idnits 2.17.1 draft-ietf-i2rs-architecture-08.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- No issues found here. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (January 7, 2015) is 3395 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- == Outdated reference: A later version (-11) exists of draft-ietf-i2rs-problem-statement-04 == Outdated reference: A later version (-13) exists of draft-ietf-idr-ls-distribution-07 == Outdated reference: A later version (-18) exists of draft-ietf-netconf-restconf-03 -- Obsolete informational reference (is this intentional?): RFC 6536 (Obsoleted by RFC 8341) Summary: 0 errors (**), 0 flaws (~~), 4 warnings (==), 2 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network Working Group A. Atlas 3 Internet-Draft Juniper Networks 4 Intended status: Informational J. Halpern 5 Expires: July 11, 2015 Ericsson 6 S. Hares 7 Huawei 8 D. Ward 9 Cisco Systems 10 T. Nadeau 11 Brocade 12 January 7, 2015 14 An Architecture for the Interface to the Routing System 15 draft-ietf-i2rs-architecture-08 17 Abstract 19 This document describes an architecture for a standard, programmatic 20 interface for state transfer in and out of the Internet routing 21 system. It describes the basic architecture, the components, and 22 their interfaces with particular focus on those to be standardized as 23 part of the Interface to Routing System (I2RS). 25 Status of This Memo 27 This Internet-Draft is submitted in full conformance with the 28 provisions of BCP 78 and BCP 79. 30 Internet-Drafts are working documents of the Internet Engineering 31 Task Force (IETF). Note that other groups may also distribute 32 working documents as Internet-Drafts. The list of current Internet- 33 Drafts is at http://datatracker.ietf.org/drafts/current/. 35 Internet-Drafts are draft documents valid for a maximum of six months 36 and may be updated, replaced, or obsoleted by other documents at any 37 time. It is inappropriate to use Internet-Drafts as reference 38 material or to cite them other than as "work in progress." 40 This Internet-Draft will expire on July 11, 2015. 42 Copyright Notice 44 Copyright (c) 2015 IETF Trust and the persons identified as the 45 document authors. All rights reserved. 47 This document is subject to BCP 78 and the IETF Trust's Legal 48 Provisions Relating to IETF Documents 49 (http://trustee.ietf.org/license-info) in effect on the date of 50 publication of this document. Please review these documents 51 carefully, as they describe your rights and restrictions with respect 52 to this document. Code Components extracted from this document must 53 include Simplified BSD License text as described in Section 4.e of 54 the Trust Legal Provisions and are provided without warranty as 55 described in the Simplified BSD License. 57 Table of Contents 59 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 60 1.1. Drivers for the I2RS Architecture . . . . . . . . . . . . 4 61 1.2. Architectural Overview . . . . . . . . . . . . . . . . . 5 62 2. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 9 63 3. Key Architectural Properties . . . . . . . . . . . . . . . . 10 64 3.1. Simplicity . . . . . . . . . . . . . . . . . . . . . . . 10 65 3.2. Extensibility . . . . . . . . . . . . . . . . . . . . . . 11 66 3.3. Model-Driven Programmatic Interfaces . . . . . . . . . . 11 67 4. Security Considerations . . . . . . . . . . . . . . . . . . . 12 68 4.1. Identity and Authentication . . . . . . . . . . . . . . . 13 69 4.2. Authorization . . . . . . . . . . . . . . . . . . . . . . 14 70 4.3. Client Redundancy . . . . . . . . . . . . . . . . . . . . 14 71 5. Network Applications and I2RS Client . . . . . . . . . . . . 15 72 5.1. Example Network Application: Topology Manager . . . . . . 15 73 6. I2RS Agent Role and Functionality . . . . . . . . . . . . . . 16 74 6.1. Relationship to its Routing Element . . . . . . . . . . . 16 75 6.2. I2RS State Storage . . . . . . . . . . . . . . . . . . . 16 76 6.2.1. I2RS Agent Failure . . . . . . . . . . . . . . . . . 17 77 6.2.2. Starting and Ending . . . . . . . . . . . . . . . . . 18 78 6.2.3. Reversion . . . . . . . . . . . . . . . . . . . . . . 18 79 6.3. Interactions with Local Config . . . . . . . . . . . . . 18 80 6.4. Routing Components and Associated I2RS Services . . . . . 19 81 6.4.1. Routing and Label Information Bases . . . . . . . . . 20 82 6.4.2. IGPs, BGP and Multicast Protocols . . . . . . . . . . 21 83 6.4.3. MPLS . . . . . . . . . . . . . . . . . . . . . . . . 21 84 6.4.4. Policy and QoS Mechanisms . . . . . . . . . . . . . . 22 85 6.4.5. Information Modeling, Device Variation, and 86 Information Relationships . . . . . . . . . . . . . . 22 87 6.4.5.1. Managing Variation: Object Classes/Types and 88 Inheritance . . . . . . . . . . . . . . . . . . . 22 89 6.4.5.2. Managing Variation: Optionality . . . . . . . . . 23 90 6.4.5.3. Managing Variation: Templating . . . . . . . . . 23 91 6.4.5.4. Object Relationships . . . . . . . . . . . . . . 24 92 6.4.5.4.1. Initialization . . . . . . . . . . . . . . . 24 93 6.4.5.4.2. Correlation Identification . . . . . . . . . 24 94 6.4.5.4.3. Object References . . . . . . . . . . . . . . 25 95 6.4.5.4.4. Active Reference . . . . . . . . . . . . . . 25 96 7. I2RS Client Agent Interface . . . . . . . . . . . . . . . . . 25 97 7.1. One Control and Data Exchange Protocol . . . . . . . . . 25 98 7.2. Communication Channels . . . . . . . . . . . . . . . . . 25 99 7.3. Capability Negotiation . . . . . . . . . . . . . . . . . 26 100 7.4. Scope Policy Specifications . . . . . . . . . . . . . . . 26 101 7.5. Connectivity . . . . . . . . . . . . . . . . . . . . . . 27 102 7.6. Notifications . . . . . . . . . . . . . . . . . . . . . . 27 103 7.7. Information collection . . . . . . . . . . . . . . . . . 28 104 7.8. Multi-Headed Control . . . . . . . . . . . . . . . . . . 28 105 7.9. Transactions . . . . . . . . . . . . . . . . . . . . . . 29 106 8. Operational and Manageability Considerations . . . . . . . . 29 107 9. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 30 108 10. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 30 109 11. Informative References . . . . . . . . . . . . . . . . . . . 30 110 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 31 112 1. Introduction 114 Routers that form the internet routing infrastructure maintain state 115 at various layers of detail and function. For example, a typical 116 router maintains a Routing Information Base (RIB), and implements 117 routing protocols such as OSPF, ISIS, and BGP to exchange 118 reachability information, topology information, protocol state, and 119 other information about the state of the network with other routers. 121 Routers convert all of this information into forwarding entries which 122 are then used to forward packets and flows between network elements. 123 The forwarding plane and the specified forwarding entries then 124 contain active state information that describes the expected and 125 observed operational behavior of the router and which is also needed 126 by the network applications. Network-oriented applications require 127 easy access to this information to learn the network topology, to 128 verify that programmed state is installed in the forwarding plane, to 129 measure the behavior of various flows, routes or forwarding entries, 130 as well as to understand the configured and active states of the 131 router. 133 This document sets out an architecture for a common, standards-based 134 interface to this information. This Interface to the Routing System 135 (I2RS) facilitates control and observation of the routing-related 136 state (for example, a Routing Element RIB manager's state), as well 137 as enabling network-oriented applications to be built on top of 138 today's routed networks. The I2RS is a programmatic asynchronous 139 interface for transferring state into and out of the internet routing 140 system. This I2RS architecture recognizes that the routing system 141 and a router's OS provide useful mechanisms that applications could 142 harness to accomplish application-level goals. 144 Fundamental to the I2RS are clear data models that define the 145 semantics of the information that can be written and read. The I2RS 146 provides a framework for registering and for requesting the 147 appropriate information for each particular application. The I2RS 148 provides a way for applications to customize network behavior while 149 leveraging the existing routing system as desired. 151 Although the I2RS architecture is general enough to support 152 information and data models for a variety of data, and aspects of the 153 I2RS solution may be useful in domains other than routing, I2RS and 154 this document are specifically focused on an interface for routing 155 data. 157 1.1. Drivers for the I2RS Architecture 159 There are four key drivers that shape the I2RS architecture. First 160 is the need for an interface that is programmatic, asynchronous, and 161 offers fast, interactive access for atomic operations. Second is the 162 access to structured information and state that is frequently not 163 directly configurable or modeled in existing implementations or 164 configuration protocols. Third is the ability to subscribe to 165 structured, filterable event notifications from the router. Fourth, 166 the operation of I2RS is to be data-model driven to facilitate 167 extensibility and provide standard data-models to be used by network 168 applications. 