idnits 2.17.1 draft-ietf-sfc-architecture-00.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 (September 8, 2014) is 3489 days in the past. Is this intentional? Checking references for intended status: Proposed Standard ---------------------------------------------------------------------------- (See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) ** Obsolete normative reference: RFC 5226 (Obsoleted by RFC 8126) == Outdated reference: A later version (-13) exists of draft-ietf-sfc-problem-statement-10 Summary: 1 error (**), 0 flaws (~~), 2 warnings (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network Working Group J. Halpern, Ed. 3 Internet-Draft Ericsson 4 Intended status: Standards Track C. Pignataro, Ed. 5 Expires: March 12, 2015 Cisco 6 September 8, 2014 8 Service Function Chaining (SFC) Architecture 9 draft-ietf-sfc-architecture-00 11 Abstract 13 This document describes an architecture for the specification, 14 creation, and ongoing maintenance of Service Function Chains (SFC) in 15 a network. It includes architectural concepts, principles, and 16 components used in the construction of composite services through 17 deployment of SFCs. This document does not propose solutions, 18 protocols, or extensions to existing protocols. 20 Status of This Memo 22 This Internet-Draft is submitted in full conformance with the 23 provisions of BCP 78 and BCP 79. 25 Internet-Drafts are working documents of the Internet Engineering 26 Task Force (IETF). Note that other groups may also distribute 27 working documents as Internet-Drafts. The list of current Internet- 28 Drafts is at http://datatracker.ietf.org/drafts/current/. 30 Internet-Drafts are draft documents valid for a maximum of six months 31 and may be updated, replaced, or obsoleted by other documents at any 32 time. It is inappropriate to use Internet-Drafts as reference 33 material or to cite them other than as "work in progress." 35 This Internet-Draft will expire on March 12, 2015. 37 Copyright Notice 39 Copyright (c) 2014 IETF Trust and the persons identified as the 40 document authors. All rights reserved. 42 This document is subject to BCP 78 and the IETF Trust's Legal 43 Provisions Relating to IETF Documents 44 (http://trustee.ietf.org/license-info) in effect on the date of 45 publication of this document. Please review these documents 46 carefully, as they describe your rights and restrictions with respect 47 to this document. Code Components extracted from this document must 48 include Simplified BSD License text as described in Section 4.e of 49 the Trust Legal Provisions and are provided without warranty as 50 described in the Simplified BSD License. 52 Table of Contents 54 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2 55 1.1. Scope . . . . . . . . . . . . . . . . . . . . . . . . . . 3 56 1.2. Assumptions . . . . . . . . . . . . . . . . . . . . . . . 3 57 1.3. Definition of Terms . . . . . . . . . . . . . . . . . . . 4 58 2. Architectural Concepts . . . . . . . . . . . . . . . . . . . 6 59 2.1. Service Function Chains . . . . . . . . . . . . . . . . . 6 60 2.2. Service Function Chain Symmetry . . . . . . . . . . . . . 7 61 2.3. Service Function Paths . . . . . . . . . . . . . . . . . 8 62 3. Architecture Principles . . . . . . . . . . . . . . . . . . . 9 63 4. Core SFC Architecture Components . . . . . . . . . . . . . . 10 64 4.1. SFC Encapsulation . . . . . . . . . . . . . . . . . . . . 11 65 4.2. Service Function (SF) . . . . . . . . . . . . . . . . . . 12 66 4.3. Service Function Forwarder (SFF) . . . . . . . . . . . . 12 67 4.3.1. Transport Derived SFF . . . . . . . . . . . . . . . . 14 68 4.4. SFC-Enabled Domain . . . . . . . . . . . . . . . . . . . 14 69 4.5. Network Overlay and Network Components . . . . . . . . . 14 70 4.6. SFC Proxy . . . . . . . . . . . . . . . . . . . . . . . . 14 71 4.7. Classification . . . . . . . . . . . . . . . . . . . . . 16 72 4.8. Re-Classification and Branching . . . . . . . . . . . . . 16 73 4.9. Shared Metadata . . . . . . . . . . . . . . . . . . . . . 17 74 5. Additional Architectural Concepts . . . . . . . . . . . . . . 17 75 5.1. The Role of Policy . . . . . . . . . . . . . . . . . . . 17 76 5.2. SFC Control Plane . . . . . . . . . . . . . . . . . . . . 18 77 5.3. Resource Control . . . . . . . . . . . . . . . . . . . . 19 78 5.4. Infinite Loop Detection and Avoidance . . . . . . . . . . 19 79 5.5. Load Balancing Considerations . . . . . . . . . . . . . . 20 80 5.6. MTU and Fragmentation Considerations . . . . . . . . . . 21 81 5.7. SFC OAM . . . . . . . . . . . . . . . . . . . . . . . . . 21 82 5.8. Resilience and Redundancy . . . . . . . . . . . . . . . . 22 83 6. Security Considerations . . . . . . . . . . . . . . . . . . . 23 84 7. Contributors and Acknowledgments . . . . . . . . . . . . . . 23 85 8. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 25 86 9. References . . . . . . . . . . . . . . . . . . . . . . . . . 25 87 9.1. Normative References . . . . . . . . . . . . . . . . . . 25 88 9.2. Informative References . . . . . . . . . . . . . . . . . 25 89 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 25 91 1. Introduction 93 This document describes an architecture used for the creation and 94 ongoing maintenance of Service Function Chains (SFC) in a network. 95 It includes architectural concepts, principles, and components. 97 An overview of the issues associated with the deployment of end-to- 98 end service function chains, abstract sets of service functions and 99 their ordering constraints that create a composite service and the 100 subsequent "steering" of traffic flows through said service 101 functions, is described in [I-D.ietf-sfc-problem-statement]. 103 This architecture presents a model addressing the problematic aspects 104 of existing service deployments, including topological independence 105 and configuration complexity. 107 Service function chains enable composite services that are 108 constructed from one or more service functions. 110 1.1. Scope 112 This document defines a framework to realize Service Function 113 Chaining (SFC) with minimum requirements on the physical topology of 114 the network. The proposed solution relies on initial packet 115 classification. Packets are initially classified at the entry point 116 of an SFC-enabled domain, and are then forwarded according to the 117 ordered set of Service Functions (SFs) that need to be enabled to 118 process these packets in the SFC-enabled domain. 