +----------------------+ || *
* | || Communications Server | | Posture Server | | Posture Validator(s) | || *
* | |+----------^------------+ +----------------+ +----------------------+ || *
* | +-----------|----------------------------------------------------------+| *
* | | | *
* | +-----------|-------------------------Endpoint or Endpoint Proxy-------+| *
* | |+----------v------------+ +----------------+ +----------------------+ || *
* | || Communications Client | | Posture Client | | Posture Collector(s) | || *
* | |+-----------------------+ +----------------+ +----------------------+ || *
* | +----------------------------------------------------------------------+| *
* +-----------------Endpoint Posture Collection Profile---------------------+ *
* *
*******************************************************************************
Figure 3: XMPP-based Architecture including EPCP
The SACM Architecture consists of a variety of SACM Components, and named components are intended to embody one or more specific capabilities. Interacting with these capabilities will require at least two levels of interface specification. The first is a logical interface specification, and the second is at least one binding to a specific transfer mechanism. An example transfer mechanism is XMPP-Grid+.
The following subsections describe some of the components, capabilities, and interfaces we may expect to see participating in a SACM Domain.
The following is a list of suggested SACM Component classes and specializations.
Repositories will have a need for fairly standard CRUD operations and query by attribute operations. Collector interfaces may enable ad hoc assessment (on-demand processing), state item watch actions (i.e. watch a particular item for particular change), persisting other behaviors (i.e. setting some mandatory reporting period). Evaluators may have their own set of interfaces, and an Assessor would represent both Collector and Evaluation interfaces, and may have additional concerns added to an Assessor Interface.
Not to be overlooked, whatever solution at which we arrive, per [RFC8248], MUST support capability negotiation. While not explicitly treated here, each interface will understand specific serializations, and other component needs to express those serializations to other components.
A capability language is fully explored in mandl-sacm-tool-capability-language (to be submitted).
Interfaces should be derived directly from identified workflows, several of which are described in this document.
The workflows described in this document should be considered as candidate workflows - informational for the purpose of discovering the necessary components and specifying their interfaces.
Information Technology asset management is easier said than done. The [CISCONTROLS] have two controls dealing with IT asset management. Control 1, Inventory and Control of Hardware Assets, states, "Actively manage (inventory, track, and correct) all hardware devices on the network so that only authorized devices are given access, and unauthorized and unmanaged devices are found and prevented from gaining access." Control 2, Inventory and Control of Software Assets, states, "Actively manage (inventory, track, and correct) all software on the network so that only authorized software is installed and can execute, and that unauthorized and unmanaged software is found and prevented from installation or execution."
In spirit, this covers all of the processing entities on your network (as opposed to things like network cables, dongles, adapters, etc.), whether physical or virtual.
An IT asset management capability needs to be able to:
Vulnerability management is a relatively established process. According to the [CISCONTROLS], continuous vulnerability management the act of continuously acquiring, assessing, and taking subsequent action on new information in order to identify and remediate vulnerabilities, therefore minimizing the window of opportunity for attackers.
A number of assumptions must be stated to clarify the scope of a vulnerability assessment workflow:
When new vulnerability description information is received by the enterprise, affected endpoints are identified and assessed. The vulnerability is said to apply to an endpoint if the endpoint satisfies the conditions expressed in the vulnerability detection data.
A vulnerability assessment (i.e. vulnerability detection) is performed in two steps:
Vulnerability detection relies on the examination of different endpoint information depending on the nature of a specific vulnerability. Common endpoint information used to detect a vulnerability includes:
In many cases, the endpoint information needed to determine an endpoint's vulnerability status will have been previously collected by the endpoint management capabilities and available in a Repository. However, in other cases, the necessary endpoint information will not be readily available in a Repository and a Collection Task will be triggered to collect it from the target endpoint. Of course, some implementations of endpoint management capabilities may prefer to enable operators to perform this collection under certain circumstances, even when sufficient information can be provided by the endpoint management capabilities (e.g. there may be freshness requirements for information).
The collection of additional endpoint information for the purpose of vulnerability assessment does not necessarily need to be a pull by the vulnerability assessment capabilities. Over time, some new pieces of information that are needed during common types of assessments might be identified. Endpoint management capabilities can be reconfigured to have this information delivered automatically. This avoids the need to trigger additional Collection Tasks to gather this information during assessments, streamlining the assessment process. Likewise, it might be observed that certain information delivered by endpoint management capabilities is rarely used. In this case, it might be useful to re-configure the endpoint management capabilities to no longer collect this information to reduce network and processing overhead. Instead, a new Collection Task can be triggered to gather this data on the rare occasions when it is needed.