170 I2RS is described as an asynchronous programmatic interface, the key 171 properties of which are described in Section 5 of 172 [I-D.ietf-i2rs-problem-statement]. 174 The I2RS architecture facilitates obtaining information from the 175 router. The I2RS architecture provides the ability to not only read 176 specific information, but also to subscribe to targeted information 177 streams and filtered and thresholded events. 179 Such an interface also facilitates the injection of ephemeral state 180 into the routing system. A non-routing protocol or application could 181 inject state into a routing element via the state-insertion 182 functionality of the I2RS and that state could then be distributed in 183 a routing or signaling protocol and/or be used locally (e.g. to 184 program the co-located forwarding plane). I2RS will only permit 185 modification of state that would be safe, conceptually, to modify via 186 local configuration; no direct manipulation of protocol-internal 187 dynamically determined data is envisioned. 189 1.2. Architectural Overview 191 Figure 1 shows the basic architecture for I2RS between applications 192 using I2RS, their associated I2RS Clients, and I2RS Agents. 193 Applications access I2RS services through I2RS clients. A single 194 client can provide access to one or more applications. In the 195 figure, Clients A and B provide access to a single application, while 196 Client P provides access to multiple applications. 198 Applications can access I2RS services through local or remote 199 clients. In the figure, Applications A and B access I2RS services 200 through local clients, while Applications C, D and E access I2RS 201 services through a remote client. The details of how applications 202 communicate with a remote client is out of scope for I2RS. 204 An I2RS Client can access one or more I2RS agents. In the figure, 205 Clients B and P access I2RS Agents 1 and 2. Likewise, an I2RS Agent 206 can provide service to one or more clients. In the figure, I2RS 207 Agent 1 provides services to Clients A, B and P while Agent 2 208 provides services to only Clients B and P. 210 I2RS agents and clients communicate with one another using an 211 asynchronous protocol. Therefore, a single client can post multiple 212 simultaneous requests, either to a single agent or to multiple 213 agents. Furthermore, an agent can process multiple requests, either 214 from a single client or from multiple clients, simultaneously. 216 The I2RS agent provides read and write access to selected data on the 217 routing element that are organized into I2RS Services. Section 4 218 describes how access is mediated by authentication and access control 219 mechanisms. In addition to read and write access, the I2RS agent 220 allows clients to subscribe to different types of notifications about 221 events affecting different object instances. An example not related 222 to the creation, modification or deletion of an object instance is 223 when a next-hop in the RIB is resolved enough to be used or when a 224 particular route is selected by the RIB Manager for installation into 225 the forwarding plane. Please see Section 7.6 and Section 7.7 for 226 details. 228 The scope of I2RS is to define the interactions between the I2RS 229 agent and the I2RS client and the associated proper behavior of the 230 I2RS agent and I2RS client. 232 ****************** ***************** ***************** 233 * Application C * * Application D * * Application E * 234 ****************** ***************** ***************** 235 ^ ^ ^ 236 | | | 237 |--------------| | |--------------| 238 | | | 239 v v v 240 *************** 241 * Client P * 242 *************** 243 ^ ^ 244 | |-------------------------| 245 *********************** | *********************** | 246 * Application A * | * Application B * | 247 * * | * * | 248 * +----------------+ * | * +----------------+ * | 249 * | Client A | * | * | Client B | * | 250 * +----------------+ * | * +----------------+ * | 251 ******* ^ ************* | ***** ^ ****** ^ ****** | 252 | | | | | 253 | |-------------| | | |-----| 254 | | -----------------------| | | 255 | | | | | 256 ************ v * v * v ********* ***************** v * v ******** 257 * +---------------------+ * * +---------------------+ * 258 * | Agent 1 | * * | Agent 2 | * 259 * +---------------------+ * * +---------------------+ * 260 * ^ ^ ^ ^ * * ^ ^ ^ ^ * 261 * | | | | * * | | | | * 262 * v | | v * * v | | v * 263 * +---------+ | | +--------+ * * +---------+ | | +--------+ * 264 * | Routing | | | | Local | * * | Routing | | | | Local | * 265 * | and | | | | Config | * * | and | | | | Config | * 266 * |Signaling| | | +--------+ * * |Signaling| | | +--------+ * 267 * +---------+ | | ^ * * +---------+ | | ^ * 268 * ^ | | | * * ^ | | | * 269 * | |----| | | * * | |----| | | * 270 * v | v v * * v | v v * 271 * +----------+ +------------+ * * +----------+ +------------+ * 272 * | Dynamic | | Static | * * | Dynamic | | Static | * 273 * | System | | System | * * | System | | System | * 274 * | State | | State | * * | State | | State | * 275 * +----------+ +------------+ * * +----------+ +------------+ * 276 * * * * 277 * Routing Element 1 * * Routing Element 2 * 278 ******************************** ******************************** 280 Figure 1: Architecture of I2RS clients and agents 282 Routing Element: A Routing Element implements some subset of the 283 routing system. It does not need to have a forwarding plane 284 associated with it. Examples of Routing Elements can include: 286 * A router with a forwarding plane and RIB Manager that runs 287 ISIS, OSPF, BGP, PIM, etc., 289 * A BGP speaker acting as a Route Reflector, 291 * An LSR that implements RSVP-TE, OSPF-TE, and PCEP and has a 292 forwarding plane and associated RIB Manager, 294 * A server that runs ISIS, OSPF, BGP and uses ForCES to control a 295 remote forwarding plane, 297 A Routing Element may be locally managed, whether via CLI, SNMP, 298 or NETCONF. 300 Routing and Signaling: This block represents that portion of the 301 Routing Element that implements part of the internet routing 302 system. It includes not merely standardized protocols (i.e. IS- 303 IS, OSPF, BGP, PIM, RSVP-TE, LDP, etc.), but also the RIB Manager 304 layer. 306 Local Config: A Routing Element will provide the ability to 307 configure and manage it. The Local Config may be provided via a 308 combination of CLI, NETCONF, SNMP, etc. The black box behavior 309 for interactions between the state that I2RS installs into the 310 routing element and the Local Config must be defined. 312 Dynamic System State: An I2RS agent needs access to state on a 313 routing element beyond what is contained in the routing subsystem. 314 Such state may include various counters, statistics, flow data, 315 and local events. This is the subset of operational state that is 316 needed by network applications based on I2RS that is not contained 317 in the routing and signaling information. How this information is 318 provided to the I2RS agent is out of scope, but the standardized 319 information and data models for what is exposed are part of I2RS. 321 Static System State: An I2RS agent needs access to static state on 322 a routing element beyond what is contained in the routing 323 subsystem. An example of such state is specifying queueing 324 behavior for an interface or traffic. How the I2RS agent modifies 325 or obtains this information is out of scope, but the standardized 326 information and data models for what is exposed are part of I2RS. 328 I2RS Agent: See the definition in Section 2. 330 Application: A network application that needs to observe the 331 network or manipulate the network to achieve its service 332 requirements. 334 I2RS Client: See the definition in Section 2. 336 As can be seen in Figure 1, an I2RS client can communicate with 337 multiple I2RS agents. An I2RS client may connect to one or more I2RS 338 agents based upon its needs. Similarly, an I2RS agent may 339 communicate with multiple I2RS clients - whether to respond to their 340 requests, to send notifications, etc. Timely notifications are 341 critical so that several simultaneously operating applications have 342 up-to-date information on the state of the network. 344 As can also be seen in Figure 1, an I2RS Agent may communicate with 345 multiple clients. Each client may send the agent a variety of write 346 operations. In order to keep the protocol simple, two clients should 347 not attempt to write (modify) the same piece of information on an 348 I2RS Agent. This is considered an error. However, such collisions 349 may happen and section 7.8 (multi-headed control) describes how the 350 I2RS agent resolves collision by first utilizing priority to resolve 351 collisions, and second by servicing the requests in a first in, first 352 served basis. The i2rs architecture includes this definition of 353 behavior for this case simply for predictability not because this is 354 an intended result. This predictability will simplify the error 355 handling and suppress oscillations. If additional error cases beyond 356 this simple treatment are required, these error cases should be 357 resolved by the network applications and management systems. 