A Reference Model for Autonomic NetworkingCiscombehring@cisco.comDepartment of Computer ScienceUniversity of AucklandPB 92019Auckland1142New Zealandbrian.e.carpenter@gmail.comCiscoeckert@cisco.comAlcatel LucentRoute de VillejustNozay91620Francelaurent.ciavaglia@alcatel-lucent.comHuawei Technologies Q14, Huawei CampusNo.156 Beiqing RoadHai-Dian District, Beijing100095P.R. Chinaleo.liubing@huawei.com
Operations and Management
ANIMA
This document describes a reference model for Autonomic Networking. The goal is to define how the various elements in an autonomic context work together, to describe their interfaces and relations. While the document is written as generally as possible, the initial solutions are limited to the chartered scope of the WG.
The document "Autonomic Networking - Definitions and Design Goals" explains the fundamental concepts behind Autonomic Networking, and defines the relevant terms in this space. In section 5 it describes a high level reference model. This document defines this reference model with more detail, to allow for functional and protocol specifications to be developed in an architecturally consistent, non-overlapping manner. While the document is written as generally as possible, the initial solutions are limited to the chartered scope of the WG.
As discussed in , the goal of this work is not to focus exclusively on fully autonomic nodes or networks. In reality, most networks will run with some autonomic functions, while the rest of the network is traditionally managed. This reference model allows for this hybrid approach.
This is a living document and will evolve with the technical solutions developed in the ANIMA WG. This section describes the various elements in a network with autonomic functions, and how these entities work together, on a high level. Subsequent sections explain the detailed inside view for each of the autonomic network elements, as well as the network functions (or interfaces) between those elements. Autonomic entities include:
Network elements: A network element can be a fully or partially autonomic node. It runs autonomic functions, and interacts with other autonomic nodes. Registrar: Security is a fundamental requirement in an autonomic network. For nodes and services to securely interact without the need to provision shared secrets, a trust infrastructure must be in place. The registrar is the trust anchor in an autonomic domain. MASA (Manufacturer Authorized Signing Authority): The MASA is a service for devices of a particular vendor. It can validate the identity of devices that are to be used in an autonomic domain, assert which device is owned by which domain, etc. This section describes all the elements in an autonomic network, their function, internal organisation and architecture. This section describes an autonomic network element and its internal architecture. The reference model explained in shows the sources of information that an autonomic service agent can leverage: Self-knowledge, network knowledge (through discovery), Intent, and feedback loops. Fundamentally, there are two levels inside an autonomic node: the level of Autonomic Service Agents, and the level of the Autonomic Networking Infrastructure, with the former using the services of the latter. illustrates this concept.
The Autonomic Networking Infrastructure (lower part of ) contains node specific data structures, for example trust information about itself and its peers, as well as a generic set of functions, independent of a particular usage. This infrastructure should be generic, and support a variety of Autonomic Service Agents (upper part of ). The Autonomic Control Plane is the summary of all interactions of the Autonomic Networking Infrastructure with other nodes and services.The use cases of "Autonomics" such as self-management, self-optimisation, etc, are implemented as Autonomic Service Agents. They use the services and data structures of the underlying autonomic networking infrastructure. The underlying Autonomic Networking Infrastructure should itself be self-managing. This section describes the registrar function in an autonomic network. It explains the tasks of a registrar element, and how registrars are placed in a network, redundancy between several, etc. [tbc]tbcInside a domain, each autonomic device needs a domain specific identifier. [tbc]Autonomic Service Agents (ASAs) need addressing to communicate with each other. This section describes the addressing approach of the Autonomic Networking Infrastructure, used by ASAs. It does NOT describe addressing approaches for the data plane of the network, which may be configured and managed in the traditional way. ASAs may provide a service to negotiate address space, or addressing mechanisms for the date plane. One use case for such an autonomic function is described in . The addressing the ASAs use is in scope for this section, the address space they negotiate for the data plane is not. It is generally desirable to make the addressing scheme of the Autonomic Networking Infrastructure as self-managing (autonomic) as possible. This section is currently under discussion. We currently believe that we should address the following questions:
How addressing inside the Autonomic Control Plane (ACP) is assigned and managed autonomically.Whether, if there is no separated ACP, Autonomic Service Agents (ASAs) shall have their own address space, or whether they should use the address space configured by the administrator.How addressing is handled in the presence of non-autonomic nodes.Prefix assignment to interfaces.Whether links and link interfaces should get routable address space, or whether link local addressing is sufficient.How the address space used in the Autonomic Networking Infrastructure is assigned and managed.It is not clear at this point whether a specific address scheme should be included in this document, or whether this document should only define the requirements. This is for further discussion.The document describes one way to autonomically assign loopback addresses to autonomic nodes, in an autonomic, self-managed way. Other ideas and suggestions are strongly encouraged. Autonomic nodes have direct interactions between themselves, which must be secured. Since an autonomic network does not rely on configuration, it is not an option to configure for example pre-shared keys. A trust infrastructure such as a PKI infrastructure must be in place. This section describes the principles of this trust infrastructure. A completely autonomic way to automatically and securely deploy such a trust infrastructure is to set up a trust anchor for the domain, and then use an approach as in the document "Bootstrapping Key Infrastructures" .This section describes how autonomic nodes interact. The totality of autonomic interactions forms the "Autonomic Control Plane". This control plane can be either implemented in the global routing table of a node, such as IGPs in today's networks; or it can be provided as an overlay network, as described in . This section describes the function of the autonomic control plane, independent of its implementation. Traditionally, most of the information a node requires is provided through configuration or northbound interfaces. An autonomic function should only minimally rely on such northbound interfaces, therefore it needs to discover resources in the network. This section describes various discovery functions in an autonomic network.Discovering nodes and their properties: A core function to establish an autonomic domain is the discovery of autonomic nodes, primarily adjacent nodes. This