Configuration management involves configuration assessment, which requires state assessment (TODO: Tie to SACM use cases). The [CISCONTROLS] specify two high-level controls concerning configuration management (Control 5 for non-network devices and Control 11 for network devices). As an aside, these controls are listed separately because many enterprises have different organizations for managing network infrastructure and workload endpoints. Merging the two controls results in a requirement to: "Establish, implement, and actively manage (track, report on, correct) the security configuration of (endpoints) using a rigorous configuration management and change control process in order to prevent attackers from exploiting vulnerable services and settings."
Typically, an enterprise will use configuration guidance from a reputable source, and from time to time they may tailor the guidance from that source prior to adopting it as part of their enterprise standard. The enterprise standard is then provided to the appropriate configuration assessment tools and they assess endpoints and/or appropriate endpoint information. A preferred flow follows:
The SACM architecture needs to support varying deployment models to accommodate the current state of the industry, but should strongly encourage event-driven approaches to monitoring configuration.
TODO
TODO
IANA tables can probably be used to make life a little easier. We would like a place to enumerate:
| [I-D.ietf-mile-xmpp-grid] | Cam-Winget, N., Appala, S., Pope, S. and P. Saint-Andre, "Using XMPP for Security Information Exchange", Internet-Draft draft-ietf-mile-xmpp-grid-09, December 2018. |
| [I-D.ietf-sacm-ecp] | Haynes, D., Fitzgerald-McKay, J. and L. Lorenzin, "Endpoint Posture Collection Profile", Internet-Draft draft-ietf-sacm-ecp-04, February 2019. |
| [RFC2119] | Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, DOI 10.17487/RFC2119, March 1997. |
| [RFC8412] | Schmidt, C., Haynes, D., Coffin, C., Waltermire, D. and J. Fitzgerald-McKay, "Software Inventory Message and Attributes (SWIMA) for PA-TNC", RFC 8412, DOI 10.17487/RFC8412, July 2018. |
This section provides a mapping of XMPP and XMPP Extensions to the relevant requirements from [RFC8248]. In the table below, the ID and Name columns provide the ID and Name of the requirement directly out of [RFC8248]. The Supported By column may contain one of several values:
If there is no entry in the Supported By column, then there is a gap that must be filled.
| ID | Name | Supported By |
|---|---|---|
| G-001 | Solution Extensibility | XMPP-Core |
| G-002 | Interoperability | XMPP |
| G-003 | Scalability | XMPP |
| G-004 | Versatility | XMPP-Core |
| G-005 | Information Extensibility | XMPP-Core |
| G-006 | Data Protection | Operational |
| G-007 | Data Partitioning | Operational |
| G-008 | Versioning and Backward Compatibility | XEP-0115/0030 |
| G-009 | Information Discovery | XEP-0030 |
| G-010 | Target Endpoint Discovery | XMPP-Core |
| G-011 | Push and Pull Access | XEP-0060/0312 |
| G-012 | SACM Component Interface | N/A |
| G-013 | Endpoint Location and Network Topology | |
| G-014 | Target Endpoint Identity | XMPP-Core |
| G-015 | Data Access Control | |
| ARCH-001 | Component Functions | XMPP |
| ARCH-002 | Scalability | XMPP-Core |
| ARCH-003 | Flexibility | XMPP-Core |
| ARCH-004 | Separation of Data and Management Functions | |
| ARCH-005 | Topology Flexibility | XMPP-Core |
| ARCH-006 | Capability Negotiation | XEP-0115/0030 |
| ARCH-007 | Role-Based Authorization | XMPP-Core |
| ARCH-008 | Context-Based Authorization | |
| ARCH-009 | Time Synchronization | Operational |
| IM-001 | Extensible Attribute Vocabulary | N/A |
| IM-002 | Posture Data Publication | N/A |
| IM-003 | Data Model Negotiation | N/A |
| IM-004 | Data Model Identification | N/A |
| IM-005 | Data Lifetime Management | N/A |
| IM-006 | Singularity and Modularity | N/A |
| DM-001 | Element Association | N/A |
| DM-002 | Data Model Structure | N/A |
| DM-003 | Search Flexibility | N/A |
| DM-004 | Full vs. Partial Updates | N/A |
| DM-005 | Loose Coupling | N/A |
| DM-006 | Data Cardinality | N/A |
| DM-007 | Data Model Negotiation | N/A |
| DM-008 | Data Origin | N/A |
| DM-009 | Origination Time | N/A |
| DM-010 | Data Generation | N/A |
| DM-011 | Data Source | N/A |
| DM-012 | Data Updates | N/A |
| DM-013 | Multiple Collectors | N/A |
| DM-014 | Attribute Extensibility | N/A |
| DM-015 | Solicited vs. Unsolicited Updates | N/A |
| DM-016 | Transfer Agnostic | N/A |
| OP-001 | Time Synchronization | |
| OP-002 | Collection Abstraction | |
| OP-003 | Collection Composition | |
| OP-004 | Attribute-Based Query | |
| OP-005 | Information-Based Query with Filtering | |
| OP-006 | Operation Scalability | |
| OP-007 | Data Abstraction | |
| OP-008 | Provider Restriction | |
| T-001 | Multiple Transfer Protocol Support | Architecture |
| T-002 | Data Integrity | Operational |
| T-003 | Data Confidentiality | Operational |
| T-004 | Transfer Protection | |
| T-005 | Transfer Reliability | |
| T-006 | Transfer-Layer Requirements | |
| T-007 | Transfer Protocol Adoption | Architecture |
Consider a policy server conforming to [RFC8322]. [RFC8322] describes a RESTful way based on the ATOM Publishing Protocol ([RFC5023]) to find specific data collections. While this represents a specific binding (i.e. RESTful API based on [RFC5023]), there is a more abstract way to look at ROLIE.