120 This document does not make any assumption on the deployment context. 121 The proposed framework covers both fixed and mobile networks. 123 The architecture described herein is assumed to be applicable to a 124 single network administrative domain. While it is possible for the 125 architectural principles and components to be applied to inter-domain 126 SFCs, these are left for future study. 128 1.2. Assumptions 130 The following assumptions are made: 132 o Not all SFs can be characterized with a standard definition in 133 terms of technical description, detailed specification, 134 configuration, etc. 136 o There is no global or standard list of SFs enabled in a given 137 administrative domain. The set of SFs varies as a function of the 138 service to be provided and according to the networking 139 environment. 141 o There is no global or standard SF chaining logic. The ordered set 142 of SFs that needs to be enabled to deliver a given service is 143 specific to each administrative entity. 145 o The chaining of SFs and the criteria to invoke them are specific 146 to each administrative entity that operates an SF-enabled domain. 148 o Several SF chaining policies can be simultaneously applied within 149 an administrative domain to meet various business requirements. 151 o No assumption is made on how FIBs and RIBs of involved nodes are 152 populated. 154 o How to bind traffic to a given SF chain is policy-based. 156 1.3. Definition of Terms 158 Network Service: An offering provided by an operator that is 159 delivered using one or more service functions. This may also be 160 referred to as a composite service. The term "service" is used 161 to denote a "network service" in the context of this document. 163 Note: Beyond this document, the term "service" is overloaded 164 with varying definitions. For example, to some a service is an 165 offering composed of several elements within the operator's 166 network, whereas for others a service, or more specifically a 167 network service, is a discrete element such as a firewall. 168 Traditionally, such services (in the latter sense) host a set of 169 service functions and have a network locator where the service 170 is hosted. 172 SFC Encapsulation: The SFC Encapsulation provides at a minimum SFP 173 identification, and is used by the SFC-aware functions, such as 174 the SFF and SFC-aware SFs. The SFC Encapsulation is not used 175 for network packet forwarding. In addition to SFP 176 identification, the SFC encapsulation carries dataplane context 177 information, also referred to as metadata. 179 Classification: Locally instantiated policy and customer/network/ 180 service profile matching of traffic flows for identification of 181 appropriate outbound forwarding actions. 183 Classifier: An element that performs Classification. 185 Service Function (SF): A function that is responsible for specific 186 treatment of received packets. A Service Function can act at 187 various layers of a protocol stack (e.g., at the network layer 188 or other OSI layers). As a logical component, a Service 189 Function can be realized as a virtual element or be embedded in 190 a physical network element. One of multiple Service Functions 191 can be embedded in the same network element. Multiple 192 occurrences of the Service Function can exist in the same 193 administrative domain. 195 One or more Service Functions can be involved in the delivery of 196 added-value services. A non-exhaustive list of Service 197 Functions includes: firewalls, WAN and application acceleration, 198 Deep Packet Inspection (DPI), LI (Lawful Intercept), server load 199 balancing, NAT44 [RFC3022], NAT64 [RFC6146], NPTv6 [RFC6296], 200 HOST_ID injection, HTTP Header Enrichment functions, TCP 201 optimizer. 203 An SF may be SFC encapsulation aware, that is it receives and 204 acts on information in the SFC encapsulation, or unaware, in 205 which case data forwarded to the SF does not contain the SFC 206 encapsulation. 208 Service Function Forwarder (SFF): A service function forwarder is 209 responsible for delivering traffic received from the network to 210 one or more connected service functions according to information 211 carried in the SFC encapsulation. 213 Service Function Chain (SFC): A service Function chain defines an 214 abstract set of service functions and ordering constraints that 215 must be applied to packets and/or frames selected as a result of 216 classification. The implied order may not be a linear 217 progression as the architecture allows for SFPs that copy to 218 more than one branch, and also allows for cases where there is 219 flexibility in the order in which services need to be applied. 220 The term service chain is often used as shorthand for service 221 function chain. 223 Service Function Path (SFP): The SFP provides a level of indirection 224 between the fully abstract notion of service chain as an 225 abstract sequence of functions to be delivered, and the fully 226 specified notion of exactly which SFF/SFs the packet will visit 227 when it actually traverses the network. By allowing the control 228 components to specify this level of indirection, the operator 229 may control the degree of SFF/SF selection authority that is 230 delegated to the network. 232 Rendered Service Path (RSP): The Service Function Path is a 233 constrained specification of where packets using a certain 234 service chain must go. While it may be so constrained as to 235 identify the exact locations, it can also be less specific. 236 Packets themselves are of course transmitted from and to 237 specific places in the network, visiting a specific sequence of 238 SFFs and SFs. This sequence of actual visits by a packet to 239 specific SFFs and SFs in the network is known as the Rendered 240 Service Path (RSP). This definition is included here for use by 241 later documents, such as when solutions may need to discuss the 242 actual sequence of locations the packets visit. 