359 In contrast, although multiple I2RS clients may need to supply data 360 into the same list (e.g. a prefix or filter list), this is not 361 considered an error and must be correctly handled. The nuances so 362 that writers do not normally collide should be handled in the 363 information models. 365 The architectural goal for the I2RS is that such errors should 366 produce predictable behaviors, and be reportable to interested 367 clients. The details of the associated policy is discussed in 368 Section 7.8. The same policy mechanism (simple priority per I2RS 369 client) applies to interactions between the I2RS agent and the 370 CLI/SNMP/NETCONF as described in Section 6.3. 372 In addition it must be noted that there may be indirect interactions 373 between write operations. A basic example of this is when two 374 different but overlapping prefixes are written with different 375 forwarding behavior. Detection and avoidance of such interactions is 376 outside the scope of the I2RS work and is left to agent design and 377 implementation. 379 2. Terminology 381 The following terminology is used in this document. 383 agent or I2RS Agent: An I2RS agent provides the supported I2RS 384 services from the local system's routing sub-systems by 385 interacting with the routing element to provide specified 386 behavior. The I2RS agent understands the I2RS protocol and can be 387 contacted by I2RS clients. 389 client or I2RS Client: A client implements the I2RS protocol, uses 390 it to communicate with I2RS Agents, and uses the I2RS services to 391 accomplish a task. It interacts with other elements of the 392 policy, provisioning, and configuration system by means outside of 393 the scope of the I2RS effort. It interacts with the I2RS agents 394 to collect information from the routing and forwarding system. 395 Based on the information and the policy oriented interactions, the 396 I2RS client may also interact with I2RS agents to modify the state 397 of their associated routing systems to achieve operational goals. 398 An I2RS client can be seen as the part of an application that uses 399 and supports I2RS and could be a software library. 401 service or I2RS Service: For the purposes of I2RS, a service refers 402 to a set of related state access functions together with the 403 policies that control their usage. The expectation is that a 404 service will be represented by a data-model. For instance, 'RIB 405 service' could be an example of a service that gives access to 406 state held in a device's RIB. 408 read scope: The set of information which the I2RS client is 409 authorized to read. The read scope specifies the access 410 restrictions to both see the existence of data and read the value 411 of that data. 413 notification scope: The set of events and associated information 414 that the I2RS Client can request be pushed by the I2RS Agent. 415 I2RS Clients have the ability to register for specific events and 416 information streams, but must be constrained by the access 417 restrictions associated with their notification scope. 419 write scope: The set of field values which the I2RS client is 420 authorized to write (i.e. add, modify or delete). This access can 421 restrict what data can be modified or created, and what specific 422 value sets and ranges can be installed. 424 scope: When unspecified as either read scope, write scope, or 425 notification scope, the term scope applies to the read scope, 426 write scope, and notification scope. 428 resources: A resource is an I2RS-specific use of memory, storage, 429 or execution that a client may consume due to its I2RS operations. 430 The amount of each such resource that a client may consume in the 431 context of a particular agent may be constrained based upon the 432 client's security role. An example of such a resource could 433 include the number of notifications registered for. These are not 434 protocol-specific resources or network-specific resources. 436 role or security role: A security role specifies the scope, 437 resources, priorities, etc. that a client or agent has. Multiple 438 identities may use the same security role. 440 identity: A client is associated with exactly one specific 441 identity. State can be attributed to a particular identity. It 442 is possible for multiple communication channels to use the same 443 identity; in that case, the assumption is that the associated 444 client is coordinating such communication. 446 secondary identity: An I2RS Client may supply a secondary opaque 447 identity that is not interpreted by the I2RS Agent. An example 448 use is when the I2RS Client is a go-between for multiple 449 applications and it is necessary to track which application has 450 requested a particular operation. 452 Groups: NETCONF Network Access [RFC6536] refers uses the term group 453 in terms of an Administrative group which supports support the 454 well-established distinction between a root account and other 455 types of less-privileged conceptual user accounts. Group still 456 refers to a single identity (e.g. root) which is shared by a group 457 of users. 459 3. Key Architectural Properties 461 Several key architectural properties for the I2RS protocol are 462 elucidated below (simplicity, extensibility, and model-driven 463 programmatic interfaces). However, some architecture principles such 464 as performance and scaling are not described below because they are 465 discussed in [I-D.ietf-i2rs-problem-statement] and because the 466 performance and scaling requires varies based on the particular use- 467 cases. 469 3.1. Simplicity 471 There have been many efforts over the years to improve the access to 472 the information available to the routing and forwarding system. 473 Making such information visible and usable to network management and 474 applications has many well-understood benefits. There are two 475 related challenges in doing so. First, the quantity and diversity of 476 information potentially available is very large. Second, the 477 variation both in the structure of the data and in the kinds of 478 operations required tends to introduce protocol complexity. 480 While the types of operations contemplated here are complex in their 481 nature, it is critical that I2RS be easily deployable and robust. 482 Adding complexity beyond what is needed to satisfy well known and 483 understood requirements would hinder the ease of implementation, the 484 robustness of the protocol, and the deployability of the protocol. 485 Overly complex data models tend to ossify information sets by 486 attempting to describe and close off every possible option, 487 complicating extensibility. 489 Thus, one of the key aims for I2RS is the keep the protocol and 490 modeling architecture simple. So for each architectural component or 491 aspect, we ask ourselves "do we need this complexity, or is the 492 behavior merely nice to have?" Protocol parsimony is clearly a goal. 494 3.2. Extensibility 496 Naturally, extensibility of the protocol and data model is very 497 important. In particular, given the necessary scope limitations of 498 the initial work, it is critical that the initial design include 499 strong support for extensibility. 501 The scope of the I2RS work is being restricted in the interests of 502 achieving a deliverable and deployable result. The I2RS Working 503 Group is modeling only a subset of the data of interest. It is 504 clearly desirable for the data models defined in the I2RS to be 505 useful in more general settings. It should be easy to integrate data 506 models from the I2RS with other data. Other work should be able to 507 easily extend it to represent additional aspects of the network 508 elements or network systems. This reinforces the criticality of 509 designing the data models to be highly extensible, preferably in a 510 regular and simple fashion. 512 The I2RS Working Group is defining operations for the I2RS protocol. 513 It would be optimistic to assume that more and different ones may not 514 be needed when the scope of I2RS increases. Thus, it is important to 515 consider extensibility not only of the underlying services' data 516 models, but also of the primitives and protocol operations. 518 3.3. Model-Driven Programmatic Interfaces 520 A critical component of I2RS is the standard information and data 521 models with their associated semantics. While many components of the 522 routing system are standardized, associated data models for them are 523 not yet available. Instead, each router uses different information, 524 different mechanisms, and different CLI which makes a standard 525 interface for use by applications extremely cumbersome to develop and 526 maintain. Well-known data modeling languages exist and may be used 527 for defining the data models for I2RS. 529 There are several key benefits for I2RS in using model-driven 530 architecture and protocol(s). First, it allows for transferring 531 data-models whose content is not explicitly implemented or 532 understood. Second, tools can automate checking and manipulating 533 data; this is particularly valuable for both extensibility and for 534 the ability to easily manipulate and check proprietary data-models. 