may either leverage existing neighbour discovery mechanisms, or new mechanisms.Discovering services: Network services such as AAA should also be discovered and not configured. Service discovery is required for such tasks. An autonomic network can either leverage existing service discovery functions, or build a new approach.Thus the discovery mechanism could either be fully integrated with negotiation and synchronization (next section) or could use an independent discovery mechanism such as DNS Service Discovery or Service Location Protocol. This choice is made independently for each Autonomic Service Agent.Autonomic nodes must negotiate and/or synchronize network parameters of any kind and complexity. This requires some form of signaling between autonomic nodes. The document "A Generic Discovery and Negotiation Protocol for Autonomic Networking" describes requirements for negotiation and synchronization in an autonomic network. It also defines a protocol, GDNP, for this purpose, including an integrated but optional discovery protocol. For network-wide synchronization using a flooding algorithm, DNCP may be used. GDNP is designed to co-exist with DNCP. The distribution of Intent is also a function of the Autonomic Control Plane. Various methods can be used to distribute Intent across an autonomic domain. An autonomic network offers through the autonomic control plane the possibility to aggregate information inside the network, before sending it to the admin of the network. While this can be seen or implemented as a specific form of negotiation, the use case is different and therefore mentioned here explicitly. Feedback loops are required in an autonomic network to allow administrator intervention, while maintaining a default behaviour. Through a feedback loop an administrator can be prompted with a default action, and has the possibility to acknowledge or override the proposed default action.All autonomic nodes in a domain must be able to communicate with each other, and with autonomic nodes outside their own domain. Therefore, an Autonomic Control Plane relies on a routing function. Different autonomic functions may conflict in setting certain parameters. For example, an energy efficiency function may want to shut down a redundant link, while a load balancing function would not want that to happen. The administrator must be able to understand and resolve such interactions, to steer autonomic network performance to a given (intended) operational point.Several interaction types may exist among autonomic functions, for example:
Cooperation: An autonomic function can improve the behavior or performance of another autonomic function, such as a traffic forecasting function used by a traffic allocation function. Dependency: An autonomic function cannot work without another one being present or accessible in the autonomic network.Conflict: A metric value conflict is a conflict where one metric is influenced by parameters of different autonomic functions. A parameter value conflict is a conflict where one parameter is modified by different autonomic functions. Solving the coordination problem beyond one-by-one cases can rapidly become intractable for large networks. Specifying a common functional block on coordination is a first step to address the problem in a systemic way. The coordination life-cycle consists in three states:
At build-time, a "static interaction map" can be constructed on the relationship of functions and attributes. This map can be used to (pre-)define policies and priorities on identified conflicts.At deploy-time, autonomic functions are not yet active/acting on the network. A "dynamic interaction map" is created for each instance of each autonomic functions and on a per resource basis, including the actions performed and their relationships. This map provides the basis to identify conflicts that will happen at run-time, categorize them and plan for the appropriate coordination strategies/mechanisms.At run-time, when conflicts happen, arbitration is driven by the coordination strategies. Also new dependencies can be observed and inferred, resulting in an update of the dynamic interaction map and adaptation of the coordination strategies and mechanisms.Multiple coordination strategies and mechanisms exists and can be devised. The set ranges from basic approaches such as random process or token-based process, to approaches based on time separation and hierarchical optimization, to more complex approaches such as multi-objective optimization, and other control theory approaches and algorithms family.A common coordination functional block is a desirable component of the ANIMA reference model. It provides a means to ensure network properties and predictable performance or behavior such as stability, and convergence, in the presence of several interacting autonomic functions.A common coordination function requires:
A common description of autonomic functions, their attributes and life-cycle.A common representation of information and knowledge (e.g., interaction maps).A common “control/command” interface between the coordination "agent" and the autonomic functions. Guidelines, recommendations or BCPs can also be provided for aspects pertaining to the coordination strategies and mechanisms.This section explains how autonomic functions can co-exist with non-autonomic functions, and how a potential overlap is managed. (tbc)This is a preliminary outline of a threat analysis, to be expanded and made more specific as the various Autonomic Networking specifications evolve.Since AN will hand over responsibility for network configuration from humans or centrally established management systems to fully distributed devices, the threat environment is also fully distributed. On the one hand, that means there is no single point of failure to act as an attractive target for bad actors. On the other hand, it means that potentially a single misbehaving autonomic device could launch a widespread attack, by misusing the distributed AN mechanisms. For example, a resource exhaustion attack could be launched by a single device requesting large amounts of that resource from all its peers, on behalf of a non-existent traffic load.
Alternatively it could simply send false information to its peers, for example by announcing resource exhaustion when this was not the case.
If security properties are managed autonomically, a misbehaving device could attempt a distributed attack by requesting all its peers to reduce security protections in some way. In general, since autonomic devices run without supervision, almost any kind of undesirable management action could in theory be attempted by a misbehaving device. If it is possible for an unauthorised device to act as an autonomic device, or for a malicious third party to inject messages appearing to come from an autonomic device, all these same risks would apply. If AN messages can be observed by a third party, they might reveal valuable information about network configuration, security precautions in use, individual users, and their traffic patterns. If encrypted, AN messages might still reveal some information via traffic analysis, but this would be quite limited (for example, this would be highly unlikely to reveal any specific information about user traffic). AN messages are liable to be exposed to third parties on any unprotected Layer 2 link, and to insider attacks even on protected Layer 2 links. This document requests no action by IANA. Many people have provided feedback and input to this document: Sheng Jiang, Roberta Maglione, Jonathan Hansford.00: Initial version.