244 SFC-enabled Domain: A network or region of a network that implements 245 SFC. An SFC-enabled Domain is limited to a single network 246 administrative domain. 248 SFC Proxy: Removes and inserts SFC encapsulation on behalf of an 249 SFC-unaware service function. SFC proxies are logical elements. 251 2. Architectural Concepts 253 The following sections describe the foundational concepts of service 254 function chaining and the SFC architecture. 256 Service Function Chaining enables the creation of composite (network) 257 services that consist of an ordered set of Service Functions (SF) 258 that must be applied to packets and/or frames selected as a result of 259 classification. Each SF is referenced using an identifier that is 260 unique within an SF-enabled domain. No IANA registry is required to 261 store the identity of SFs. 263 Service Function Chaining is a concept that provides for more than 264 just the application of an ordered set of SFs to selected traffic; 265 rather, it describes a method for deploying SFs in a way that enables 266 dynamic ordering and topological independence of those SFs as well as 267 the exchange of metadata between participating entities. 269 2.1. Service Function Chains 271 In most networks services are constructed as abstract sequences of 272 SFs that represent SFCs. At a high level, an SFC is an abstracted 273 view of a service that specifies the set of required SFs as well as 274 the order in which they must be executed. Graphs, as illustrated in 275 Figure 1, define each SFC. A given SF can be part of zero, one, or 276 many SFCs. A given SF can appear one time or multiple times in a 277 given SFC. 279 SFCs can start from the origination point of the service function 280 graph (i.e.: node 1 in Figure 1), or from any subsequent node in the 281 graph. SFs may therefore become branching nodes in the graph, with 282 those SFs selecting edges that move traffic to one or more branches. 283 An SFC can have more than one terminus. 285 ,-+-. ,---. ,---. ,---. 286 / \ / \ / \ / \ 287 ( 1 )+--->( 2 )+---->( 6 )+---->( 8 ) 288 \ / \ / \ / \ / 289 `---' `---' `---' `---' 291 ,-+-. ,---. ,---. ,---. ,---. 292 / \ / \ / \ / \ / \ 293 ( 1 )+--->( 2 )+---->( 3 )+---->( 7 )+---->( 9 ) 294 \ / \ / \ / \ / \ / 295 `---' `---' `---' `---' `---' 297 ,-+-. ,---. ,---. ,---. ,---. 298 / \ / \ / \ / \ / \ 299 ( 1 )+--->( 7 )+---->( 8 )+---->( 4 )+---->( 7 ) 300 \ / \ / \ / \ / \ / 301 `---' `---' `---' `---' `---' 303 Figure 1: Service Function Chain Graphs 305 2.2. Service Function Chain Symmetry 307 SFCs may be unidirectional or bidirectional. A unidirectional SFC 308 requires that traffic be forwarded through the ordered SFs in one 309 direction (SF1 -> SF2 -> SF3), whereas a bidirectional SFC requires a 310 symmetric path (SF1 -> SF2 -> SF3 and SF3 -> SF2 -> SF1), and in 311 which the SF instances are the same in opposite directions. A hybrid 312 SFC has attributes of both unidirectional and bidirectional SFCs; 313 that is to say some SFs require symmetric traffic, whereas other SFs 314 do not process reverse traffic or are independent of the 315 corresponding forward traffic. 317 SFCs may contain cycles; that is traffic may need to traverse one or 318 more SFs within an SFC more than once. Solutions will need to ensure 319 suitable disambiguation for such situations. 321 The architectural allowance that is made for SFPs that delegate 322 choice to the network for which SFs or SFFs a packet will visit 323 creates potential issues here. A solution that allows such 324 delegation needs to also describe how the solution ensures that those 325 service chains that require service function chain symmetry can 326 achieve that. 328 Further, there are state tradeoffs in symmetry. Symmetry may be 329 realized in several ways depending on the SFF and classifier 330 functionality. In some cases, "mirrored" classification (S -> D and 331 D -> S) policy may be deployed, whereas in others shared state 332 between classifiers may be used to ensure that symmetric flows are 333 correctly identified, then steered along the required SFP. At a high 334 level, there are various common cases. In a non-exhaustive way, 335 there can be for example: a single classifier (or a small number of 336 classifiers), in which case both incoming and outgoing flows could be 337 recognized at the same classifier, so the synchronization would be 338 feasible by internal mechanisms internal to the classifier. Another 339 case is the one of stateful classifiers where several classifiers may 340 be clustered and share state. Lastly, another case entails fully 341 distributed classifiers, where synchronization needs to be provided 342 through unspecified means. This is a non-comprehensive list of 343 common cases. 345 2.3. Service Function Paths 347 A service function path (SFP) is a mechanism used by service chaining 348 to express the result of applying more granular policy and 349 operational constraints to the abstract requirements of a service 350 chain (SFC). This architecture does not mandate the degree of 351 specificity of the SFP. Architecturally, within the same SFC-enabled 352 domain, some SFPs may be fully specified, selecting exactly which SFF 353 and which SF are to be visited by packets using that SFP, while other 354 SFPs may be quite vague, deferring to the SFF the decisions about the 355 exact sequence of steps to be used to realize the SFC. The 356 specificity may be anywhere in between these extremes. 358 As an example of such an intermediate specificity, there may be two 359 SFPs associated with a given SFC, where one SFP says essentially that 360 any order of SFF and SF may be used as long as it is within data 361 center 1, and where the second SFP allows the same latitude, but only 362 within data center 2. 364 Thus, the policies and logic of SFP selection or creation (depending 365 upon the solution) produce what may be thought of as a constrained 366 version of the original SFC. Since multiple policies may apply to 367 different traffic that uses the same SFC, it also follows that there 368 may be multiple SFPs may be associated with a single SFC. 370 The architecture allows for the same SF to be reachable through 371 multiple SFFs. In these cases, some SFPs may constrain which SFF is 372 used to reach which SF, while some SFPs may leave that decision to 373 the SFF itself. 