536 The different services provided by I2RS can correspond to separate 537 data-models. An I2RS agent may indicate which data-models are 538 supported. 540 4. Security Considerations 542 This I2RS architecture describes interfaces that clearly require 543 serious consideration of security. First, here is a brief 544 description of the assumed security environment for I2RS. The I2RS 545 Agent associated with a Routing Element is a trusted part of that 546 Routing Element. For example, it may be part of a vendor-distributed 547 signed software image for the entire Routing Element or it may be 548 trusted signed application that an operator has installed. The I2RS 549 Agent is assumed to have a separate authentication and authorization 550 channel by which it can validate both the identity and permissions 551 associated with an I2RS Client. To support numerous and speedy 552 interactions between the I2RS Agent and I2RS Client, it is assumed 553 that the I2RS Agent can also cache that particular I2RS Clients are 554 trusted and their associated authorized scope. This implies that the 555 permission information may be old either in a pull model until the 556 I2RS Agent re-requests it, or in a push model until the 557 authentication and authorization channel can notify the I2RS Agent of 558 changes. 560 Mutual authentication between the I2RS Client and I2RS Agent is 561 required. An I2RS Client must be able to trust that the I2RS Agent 562 is attached to the relevant Routing Element so that write/modify 563 operations are correctly applied and so that information received 564 from the I2RS Agent can be trusted by the I2RS Client. 566 An I2RS Client is not automatically trustworthy. It has identity 567 information and applications using that I2RS Client should be aware 568 of the scope limitations of that I2RS Client. If the I2RS Client is 569 acting as a broker for multiple applications, managing the security, 570 authentication and authorization for that communication is out of 571 scope; nothing prevents I2RS and a separate authentication and 572 authorization channel from being used. Regardless of mechanism, an 573 I2RS Client that is acting as a broker is responsible for determining 574 that applications using it are trusted and permitted to make the 575 particular requests. 577 Different levels of integrity, confidentiality, and replay protection 578 are relevant for different aspects of I2RS. The primary 579 communication channel that is used for client authentication and then 580 used by the client to write data requires integrity, confidentiality 581 and replay protection. Appropriate selection of a default required 582 transport protocol is the preferred way of meeting these 583 requirements. 585 Other communications via I2RS may not require integrity, 586 confidentiality, and replay protection. For instance, if an I2RS 587 Client subscribes to an information stream of prefix announcements 588 from OSPF, those may require integrity but probably not 589 confidentiality or replay protection. Similarly, an information 590 stream of interface statistics may not even require guaranteed 591 delivery. In Section 7.2, more reasoning for multiple communication 592 channels is provided. From the security perspective, it is critical 593 to realize that an I2RS Agent may open a new communication channel 594 based upon information provided by an I2RS Client (as described in 595 Section 7.2). For example, a I2RS client may request notifications 596 of certain events and the agent will open a communication channel to 597 report such events. Therefore, to avoid an indirect attack, such a 598 request must be done in the context of an authenticated and 599 authorized client whose communications cannot have been altered. 601 4.1. Identity and Authentication 603 As discussed above, all control exchanges between the I2RS client and 604 agent should be authenticated and integrity protected (such that the 605 contents cannot be changed without detection). Further, manipulation 606 of the system must be accurately attributable. In an ideal 607 architecture, even information collection and notification should be 608 protected; this may be subject to engineering tradeoffs during the 609 design. 611 I2RS clients may be operating on behalf of other applications. While 612 those applications' identities are not needed for authentication or 613 authorization, each application should have a unique opaque 614 identifier that can be provided by the I2RS client to the I2RS agent 615 for purposes of tracking attribution of operations to support 616 functionality such as troubleshooting and logging of network changes. 618 4.2. Authorization 620 All operations using I2RS, both observation and manipulation, should 621 be subject to appropriate authorization controls. Such authorization 622 is based on the identity and assigned role of the I2RS client 623 performing the operations and the I2RS agent in the network element. 624 (Multiple Identities may use the same role). 626 I2RS Agents, in performing information collection and manipulation, 627 will be acting on behalf of the I2RS clients. As such, each 628 operation authorization will be based on the lower of the two 629 permissions of the agent itself and of the authenticated client. The 630 mechanism by which this authorization is applied within the device is 631 outside of the scope of I2RS. 633 The appropriate or necessary level of granularity for scope can 634 depend upon the particular I2RS Service and the implementation's 635 granularity. An approach to a similar access control problem is 636 defined in the NetConf Access Control Model (NACM) [RFC6536]; it 637 allows arbitrary access to be specified for a data node instance 638 identifier while defining meaningful manipulable defaults. The 639 identity within NACM [RFC6536] can be specify as either a user name 640 or a group user name (e.g. Root), and this name is linked a scope 641 policy that contained in a a set of access control rules. Similarly, 642 it is expected the I2RS identity links to one role which has a scope 643 policy specified by a set of access control rules. This scope policy 644 is can be provided via Local Config, exposed as an I2RS Service for 645 manipulation by authorized clients, or via some other method (e.g. 646 AAA service) 648 When an I2RS client is authenticated, its identity is provided to the 649 I2RS Agent, and this identity links to a role which links to the 650 scope policy. Multiple identities may link to the same role (e.g 651 ability to read I2RS RIB). 653 4.3. Client Redundancy 655 I2RS must support client redundancy. At the simplest, this can be 656 handled by having a primary and a backup network application that 657 both use the same client identity and can successfully authenticate 658 as such. Since I2RS does not require a continuous transport 659 connection and supports multiple transport sessions, this can provide 660 some basic redundancy. However, it does not address the need for 661 troubleshooting and logging of network changes to be informed about 662 which network application is actually active. At a minimum, basic 663 transport information about each connection and time can be logged 664 with the identity. 666 5. Network Applications and I2RS Client 668 I2RS is expected to be used by network-oriented applications in 669 different architectures. While the interface between a network- 670 oriented application and the I2RS client is outside the scope of 671 I2RS, considering the different architectures is important to 672 sufficiently specify I2RS. 674 In the simplest architecture of direct access, a network-oriented 675 application has an I2RS client as a library or driver for 676 communication with routing elements. 678 In the broker architecture, multiple network-oriented applications 679 communicate in an unspecified fashion to a broker application that 680 contains an I2RS Client. That broker application requires additional 681 functionality for authentication and authorization of the network- 682 oriented applications; such functionality is out of scope for I2RS 683 but similar considerations to those described in Section 4.2 do 684 apply. As discussed in Section 4.1, the broker I2RS Client should 685 determine distinct opaque identifiers for each network-oriented 686 application that is using it. The broker I2RS Client can pass along 687 the appropriate value as a secondary identifier which can be used for 688 tracking attribution of operations. 