ROLIE provides notional workspaces and collections, and provides the concept of information categories and links. Strictly speaking, these are logical concepts independent of the RESTful binding ROLIE specifies. In other words, ROLIE binds a logical interface (i.e. GET workspace, GET collection, SET entry, and so on) to a specific mechanism (namely an ATOM Publication Protocol extension).
It is not inconceivable to believe there could be a different interface mechanism, or a connector, providing these same operations using XMPP-Grid as the transfer mechanism.
Even if a [RFC8322] server were external to an organization, there would be a need for a policy source inside the organization as well, and it may be preferred for such a policy source to be connected directly to the ecosystem's communication infrastructure.
The SACM working group has accepted work on the Endpoint Posture Collection Profile [I-D.ietf-sacm-ecp], which describes a collection architecture and may be viewed as a collector coupled with a collection-specific repository.
Posture Manager Endpoint
Orchestrator +---------------+ +---------------+
+--------+ | | | |
| | | +-----------+ | | +-----------+ |
| |<---->| | Posture | | | | Posture | |
| | pub/ | | Validator | | | | Collector | |
| | sub | +-----------+ | | +-----------+ |
+--------+ | | | | | |
| | | | | |
Evaluator Repository | | | | | |
+------+ +--------+ | +-----------+ |<-------| +-----------+ |
| | | | | | Posture | | report | | Posture | |
| | | | | | Collection| | | | Collection| |
| |<-----> | |<-----| | Manager | | query | | Engine | |
| |request/| | store| +-----------+ |------->| +-----------+ |
| |respond | | | | | |
| | | | | | | |
+------+ +--------+ +---------------+ +---------------+
Figure 4: EPCP Collection Architecture
In Figure 4, any of the communications between the Posture Manager and EPCP components to its left could be performed directly or indirectly using a given message transfer mechanism. For example, the pub/sub interface between the Orchestrator and the Posture Manager could be using a proprietary method or using [I-D.ietf-mile-xmpp-grid] or some other pub/sub mechanism. Similarly, the store connection from the Posture Manager to the Repository could be performed internally to a given implementation, via a RESTful API invocation over HTTPS, or even over a pub/sub mechanism.
Our assertion is that the Evaluator, Repository, Orchestrator, and Posture Manager all have the potential to represent SACM Components with specific capability interfaces that can be logically specified, then bound to one or more specific transfer mechanisms (i.e. RESTful API, [RFC8322], [I-D.ietf-mile-xmpp-grid], and so on).
[NIST800126], also known as SCAP 1.3, provides the technical specifications for a "datastream collection". The specification describes the "datastream collection" as being "composed of SCAP data streams and SCAP source components". A "datastream" provides an encapsulation of the SCAP source components required to, for example, perform configuration assessment on a given endpoint. These source components include XCCDF checklists, OVAL Definitions, and CPE Dictionary information. A single "datastream collection" may encapsulate multiple "datastreams", and reference any number of SCAP components. Datastream collections were intended to provide an envelope enabling transfer of SCAP data more easily.
The [NIST800126] specification also defines the "SCAP result data stream" as being conformant to the Asset Reporting Format specification, defined in [NISTIR7694]. The Asset Reporting Format provides an encapsulation of the SCAP source components, Asset Information, and SCAP result components, such as system characteristics and state evaluation results.
What [NIST800126]did not do is specify the interface for finding or acquiring source datastream information, nor an interface for publishing result information. Discovering the actual resources for this information could be done via ROLIE, as described in the Policy Services section above, but other repositories of SCAP data exist as well.
[draft-birkholz-sacm-yang-content] illustrates a SACM Component incorporating a YANG Push client function and an XMPP-grid publisher function. [draft-birkholz-sacm-yang-content] further states "the output of the YANG Push client function is encapsulated in a SACM Content Element envelope, which is again encapsulated in a SACM statement envelope" which are published, essentially, via an XMPP-Grid Connector for SACM Components also part of the XMPP-Grid.
This is a specific example of an existing collection mechanism being adapted to the XMPP-Grid message transfer system.