375 Further, the architecture allows for two or more SFs to be attached 376 to the same SFF, and possibly connected via internal means allowing 377 more effective communication. In these cases, some solutions or 378 deployments may choose to use some form of internal inter-process or 379 inter-VM messaging (communication behind the virtual switching 380 element) that is optimized for such an environment. This must be 381 coordinated with the SFF so that the service function forwarding can 382 properly perform its job. Implementation details of such mechanisms 383 are considered out of scope for this document, and can include a 384 spectrum of methods: for example situations including all next-hops 385 explicitly, others where a list of possible next-hops is provided and 386 the selection is local, or cases with just an identifier, where all 387 resolution is local. 389 This architecture also allows the same SF to be part of multiple 390 SFPs. 392 3. Architecture Principles 394 Service function chaining is predicated on several key architectural 395 principles: 397 1. Topological independence: no changes to the underlay network 398 forwarding topology - implicit, or explicit - are needed to 399 deploy and invoke SFs or SFCs. 401 2. Plane separation: dynamic realization of SFPs is separated from 402 packet handling operations (e.g., packet forwarding). 404 3. Classification: traffic that satisfies classification rules is 405 forwarded according to a specific SFP. For example, 406 classification can be as simple as an explicit forwarding entry 407 that forwards all traffic from one address into the SFP. 408 Multiple classification points are possible within an SFC (i.e. 409 forming a service graph) thus enabling changes/updates to the SFC 410 by SFs. 412 Classification can occur at varying degrees of granularity; for 413 example, classification can use a 5-tuple, a transport port or 414 set of ports, part of the packet payload, or it can come from 415 external systems. 417 4. Shared Metadata: Metadata/context data can be shared amongst SFs 418 and classifiers, between SFs, and between external systems and 419 SFs (e.g., orchestration). 421 Generally speaking, metadata can be thought of as providing and 422 sharing the result of classification (that occurs within the SFC- 423 enabled domain, or external to it) along an SFP. For example, an 424 external repository might provide user/subscriber information to 425 a service chain classifier. This classifier could in turn impose 426 that information in the SFC encapsulation for delivery to the 427 requisite SFs. The SFs could in turn utilize the user/subscriber 428 information for local policy decisions. 430 5. Service definition independence: The SFC architecture does not 431 depend on the details of SFs themselves. Additionally, no IANA 432 registry is required to store the list of SFs. 434 6. Service function chain independence: The creation, modification, 435 or deletion of an SFC has no impact on other SFCs. The same is 436 true for SFPs. 438 7. Heterogeneous control/policy points: The architecture allows SFs 439 to use independent mechanisms (out of scope for this document) to 440 populate and resolve local policy and (if needed) local 441 classification criteria. 443 4. Core SFC Architecture Components 445 At a very high level, the logical architecture of an SFC-enabled 446 Domain comprises: 448 o . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 449 . +--------------+ +------------------~~~ 450 . | Service | SFC | Service +---+ +---+ 451 . |Classification| Encapsulation | Function |sf1|...|sfn| 452 +---->| Function |+---------------->| Path +---+ +---+ 453 . +--------------+ +------------------~~~ 454 . SFC-enabled Domain 455 o . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 457 Figure 2: Service Function Chain Architecture 459 The following sub-sections provide details on each logical component 460 that form the basis of the SFC architecture. A detailed overview of 461 how each of these architectural components interact is provided in 462 Figure 3: 464 +----------------+ +----------------+ 465 | SFC-aware | | SFC-unaware | 466 |Service Function| |Service Function| 467 +-------+--------+ +-------+--------+ 468 | | 469 SFC Encapsulation No SFC Encapsulation 470 | SFC | 471 +----+ +----------------+ Encapsulation +---------+ 472 | SF |-----------------+ \ +------------|SFC Proxy| 473 +----+ ... ----------+ \ \ / +---------+ 474 \ \ \ / 475 +-------+--------+ 476 | SF Forwarder | 477 | (SFF) | 478 +-------+--------+ 479 | 480 SFC Encapsulation 481 | 482 ... SFC-enabled Domain ... 483 | 484 Network Overlay Transport 485 | 486 _,....._ 487 ,-' `-. 488 / `. 489 | Network | 490 `. / 491 `.__ __,-' 492 `'''' 494 Figure 3: Service Function Chain Architecture Components 496 4.1. SFC Encapsulation 498 The SFC encapsulation enables service function path selection. It 499 also enables the sharing of metadata/context information when such 500 metadata exchange is required. 502 The SFC encapsulation provides explicit information used to identify 503 the SFP. However, the SFC encapsulation is not a transport 504 encapsulation itself: it is not used to forward packets within the 505 network fabric. If packets need to flow between separate physical 506 platforms, the SFC encapsulation therefore relies on an outer network 507 transport. Transit forwarders -- such as router and switches -- 508 simply forward SFC encapsulated packets based on the outer (non-SFC) 509 encapsulation. 511 One of the key architecture principles of SFC is that the SFC 512 encapsulation remain transport independent. As such any network 513 transport protocol may be used to carry the SFC encapsulated traffic. 515 4.2. Service Function (SF) 517 The concept of an SF evolves; rather than being viewed as a bump in 518 the wire, an SF becomes a resource within a specified administrative 519 domain that is available for consumption as part of a composite 520 service. SFs send/receive data to/from one or more SFFs. SFC-aware 521 SFs receive this traffic with the SFC encapsulation. 