690 In a third architecture, a routing element or network-oriented 691 application that uses an I2RS Client to access services on a 692 different routing element may also contain an I2RS agent to provide 693 services to other network-oriented applications. However, where the 694 needed information and data models for those services differs from 695 that of a conventional routing element, those models are, at least 696 initially, out of scope for I2RS. Below is an example of such a 697 network application 699 5.1. Example Network Application: Topology Manager 701 A Topology Manager includes an I2RS client that uses the I2RS data 702 models and protocol to collect information about the state of the 703 network by communicating directly with one or more I2RS agents. From 704 these I2RS agents, the Topology Manager collects routing 705 configuration and operational data, such as interface and label- 706 switched path (LSP) information. In addition, the Topology Manager 707 may collect link-state data in several ways - either via I2RS models, 708 by peering with BGP-LS[I-D.ietf-idr-ls-distribution] or listening 709 into the IGP. 711 The set of functionality and collected information that is the 712 Topology Manager may be embedded as a component of a larger 713 application, such as a path computation application. As a stand- 714 alone application, the Topology Manager could be useful to other 715 network applications by providing a coherent picture of the network 716 state accessible via another interface. That interface might use the 717 same I2RS protocol and could provide a topology service using 718 extensions to the I2RS data models. 720 6. I2RS Agent Role and Functionality 722 The I2RS Agent is part of a routing element. As such, it has 723 relationships with that routing element as a whole, and with various 724 components of that routing element. 726 6.1. Relationship to its Routing Element 728 A Routing Element may be implemented with a wide variety of different 729 architectures: an integrated router, a split architecture, 730 distributed architecture, etc. The architecture does not need to 731 affect the general I2RS agent behavior. 733 For scalability and generality, the I2RS agent may be responsible for 734 collecting and delivering large amounts of data from various parts of 735 the routing element. Those parts may or may not actually be part of 736 a single physical device. Thus, for scalability and robustness, it 737 is important that the architecture allow for a distributed set of 738 reporting components providing collected data from the I2RS agent 739 back to the relevant I2RS clients. There may be multiple I2RS Agents 740 within the same router. In such a case, they must have non- 741 overlapping sets of information which they manipulate. 743 To facilitate operations, deployment and troubleshooting, it is 744 important that traceability of the I2RS Agent's requests and actions 745 be supported via a common data model. 747 6.2. I2RS State Storage 749 State modification requests are sent to the I2RS agent in a routing 750 element by I2RS clients. The I2RS agent is responsible for applying 751 these changes to the system, subject to the authorization discussed 752 above. The I2RS agent will retain knowledge of the changes it has 753 applied, and the client on whose behalf it applied the changes. The 754 I2RS agent will also store active subscriptions. These sets of data 755 form the I2RS data store. This data is retained by the agent until 756 the state is removed by the client, overridden by some other 757 operation such as CLI, or the device reboots. Meaningful logging of 758 the application and removal of changes is recommended. I2RS applied 759 changes to the routing element state will not be retained across 760 routing element reboot. The I2RS data store is not preserved across 761 routing element reboots; thus the I2RS agent will not attempt to 762 reapply such changes after a reboot. 764 6.2.1. I2RS Agent Failure 766 It is expected that an I2RS Agent may fail independently of the 767 associated routing element. This could happen because I2RS is 768 disabled on the routing element or because the I2RS Agent, a separate 769 process or even running on a separate processor, experiences an 770 unexpected failure. Just as routing state learned from a failed 771 source is removed, the ephemeral I2RS state will usually be removed 772 shortly after the failure is detected or as part of a graceful 773 shutdown process. To handle I2RS Agent failure, the I2RS Agent must 774 use two different notifications. 776 NOTIFICATION_I2RS_AGENT_STARTING: This notification identifies that 777 the associated I2RS Agent has started. It includes an agent-boot- 778 count that indicates how many times the I2RS Agent has restarted 779 since the associated routing element restarted. The agent-boot- 780 count allows an I2RS Client to determine if the I2RS Agent has 781 restarted. 783 NOTIFICATION_I2RS_AGENT_TERMINATING: This notification reports that 784 the associated I2RS Agent is shutting down gracefully. Ephemeral 785 state will be removed. It can optionally include a timestamp 786 indicating when the I2RS Agent will shutdown. Use of this 787 timestamp assumes that time synchronization has been done and the 788 timestamp should not have granularity finer than one second 789 because better accuracy of shutdown time is not guaranteed. 791 There are two different failure types that are possible and each has 792 different behavior. 794 Unexpected failure: In this case, the I2RS Agent has unexpectedly 795 crashed and thus cannot notify its clients of anything. Since 796 I2RS does not require a persistent connection between the I2RS 797 Client and I2RS Agent, it is necessary to have a mechanism for the 798 I2RS Agent to notify I2RS Clients that had subscriptions or 799 written ephemeral state; such I2RS Clients should be cached by the 800 I2RS Agent's system in persistent storage. When the I2RS Agent 801 starts, it should send a NOTIFICATION_I2RS_AGENT_STARTING to each 802 cached I2RS Client. 804 Graceful failure: In this case, the I2RS Agent can do specific 805 limited work as part of the process of being disabled. The I2RS 806 Agent must send a NOTIFICATION_I2RS_AGENT_TERMINATING to all its 807 cached I2RS Clients. 809 6.2.2. Starting and Ending 811 When an I2RS client applies changes via the I2RS protocol, those 812 changes are applied and left until removed or the routing element 813 reboots. The network application may make decisions about what to 814 request via I2RS based upon a variety of conditions that imply 815 different start times and stop times. That complexity is managed by 816 the network application and is not handled by I2RS. 818 6.2.3. Reversion 820 An I2RS Agent may decide that some state should no longer be applied. 821 An I2RS Client may instruct an Agent to remove state it has applied. 822 In all such cases, the state will revert to what it would have been 823 without the I2RS client-agent interaction; that state is generally 824 whatever was specified via the CLI, NETCONF, SNMP, etc. I2RS Agents 825 will not store multiple alternative states, nor try to determine 826 which one among such a plurality it should fall back to. Thus, the 827 model followed is not like the RIB, where multiple routes are stored 828 at different preferences. (For I2RS state in the presence of two 829 I2RS clients, please see section 1.2 and section 7.8) 831 An I2RS Client may register for notifications, subject to its 832 notification scope, regarding state modification or removal by a 833 particular I2RS Client. 835 6.3. Interactions with Local Config 837 Changes may originate from either Local Config or from I2RS. The 838 modifications and data stored by I2RS are separate from the local 839 device configuration, but conflicts between the two must be resolved 840 in a deterministic manner that respects operator-applied policy. 841 That policy can determine whether Local Config overrides a particular 842 I2RS client's request or vice versa. To achieve this end, either by 843 default Local Config always wins or, optionally, a routing element 844 may permit a priority to be configured on the device for the Local 845 Config mechanism. The policy mechanism in the later case is 846 comparing the I2RS client's priority with that priority assigned to 847 the Local Config. 849 When the Local Config always wins, some communication between that 850 subsystem and the I2RS Agent is still necessary. That communication 851 contains the details of each specific device configuration change 852 that the I2RS Agent is permitted to modify. In addition, when the 853 system determines, that a client's I2RS state is preempted, the I2RS 854 agent must notify the affected I2RS clients; how the system 855 determines this is implementation-dependent. 857 It is critical that policy based upon the source is used because the 858 resolution cannot be time-based. Simply allowing the most recent 859 state to prevail could cause race conditions where the final state is 860 not repeatably deterministic. 862 6.4. Routing Components and Associated I2RS Services 864 For simplicity, each logical protocol or set of functionality that 865 can be compactly described in a separable information and data model 866 is considered as a separate I2RS Service. A routing element need not 867 implement all routing components described nor provide the associated 868 I2RS services. I2RS Services should include a capability model so 869 that peers can determine which parts of the service are supported. 870 Each I2RS Service requires an information model that describes at 871 least the following: data that can be read, data that can be written, 872 notifications that can be subscribed to, and the capability model 873 mentioned above. 