523 While the SFC architecture defines a new encapsulation - the SFC 524 encapsulation - and several logical components for the construction 525 of SFCs, existing SF implementations may not have the capabilities to 526 act upon or fully integrate with the new SFC encapsulation. In order 527 to provide a mechanism for such SFs to participate in the 528 architecture, an SFC proxy function is defined. The SFC proxy acts 529 as a gateway between the SFC encapsulation and SFC-unaware SFs. The 530 integration of SFC-unaware service functions is discussed in more 531 detail in the SFC proxy section. 533 This architecture allows an SF to be part of multiple SFPs and SFCs. 535 4.3. Service Function Forwarder (SFF) 537 The SFF is responsible for forwarding packets and/or frames received 538 from the network to one or more SFs associated with a given SFF using 539 information conveyed in the SFC encapsulation. Traffic from SFs 540 eventually returns to the same SFF, which is responsible for putting 541 it back onto the network. 543 The collection of SFFs and associated SFs creates a service plane 544 overlay in which SFC-aware SFs, as well as SFC-unaware SFs reside. 545 Within this service plane, the SFF component connects different SFs 546 that form a service function path. 548 SFFs maintain the requisite SFP forwarding information. SFP 549 forwarding information is associated with a service path identifier 550 that is used to uniquely identify an SFP. The service forwarding 551 state enables an SFF to identify which SFs of a given SFP should be 552 applied, and in what order, as traffic flows through the associated 553 SFP. While there may appear to the SFF to be only one available way 554 to deliver the given SF, there may also be multiple choices allowed 555 by the constraints of the SFP. 557 If there are multiple choices, the SFF needs to preserve the property 558 that all packets of a given flow are handled the same way, since the 559 SF may well be stateful. 561 The SFF also has the information to allow it to forward packets to 562 the next SFF after applying local service functions. Again, while 563 there may be only a single choice available, the architecture allows 564 for multiple choices for the next SFF. As with SFs, the solution 565 needs to operate such that the behavior with regard to specific flows 566 (see the Rendered Service Path) is stable. It should be noted that 567 the selection of available SFs and next SFFs may be interwoven when 568 an SFF supports multiple distinct service functions and the same 569 service function is available at multiple SFFs. Solutions need to be 570 clear about what is allowed in these cases. 572 It should be noted that even when the SFF supports and utilizes 573 multiple choices, the decision as to whether to use flow-specific 574 mechanisms or coarser grained means to ensure that the behavior of 575 specific flows is stable is a matter for specific solutions and 576 specific implementations. 578 The SFF component has the following primary responsibilities: 580 1. SFP forwarding : Traffic arrives at an SFF from the network. The 581 SFF determines the appropriate SF the traffic should be forwarded 582 to via information contained in the SFC encapsulation. Post-SF, 583 the traffic is returned to the SFF, and if needed forwarded to 584 another SF associated with that SFF. If there is another non- 585 local (i.e., different SFF) hop in the SFP, the SFF further 586 encapsulates the traffic in the appropriate network transport 587 protocol and delivers it to the network for delivery to the next 588 SFF along the path. Related to this forwarding responsibility, 589 an SFF should be able to interact with metadata. 591 2. Terminating SFPs : An SFC is completely executed when traffic has 592 traversed all required SFs in a chain. When traffic arrives at 593 the SFF after the last SF has finished processing it, the final 594 SFF knows from the service forwarding state that the SFC is 595 complete. The SFF removes the SFC encapsulation and delivers the 596 packet back to the network for forwarding. 598 3. Maintaining flow state: In some cases, the SFF may be stateful. 599 It creates flows and stores flow-centric information. This state 600 information may be used for a range of SFP-related tasks such as 601 ensuring consistent treatment of all packets in a given flow, 602 ensuring symmetry or for state-aware SFC Proxy functionality (see 603 Section 4.8). 605 4.3.1. Transport Derived SFF 607 Service function forwarding, as described above, directly depends 608 upon the use of the service path information contained in the SFC 609 encapsulation. However, existing implementations may not be able to 610 act on the SFC encapsulation. These platforms may opt to use 611 existing transport information if it can be arranged to provide 612 explicit service path information. 614 This results in the same architectural behavior and meaning for 615 service function forwarding and service function paths. It is the 616 responsibility of the control components to ensure that the transport 617 path executed in such a case is fully aligned with the path 618 identified by the information in the service chaining encapsulation. 620 4.4. SFC-Enabled Domain 622 Specific features may need to be enforced at the boundaries of an 623 SFC-enabled domain, for example to avoid leaking SFC information. 624 Using the term node to refer generically to an entity that is 625 performing a set of functions, in this context, an SFC Boundary Node 626 denotes a node that connects one SFC-enabled domain to a node either 627 located in another SFC-enabled domain or in a domain that is SFC- 628 unaware. 630 An SFC Boundary node can act as egress or ingress. An SFC Egress 631 Node denotes a SFC Boundary Node that handles traffic leaving the 632 SFC-enabled domain the Egress Node belongs to. Such a node is 633 required to remove any information specific to the SFC Domain, 634 typically the SFC Encapsulation. An SFC Ingress Node denotes an SFC 635 Boundary Node that handles traffic entering the SFC-enabled domain. 636 In most solutions and deployments this will need to include a 637 classifier, and will be responsible for adding the SFC encapsulation 638 to the packet. 640 4.5. Network Overlay and Network Components 642 Underneath the SFF there are components responsible for performing 643 the transport (overlay) forwarding. They do not consult the SFC 644 encapsulation or inner payload for performing this forwarding. They 645 only consult the outer-transport encapsulation for the transport 646 (overlay) forwarding. 648 4.6. SFC Proxy 650 In order for the SFC architecture to support SFC-unaware SFs (.e.g 651 legacy service functions) a logical SFC proxy function may be used. 653 This function sits between an SFF and one or more SFs to which the 654 SFF is directing traffic. 656 The proxy accepts packets from the SFF on behalf of the SF. It 657 removes the SFC encapsulation, and then uses a local attachment 658 circuit to deliver packets to SFC unaware SFs. It also receives 659 packets back from the SF, reapplies the SFC encapsulation, and 660 returns them to the SFF for processing along the service function 661 path. 663 Thus, from the point of view of the SFF, the SFC proxy appears to be 664 part of an SFC aware SF. 666 Communication details between the SFF and the SFC Proxy are the same 667 as those between the SFF and an SFC aware SF. The details of that 668 are not part of this architecture. The details of the communication 669 methods over the local attachment circuit between the SFC proxy and 670 the SFC-unaware SF are dependent upon the specific behaviors and 671 capabilities of that SFC-unaware SF, and thus are also out of scope 672 for this architecture. 674 Specifically, for traffic received from the SFF intended for the SF 675 the proxy is representing, the SFC proxy: 677 o Removes the SFC encapsulation from SFC encapsulated packets. 679 o Identifies the required SF to be applied based on available 680 information including that carried in the SFC encapsulation. 682 o Selects the appropriate outbound local attachment circuit through 683 which the next SF for this SFP is reachable. This is derived from 684 the identification of the SF carried in the SFC encapsulation, and 685 may include local techniques. Examples of a local attachment 686 circuit include, but are not limited to, VLAN, IP-in-IP, L2TPv3, 687 GRE, VXLAN. 689 o Forwards the original payload via the selected local attachment 690 circuit to the appropriate SF. 692 When traffic is returned from the SF: 694 o Applies the required SFC encapsulation. The determination of the 695 encapsulation details may be inferred by the local attachment 696 circuit through which the packet and/or frame was received, or via 697 packet classification, or other local policy. In some cases, 698 packet ordering or modification by the SF may necessitate 699 additional classification in order to re-apply the correct SFC 700 encapsulation. 702 o Delivers the packet with the SFC Encapsulation to the SFF, as 703 would happen with packets returned from an SFC-aware SF. 705 Alternatively, a service provider may decide to exclude legacy 706 service functions from an SFC domain. 708 4.7. Classification 710 Traffic from the network that satisfies classification criteria is 711 directed into an SFP and forwarded to the requisite service 712 function(s). Classification is handled by a service classification 713 function; initial classification occurs at the ingress to the SFC 714 domain. The granularity of the initial classification is determined 715 by the capabilities of the classifier and the requirements of the SFC 716 policy. For instance, classification might be relatively coarse: all 717 packets from this port are subject to SFC policy X and directed into 718 SFP A, or quite granular: all packets matching this 5-tuple are 719 subject to SFC policy Y and directed into SFP B. 721 As a consequence of the classification decision, the appropriate SFC 722 encapsulation is imposed on the data, and a suitable SFP is selected 723 or created. Classification results in attaching the traffic to a 724 specific SFP. 726 4.8. Re-Classification and Branching 728 The SFC architecture supports re-classification (or non-initial 729 classification) as well. As packets traverse an SFP, re- 730 classification may occur - typically performed by a classification 731 function co-resident with a service function. Reclassification may 732 result in the selection of a new SFP, an update of the associated 733 metadata, or both. This is referred to as "branching". 735 For example, an initial classification results in the selection of 736 SFP A: DPI_1 --> SLB_8. However, when the DPI service function is 737 executed, attack traffic is detected at the application layer. DPI_1 738 re-classifies the traffic as attack and alters the service path to 739 SFP B, to include a firewall for policy enforcement: dropping the 740 traffic: DPI_1 --> FW_4. Subsequent to FW_4, surviving traffic would 741 be returned to the original SFF. In this simple example, the DPI 742 service function re-classifies the traffic based on local application 743 layer classification capabilities (that were not available during the 744 initial classification step). 746 When traffic arrives after being steered through an SFC-unaware SF, 747 the SFC Proxy must perform re-classification of traffic to determine 748 the SFP. The SFC Proxy is concerned with re-attaching information 749 for SFC-unaware SFs, and a stateful SFC Proxy simplifies such 750 classification to a flow lookup. 752 4.9. Shared Metadata 754 Sharing metadata allows the network to provide network-derived 755 information to the SFs, SF-to-SF information exchange and the sharing 756 of service-derived information to the network. Some SFCs may not 757 require metadata exchange. SFC infrastructure enables the exchange 758 of this shared data along the SFP. The shared metadata serves 759 several possible roles within the SFC architecture: 761 o Allows elements that typically operate as ships in the night to 762 exchange information. 764 o Encodes information about the network and/or data for post- 765 service forwarding. 767 o Creates an identifier used for policy binding by SFs. 769 Context information can be derived in several ways: 771 o External sources 773 o Network node classification 775 o Service function classification 777 5. Additional Architectural Concepts 779 There are a number of issues which solutions need to address, and 780 which the architecture informs but does not determine. This section 781 lays out some of those concepts. 