875 The initial services included in the I2RS architecture are as 876 follows. 878 *************************** ************** ***************** 879 * I2RS Protocol * * * * Dynamic * 880 * * * Interfaces * * Data & * 881 * +--------+ +-------+ * * * * Statistics * 882 * | Client | | Agent | * ************** ***************** 883 * +--------+ +-------+ * 884 * * ************** ************* 885 *************************** * * * * 886 * Policy * * Base QoS * 887 ******************** ******** * Templates * * Templates * 888 * +--------+ * * * * * ************* 889 * BGP | BGP-LS | * * PIM * ************** 890 * +--------+ * * * 891 ******************** ******** **************************** 892 * MPLS +---------+ +-----+ * 893 ********************************** * | RSVP-TE | | LDP | * 894 * IGPs +------+ +------+ * * +---------+ +-----+ * 895 * +--------+ | OSPF | | ISIS | * * +--------+ * 896 * | Common | +------+ +------+ * * | Common | * 897 * +--------+ * * +--------+ * 898 ********************************** **************************** 900 ************************************************************** 901 * RIB Manager * 902 * +-------------------+ +---------------+ +------------+ * 903 * | Unicast/multicast | | Policy-Based | | RIB Policy | * 904 * | RIBs & LIBs | | Routing | | Controls | * 905 * | route instances | | (ACLs, etc) | +------------+ * 906 * +-------------------+ +---------------+ * 907 ************************************************************** 909 Figure 2: Anticipated I2RS Services 911 There are relationships between different I2RS Services - whether 912 those be the need for the RIB to refer to specific interfaces, the 913 desire to refer to common complex types (e.g. links, nodes, IP 914 addresses), or the ability to refer to implementation-specific 915 functionality (e.g. pre-defined templates to be applied to interfaces 916 or for QoS behaviors that traffic is direct into). Section 6.4.5 917 discusses information modeling constructs and the range of 918 relationship types that are applicable. 920 6.4.1. Routing and Label Information Bases 922 Routing elements may maintain one or more Information Bases. 923 Examples include Routing Information Bases such as IPv4/IPv6 Unicast 924 or IPv4/IPv6 Multicast. Another such example includes the MPLS Label 925 Information Bases, per-platform or per-interface. This 926 functionality, exposed via an I2RS Service, must interact smoothly 927 with the same mechanisms that the routing element already uses to 928 handle RIB input from multiple sources, so as to safely change the 929 system state. Conceptually, this can be handled by having the I2RS 930 Agent communicate with a RIB Manager as a separate routing source. 932 The point-to-multipoint state added to the RIB does not need to match 933 to well-known multicast protocol installed state. The I2RS Agent can 934 create arbitrary replication state in the RIB, subject to the 935 advertised capabilities of the routing element. 937 6.4.2. IGPs, BGP and Multicast Protocols 939 A separate I2RS Service can expose each routing protocol on the 940 device. Such I2RS services may include a number of different kinds 941 of operations: 943 o reading the various internal RIB(s) of the routing protocol is 944 often helpful for understanding the state of the network. 945 Directly writing to these protocol-specific RIBs or databases is 946 out of scope for I2RS. 948 o reading the various pieces of policy information the particular 949 protocol instance is using to drive its operations. 951 o writing policy information such as interface attributes that are 952 specific to the routing protocol or BGP policy that may indirectly 953 manipulate attributes of routes carried in BGP. 955 o writing routes or prefixes to be advertised via the protocol. 957 o joining/removing interfaces from the multicast trees 959 o subscribing to an information stream of route changes 961 o receiving notifications about peers coming up or going down 963 For example, the interaction with OSPF might include modifying the 964 local routing element's link metrics, announcing a locally-attached 965 prefix, or reading some of the OSPF link-state database. However, 966 direct modification of the link-state database must not be allowed in 967 order to preserve network state consistency. 969 6.4.3. MPLS 971 I2RS Services will be needed to expose the protocols that create 972 transport LSPs (e.g. LDP and RSVP-TE) as well as protocols (e.g. 973 BGP, LDP) that provide MPLS-based services (e.g. pseudowires, L3VPNs, 974 L2VPNs, etc). This should include all local information about LSPs 975 originating in, transiting, or terminating in this Routing Element. 977 6.4.4. Policy and QoS Mechanisms 979 Many network elements have separate policy and QoS mechanisms, 980 including knobs which affect local path computation and queue control 981 capabilities. These capabilities vary widely across implementations, 982 and I2RS cannot model the full range of information collection or 983 manipulation of these attributes. A core set does need to be 984 included in the I2RS information models and supported in the expected 985 interfaces between the I2RS Agent and the network element, in order 986 to provide basic capabilities and the hooks for future extensibility. 988 By taking advantage of extensibility and sub-classing, information 989 models can specify use of a basic model that can be replaced by a 990 more detailed model. 992 6.4.5. Information Modeling, Device Variation, and Information 993 Relationships 995 I2RS depends heavily on information models of the relevant aspects of 996 the Routing Elements to be manipulated. These models drive the data 997 models and protocol operations for I2RS. It is important that these 998 informational models deal well with a wide variety of actual 999 implementations of Routing Elements, as seen between different 1000 products and different vendors. There are three ways that I2RS 1001 information models can address these variations: class or type 1002 inheritance, optional features, and templating. 1004 6.4.5.1. Managing Variation: Object Classes/Types and Inheritance 1006 Information modelled by I2RS from a Routing Element can be described 1007 in terms of classes or types or object. Different valid inheritance 1008 definitions can apply. What is appropriate for I2RS to use is not 1009 determined in this architecture; for simplicity, class and subclass 1010 will be used as the example terminology. This I2RS architecture does 1011 require the ability to address variation in Routing Elements by 1012 allowing information models to define parent or base classes and 1013 subclasses. 1015 The base or parent class defines the common aspects that all Routing 1016 Elements are expected to support. Individual subclasses can 1017 represent variations and additional capabilities. When applicable, 1018 there may be several levels of refinement. The I2RS protocol can 1019 then provide mechanisms to allow an I2RS client to determine which 1020 classes a given I2RS Agent has available. Clients which only want 1021 basic capabilities can operate purely in terms of base or parent 1022 classes, while a client needing more details or features can work 1023 with the supported sub-class(es). 1025 As part of I2RS information modeling, clear rules should be specified 1026 for how the parent class and subclass can relate; for example, what 1027 changes can a subclass make to its parent? The description of such 1028 rules should be done so that it can apply across data modeling tools 1029 until the I2RS data modeling language is selected. 1031 6.4.5.2. Managing Variation: Optionality 1033 I2RS Information Models must be clear about what aspects are 1034 optional. For instance, must an instance of a class always contain a 1035 particular data field X? If so, must the client provide a value for 1036 X when creating the object or is there a well-defined default value? 1037 From the Routing Element perspective, in the above example, each 1038 Information model should provide information that: 1040 o Is X required for the data field to be accepted and applied? 1042 o If X is optional, then how does "X" as an optional portion of data 1043 field interact with the required aspects of the data field? 1045 o Does the data field have defaults for the mandatory portion of the 1046 field and the optional portions of the field 1048 o Is X required to be within a particular set of values (e.g. range, 1049 length of strings)? 1051 The information model needs to be clear about what read or write 1052 values are set by client and what responses or actions are required 1053 by the agent. It is important to indicate what is required or 1054 optional in client values and agent responses/actions. 1056 6.4.5.3. Managing Variation: Templating 1058 A template is a collection of information to address a problem; it 1059 cuts across the notions of class and object instances. A template 1060 provides a set of defined values for a set of information fields and 1061 can specify a set of values that must be provided to complete the 1062 template. Further, a flexible template scheme may that some of the 1063 defined values can be over-written. 