783 5.1. The Role of Policy 785 Much of the behavior of service chains is driven by operator and per- 786 customer policy. This architecture is structured to isolate the 787 policy interactions from the data plane and control logic. 789 Specifically, it is assumed that the service chaining control plane 790 creates the service paths. The service chaining data plane is used 791 to deliver the classified packets along the service chains to the 792 intended service functions. 794 Policy, in contrast, interacts with the system in other places. 795 Policies and policy engines may monitor service functions to decide 796 if additional (or fewer) instances of services are needed. When 797 applicable, those decisions may in turn result in interactions that 798 direct the control logic to change the SFP placement or packet 799 classification rules. 801 Similarly, operator service policy, often managed by operational or 802 business support systems (OSS or BSS), will frequently determine what 803 service functions are available. Operator service policies also 804 determine which sequences of functions are valid and are to be used 805 or made available. 807 The offering of service chains to customers, and the selection of 808 which service chain a customer wishes to use, are driven by a 809 combination of operator and customer policies using appropriate 810 portals in conjunction with the OSS and BSS tools. These selections 811 then drive the service chaining control logic, which in turn 812 establishes the appropriate packet classification rules. 814 5.2. SFC Control Plane 816 This is part of the overall architecture but outside the scope of 817 this document. 819 The SFC control plane is responsible for constructing SFPs, 820 translating SFCs to forwarding paths and propagating path information 821 to participating nodes to achieve requisite forwarding behavior to 822 construct the service overlay. For instance, an SFC construction may 823 be static; selecting exactly which SFFs and which SFs from those SFFs 824 are to be used, or it may be dynamic, allowing the network to perform 825 some or all of the choices of SFF or SF to use to deliver the 826 selected service chain within the constraints represented by the 827 service path. 829 In the SFC architecture, SFs are resources; the control plane manages 830 and communicates their capabilities, availability and location in 831 fashions suitable for the transport and SFC operations in use. The 832 control plane is also responsible for the creation of the context 833 (see below). The control plane may be distributed (using new or 834 existing control plane protocols), or be centralized, or a 835 combination of the two. 837 The SFC control plane provides the following functionality: 839 1. An SFC-enabled domain wide view of all available service function 840 resources as well as the network locators through which they are 841 reachable. 843 2. Uses SFC policy to construct service function chains, and 844 associated service function paths. 846 3. Selection of specific SFs for a requested SFC, either statically 847 (using specific SFs) or dynamically (using service explicit SFs 848 at the time of delivering traffic to them). 850 4. Provides requisite SFC data plane information to the SFC 851 architecture components, most notably the SFF. 853 5. Allocation of metadata associated with a given SFP and 854 propagation of the metadata to relevant SFs and/or SFC 855 encapsulation-proxies or their respective policy planes. 857 5.3. Resource Control 859 The SFC system may be responsible for managing all resources 860 necessary for the SFC components to function. This includes network 861 constraints used to plan and choose network path(s) between service 862 function forwarders, network communication paths between service 863 function forwarders and their attached service functions, 864 characteristics of the nodes themselves such as memory, number of 865 virtual interfaces, routes, and instantiation, configuration, and 866 deletion of SFs. 868 The SFC system will also be required to reflect policy decisions 869 about resource control, as expressed by other components in the 870 system. 872 While all of these aspects are part of the overall system, they are 873 beyond the scope of this architecture. 875 5.4. Infinite Loop Detection and Avoidance 877 This SFC architecture is predicated on topological independence from 878 the underlying forwarding topology. Consequently, a service topology 879 is created by Service Function Paths or by the local decisions of the 880 Service Function Forwarders based on the constraints expressed in the 881 SFP. Due to the overlay constraints, the packet-forwarding path may 882 need to visit the same SFF multiple times, and in some less common 883 cases may even need to visit the same SF more than once. The Service 884 Chaining solution needs to permit these limited and policy-compliant 885 loops. At the same time, the solutions must ensure that indefinite 886 and unbounded loops cannot be formed, as such would consume unbounded 887 resources without delivering any value. 889 In other words, this architecture prevents infinite Service Function 890 Loops, even when Service Functions may be invoked multiple times in 891 the same SFP. 893 5.5. Load Balancing Considerations 895 Supporting function elasticity and high-availability should not 896 overly complicate SFC or lead to unnecessary scalability problems. 898 In the simplest case, where there is only a single function in the 899 SPF (the next hop is either the destination address of the flow or 900 the appropriate next hop to that destination), one could argue that 901 there may be no need for SFC. 903 In the cases where the classifier is separate from the single 904 function or a function at the terminal address may need sub-prefix or 905 per-subscriber metadata, a single SPF exists (the metadata changes 906 but the SPF does not), regardless of the number of potential terminal 907 addresses for the flow. This is the case of the simple load 908 balancer. See Figure 4. 