1065 For instance, assigning traffic to a particular service class might 1066 be done by specifying a template Queueing with a parameter to 1067 indicate Gold, Silver, or Best Effort. The details of how that is 1068 carried out are not modeled. This does assume that the necessary 1069 templates are made available on the Routing Element via some 1070 mechanism other than I2RS. The idea is that by providing suitable 1071 templates for tasks that need to be accomplished, with templates 1072 implemented differently for different kinds of Routing Elements, the 1073 client can easily interact with the Routing Element without concern 1074 for the variations which are handled by values included in the 1075 template. 1077 If implementation variation can be exposed in other ways, templates 1078 may not be needed. However, templates themselves could be objects 1079 referenced in the protocol messages, with Routing Elements being 1080 configured with the proper templates to complete the operation. This 1081 is a topic for further discussion. 1083 6.4.5.4. Object Relationships 1085 Objects (in a Routing Element or otherwise) do not exist in 1086 isolation. They are related to each other. One of the important 1087 things a class definition does is represent the relationships between 1088 instances of different classes. These relationships can be very 1089 simple, or quite complicated. The following lists the information 1090 relationships that the information models need to support. 1092 6.4.5.4.1. Initialization 1094 The simplest relationship is that one object instance is initialized 1095 by copying another. For example, one may have an object instance 1096 that represents the default setup for a tunnel, and all new tunnels 1097 have fields copied from there if they are not set as part of 1098 establishment. This is closely related to the templates discussed 1099 above, but not identical. Since the relationship is only momentary 1100 it is often not formally represented in modeling, but only captured 1101 in the semantic description of the default object. 1103 6.4.5.4.2. Correlation Identification 1105 Often, it suffices to indicate in one object that it is related to a 1106 second object, without having a strong binding between the two. So 1107 an Identifier is used to represent the relationship. This can be 1108 used to allow for late binding, or a weak binding that does not even 1109 need to exist. A policy name in an object might indicate that if a 1110 policy by that name exists, it is to be applied under some 1111 circumstance. In modeling this is often represented by the type of 1112 the value. 1114 6.4.5.4.3. Object References 1116 Sometimes the relationship between objects is stronger. A valid ARP 1117 entry has to point to the active interface over which it was derived. 1118 This is the classic meaning of an object reference in programming. 1119 It can be used for relationships like containment or dependence. 1120 This is usually represented by an explicit modeling link. 1122 6.4.5.4.4. Active Reference 1124 There is an even stronger form of coupling between objects if changes 1125 in one of the two objects are always to be reflected in the state of 1126 the other. For example, if a Tunnel has an MTU (maximum transmit 1127 unit), and link MTU changes need to immediately propagate to the 1128 Tunnel MTU, then the tunnel is actively coupled to the link 1129 interface. This kind of active state coupling implies some sort of 1130 internal bookkeeping to ensure consistency, often conceptualized as a 1131 subscription model across objects. 1133 7. I2RS Client Agent Interface 1135 7.1. One Control and Data Exchange Protocol 1137 As agreed by the I2RS working group, this I2RS architecture assumes 1138 that there is a single I2RS protocol for control and data exchange; 1139 that protocol will be based on NETCONF[RFC6241] and RESTCONF 1140 [I-D.ietf-netconf-restconf]. This helps meet the goal of simplicity 1141 and thereby enhances deployability. That protocol may need to use 1142 several underlying transports (TCP, SCTP (stream control transport 1143 protocol), DCCP (Datagram Congestion Control Protocol)), with 1144 suitable authentication and integrity protection mechanisms. These 1145 different transports can support different types of communication 1146 (e.g. control, reading, notifications, and information collection) 1147 and different sets of data. Whatever transport is used for the data 1148 exchange, it must also support suitable congestion control 1149 mechanisms. The transports chosen should be operator and implementor 1150 friendly to ease adoption. 1152 7.2. Communication Channels 1154 Multiple communication channels and multiple types of communication 1155 channels are required. There may be a range of requirements (e.g. 1156 confidentiality, reliability), and to support the scaling there may 1157 need to be channels originating from multiple sub-components of a 1158 routing element and/or to multiple parts of an I2RS client. All such 1159 communication channels will use the same higher level protocol. Use 1160 of additional channels for communication will be coordinated between 1161 the I2RS client and the I2RS agent. 1163 I2RS protocol communication can be delivered in-band via the routing 1164 system's data plane. I2RS protocol communication might be delivered 1165 out-of-band via a management interface. Depending on what operations 1166 are requested, it is possible for the I2RS protocol communication to 1167 cause the in-band communication channels to stop working; this could 1168 cause the I2RS agent to become unreachable across that communication 1169 channel. 1171 7.3. Capability Negotiation 1173 The support for different protocol capabilities and I2RS Services 1174 will vary across I2RS Clients and Routing Elements supporting I2RS 1175 Agents. Since each I2RS Service is required to include a capability 1176 model (see Section 6.4), negotiation at the protocol level can be 1177 restricted to protocol specifics and which I2RS Services are 1178 supported. 1180 Capability negotiation (such as which transports are supported beyond 1181 the minimum required to implement) will clearly be necessary. It is 1182 important that such negotiations be kept simple and robust, as such 1183 mechanisms are often a source of difficulty in implementation and 1184 deployment. 1186 The protocol capability negotiation can be segmented into the basic 1187 version negotiation (required to ensure basic communication), and the 1188 more complex capability exchange which can take place within the base 1189 protocol mechanisms. In particular, the more complex protocol and 1190 mechanism negotiation can be addressed by defining information models 1191 for both the I2RS Agent and the I2RS Client. These information 1192 models can describe the various capability options. This can then 1193 represent and be used to communicate important information about the 1194 agent, and the capabilities thereof. 1196 7.4. Scope Policy Specifications 1198 As section 4.1 and 4.2 describe, each I2RS Client will have a unique 1199 identity and it may have a secondary identity (see section 2) to aid 1200 in troubleshooting. As section 4 indicates, all authentication and 1201 authorization mechanisms are based on the primary Identity which 1202 links to a role with scope policy for for reading data, for writing 1203 data, and limitations on the resources that can be consumed. 1204 Specifications for scope policy need to specify the data and value 1205 ranges for portion of scope policy. 1207 7.5. Connectivity 1209 A client may or may not maintain an active communication channel with 1210 an agent. Therefore, an agent may need to open a communication 1211 channel to the client to communicate previously requested 1212 information. The lack of an active communication channel does not 1213 imply that the associated client is non-functional. When 1214 communication is required, the agent or client can open a new 1215 communication channel. 1217 State held by an agent that is owned by a client should not be 1218 removed or cleaned up when a client is no longer communicating - even 1219 if the agent cannot successfully open a new communication channel to 1220 the client. 1222 For many applications, it may be desirable to clean up state if a 1223 network application dies before removing the state it has created. 1224 Typically, this is dealt with in terms of network application 1225 redundancy. If stronger mechanisms are desired, mechanisms outside 1226 of I2RS may allow a supervisory network application to monitor I2RS 1227 clients, and based on policy known to the supervisor clean up state 1228 if applications die. More complex mechanism instantiated in the I2RS 1229 agent would add complications to the I2RS protocol and are thus left 1230 for future work. 1232 Some examples of such a mechanism include the following. In one 1233 option, the client could request state clean-up if a particular 1234 transport session is terminated. The second is to allow state 1235 expiration, expressed as a policy associated with the I2RS client's 1236 role. The state expiration could occur after there has been no 1237 successful communication channel to or from the I2RS client for the 1238 policy-specified duration. 1240 7.6. Notifications 1242 As with any policy system interacting with the network, the I2RS 1243 Client needs to be able to receive notifications of changes in 1244 network state. Notifications here refers to changes which are 1245 unanticipated, represent events outside the control of the systems 1246 (such as interface failures on controlled devices), or are 1247 sufficiently sparse as to be anomalous in some fashion. A 1248 notification may also be due to a regular event. 1250 Such events may be of interest to multiple I2RS Clients controlling 1251 data handled by an I2RS Agent, and to multiple other I2RS clients 1252 which are collecting information without exerting control. The 1253 architecture therefore requires that it be practical for I2RS Clients 1254 to register for a range of notifications, and for the I2RS Agents to 1255 send notifications to a number of Clients. The I2RS Client should be 1256 able to filter the specific notifications that will be received; the 1257 specific types of events and filtering operations can vary by 1258 information model and need to be specified as part of the information 1259 model. 1261 The I2RS information model needs to include representation of these 1262 events. As discussed earlier, the capability information in the 1263 model will allow I2RS clients to understand which events a given I2RS 1264 Agent is capable of generating. 1266 For performance and scaling by the I2RS client and general 1267 information confidentiality, an I2RS Client needs to be able to 1268 register for just the events it is interested in. It is also 1269 possible that I2RS might provide a stream of notifications via a 1270 publish/subscribe mechanism that is not amenable to having the I2RS 1271 agent do the filtering. 1273 7.7. Information collection 1275 One of the other important aspects of the I2RS is that it is intended 1276 to simplify collecting information about the state of network 1277 elements. This includes both getting a snapshot of a large amount of 1278 data about the current state of the network element, and subscribing 1279 to a feed of the ongoing changes to the set of data or a subset 1280 thereof. This is considered architecturally separate from 1281 notifications due to the differences in information rate and total 1282 volume. 1284 7.8. Multi-Headed Control 1286 As was described earlier, an I2RS Agent interacts with multiple I2RS 1287 Clients who are actively controlling the network element. From an 1288 architecture and design perspective, the assumption is that by means 1289 outside of this system the data to be manipulated within the network 1290 element is appropriately partitioned so that any given piece of 1291 information is only being manipulated by a single I2RS Client. 1293 Nonetheless, unexpected interactions happen and two (or more) I2RS 1294 clients may attempt to manipulate the same piece of data. This is 1295 considered an error case. This architecture does not attempt to 1296 determine what the right state of data should be when such a 1297 collision happens. Rather, the architecture mandates that there be 1298 decidable means by which I2RS Agents handle the collisions. The 1299 mechanism for ensuring predictability is to have a simple priority 1300 associated with each I2RS clients, and the highest priority change 1301 remains in effect. In the case of priority ties, the first client 1302 whose attribution is associated with the data will keep control. 1304 In order for this approach to multi-headed control to be useful for 1305 I2RS Clients, it is important that it is possible for an I2RS Client 1306 to register for changes to any changes made by I2RS to data that it 1307 may care about. This is included in the I2RS event mechanisms. This 1308 also needs to apply to changes made by CLI/NETCONF/SNMP within the 1309 write-scope of the I2RS Agent, as the same priority mechanism (even 1310 if it is "CLI always wins") applies there. The I2RS client may then 1311 respond to the situation as it sees fit. 1313 7.9. Transactions 1315 In the interest of simplicity, the I2RS architecture does not include 1316 multi-message atomicity and rollback mechanisms. Rather, it includes 1317 a small range of error handling for a set of operations included in a 1318 single message. An I2RS Client may indicate one of the following 1319 three error handling for a given message with multiple operations 1320 which it sends to an I2RS Agent: 1322 Perform all or none: This traditional SNMP semantic indicates that 1323 other I2RS agent will keep enough state when handling a single 1324 message to roll back the operations within that message. Either 1325 all the operations will succeed, or none of them will be applied 1326 and an error message will report the single failure which caused 1327 them not to be applied. This is useful when there are, for 1328 example, mutual dependencies across operations in the message. 1330 Perform until error: In this case, the operations in the message 1331 are applied in the specified order. When an error occurs, no 1332 further operations are applied, and an error is returned 1333 indicating the failure. This is useful if there are dependencies 1334 among the operations and they can be topologically sorted. 1336 Perform all storing errors: In this case, the I2RS Agent will 1337 attempt to perform all the operations in the message, and will 1338 return error indications for each one that fails. This is useful 1339 when there is no dependency across the operation, or where the 1340 client would prefer to sort out the effect of errors on its own. 1342 In the interest of robustness and clarity of protocol state, the 1343 protocol will include an explicit reply to modification or write 1344 operations even when they fully succeed. 1346 8. Operational and Manageability Considerations 1348 In order to facilitate troubleshooting of routing elements 1349 implementing I2RS agents, those routing elements should provide for a 1350 mechanism to show actively provisioned I2RS state and other I2RS 1351 Agent internal information. Note that this information may contain 1352 highly sensitive material subject to the Security Considerations of 1353 any data models implemented by that Agent and thus must be protected 1354 according to those considerations. Preferably, this mechanism should 1355 use a different privileged means other than simply connecting as an 1356 I2RS client to learn the data. Using a different mechanism should 1357 improve traceability and failure management. 1359 Manageability plays a key aspect in I2RS. Some initial examples 1360 include: 1362 Resource Limitations: Using I2RS, applications can consume 1363 resources, whether those be operations in a time-frame, entries in 1364 the RIB, stored operations to be triggered, etc. The ability to 1365 set resource limits based upon authorization is important. 1367 Configuration Interactions: The interaction of state installed via 1368 the I2RS and via a router's configuration needs to be clearly 1369 defined. As described in this architecture, a simple priority 1370 that is configured is used to provide sufficient policy 1371 flexibility. 1373 9. IANA Considerations 1375 This document includes no request to IANA. 1377 10. Acknowledgements 1379 Significant portions of this draft came from draft-ward-i2rs- 1380 framework-00 and draft-atlas-i2rs-policy-framework-00. 1382 The authors would like to thank Nitin Bahadur, Shane Amante, Ed 1383 Crabbe, Ken Gray, Carlos Pignataro, Wes George, Ron Bonica, Joe 1384 Clarke, Juergen Schoenwalder, Jeff Haas, Jamal Hadi Salim, Scott 1385 Brim, Thomas Narten, Dean Bogdanovi, Tom Petch, Robert Raszuk, 1386 Sriganesh Kini, John Mattsson, Nancy Cam-Winget, DaCheng Zhang, Qin 1387 Wu, Ahmed Abro, Salman Asadullah, and Eric Yu. for their suggestions 1388 and review. 1390 11. Informative References 1392 [I-D.ietf-i2rs-problem-statement] 1393 Atlas, A., Nadeau, T., and D. Ward, "Interface to the 1394 Routing System Problem Statement", draft-ietf-i2rs- 1395 problem-statement-04 (work in progress), June 2014. 1397 [I-D.ietf-idr-ls-distribution] 1398 Gredler, H., Medved, J., Previdi, S., Farrel, A., and S. 1399 Ray, "North-Bound Distribution of Link-State and TE 1400 Information using BGP", draft-ietf-idr-ls-distribution-07 1401 (work in progress), November 2014. 1403 [I-D.ietf-netconf-restconf] 1404 Bierman, A., Bjorklund, M., and K. Watsen, "RESTCONF 1405 Protocol", draft-ietf-netconf-restconf-03 (work in 1406 progress), October 2014. 1408 [RFC6241] Enns, R., Bjorklund, M., Schoenwaelder, J., and A. 1409 Bierman, "Network Configuration Protocol (NETCONF)", RFC 1410 6241, June 2011. 1412 [RFC6536] Bierman, A. and M. Bjorklund, "Network Configuration 1413 Protocol (NETCONF) Access Control Model", RFC 6536, March 1414 2012. 1416 Authors' Addresses 1418 Alia Atlas 1419 Juniper Networks 1420 10 Technology Park Drive 1421 Westford, MA 01886 1422 USA 1424 Email: akatlas@juniper.net 1426 Joel Halpern 1427 Ericsson 1429 Email: Joel.Halpern@ericsson.com 1431 Susan Hares 1432 Huawei 1434 Email: shares@ndzh.com 1435 Dave Ward 1436 Cisco Systems 1437 Tasman Drive 1438 San Jose, CA 95134 1439 USA 1441 Email: wardd@cisco.com 1443 Thomas D. Nadeau 1444 Brocade 1446 Email: tnadeau@lucidvision.com