910 +---+ +---++--->web server 911 source+-->|sff|+-->|sf1|+--->web server 912 +---+ +---++--->web server 914 Figure 4: Simple Load Balancing 916 By extrapolation, in the case where intermediary functions within a 917 chain had similar "elastic" behaviors, we do not need separate chains 918 to account for this behavior - as long as the traffic coalesces to a 919 common next-hop after the point of elasticity. 921 In Figure 5, we have a chain of five service functions between the 922 traffic source and its destination. 924 +---+ +---+ +---+ +---+ +---+ +---+ 925 |sf2| |sf2| |sf3| |sf3| |sf4| |sf4| 926 +---+ +---+ +---+ +---+ +---+ +---+ 927 | | | | | | 928 +-----+-----+ +-----+-----+ 929 | | 930 + + 931 +---+ +---+ +---+ +---+ +---+ 932 source+-->|sff|+-->|sff|+--->|sff|+--->|sff|+-->|sff|+-->destination 933 +---+ +---+ +---+ +---+ +---+ 934 + + + 935 | | | 936 +---+ +---+ +---+ 937 |sf1| |sf3| |sf5| 938 +---+ +---+ +---+ 940 Figure 5: Load Balancing 942 This would be represented as one service function path: 943 sf1->sf2->sf3->sf4->sf5. The SFF is a logical element, which may be 944 made up of one or multiple components. In this architecture, the SFF 945 may handle load distribution based on policy. 947 It can also be seen in the above that the same service function may 948 be reachable through multiple SFFs, as discussed earlier. The 949 selection of which SFF to use to reach SF3 may be made by the control 950 logic in defining the SFP, or may be left to the SFFs themselves, 951 depending upon policy, solution, and deployment constraints. In the 952 latter case, it needs to be assured that exactly one SFF takes 953 responsibility to steer traffic through SF3. 955 5.6. MTU and Fragmentation Considerations 957 This architecture prescribes additional information being added to 958 packets to identify service function paths and often to represent 959 metadata. It also envisions adding transport information to carry 960 packets along service function paths, at least between service 961 function forwarders. This added information increases the size of 962 the packet to be carried by service chaining. Such additions could 963 potentially increase the packet size beyond the MTU supported on some 964 or all of the media used in the service chaining domain. 966 Such packet size increases can thus cause operational MTU problems. 967 Requiring fragmentation and reassembly in an SFF would be a major 968 processing increase, and might be impossible with some transports. 969 Expecting service functions to deal with packets fragmented by the 970 SFC function might be onerous even when such fragmentation was 971 possible. Thus, at the very least, solutions need to pay attention 972 to the size cost of their approach. There may be alternative or 973 additional means available, although any solution needs to consider 974 the tradeoffs. 976 These considerations apply to any generic architecture that increases 977 the header size. There are also more specific MTU considerations: 978 Effects on Path MTU Discovery (PMTUD) as well as deployment 979 considerations. Deployments within a single administrateive control 980 or even a single Data Center complex can afford more flexibility in 981 dealing with larger packets, and deploying existing mitigations that 982 decrease the likelihood of fragmentation or discard. 984 5.7. SFC OAM 986 Operations, Administration, and Maintenance (OAM) tools are an 987 integral part of the architecture. These serve various purposes, 988 including fault detection and isolation, and performance management. 989 For example, there are many advantages of SFP liveness detection, 990 including status reporting, support for resiliency operations and 991 policies, and an enhanced ability to balance load. 993 Service Function Paths create a services topology, and OAM performs 994 various functions within this service layer. Furthermore, SFC OAM 995 follows the same architectural principles of SFC in general. For 996 example, topological independence (including the ability to run OAM 997 over various overlay technologies) and classification-based policy. 999 We can subdivide the SFC OAM architecture in two parts: 1001 o In-band: OAM packets follow the same path and share fate with user 1002 packets, within the service topology. For this, they also follow 1003 the architectural principle of consistent policy identifiers, and 1004 use the same path IDs as the service chain data packets. Load 1005 balancing and SFC encapsulation with packet forwarding are 1006 particularly important here. 1008 o Out-of-band: reporting beyond the actual data plane. An 1009 additional layer beyond the data-plane OAM allows for additional 1010 alerting and measurements. 1012 This architecture prescribes end-to-end SFP OAM functions, which 1013 implies SFF understanding of whether an in-band packet is an OAM or 1014 user packet. However, service function validation is outside of the 1015 scope of this architecture, and application-level OAM is not what 1016 this architecture prescribes. 1018 Some of the detailed functions performed by SFC OAM include fault 1019 detection and isolation in a Service Function Path or a Service 1020 Function, verification that connectivity using SFPs is both effective 1021 and directing packets to the intended service functions, service path 1022 tracing, diagnostic and fault isolation, alarm reporting, performance 1023 measurement, locking and testing of service functions, validation 1024 with the control plane (see Section 5.2), and also allow for vendor- 1025 specific as well as experimental functions. SFC should leverage, and 1026 if needed extend relevant existing OAM mechanisms. 1028 5.8. Resilience and Redundancy 1030 As a practical operational requirement, any service chaining solution 1031 needs to be able to respond effectively, and usually very quickly, to 1032 failure conditions. These may be failures of connectivity in the 1033 network between SFFs, failures of SFFs, or failures of SFs. Per-SF 1034 state, as for example stateful-firewall state, is the responsibility 1035 of the SF, and not addressed by this architecture. 1037 Multiple techniques are available to address this issue. Solutions 1038 can describe both what they require and what they allow to address 1039 failure. Solutions can make use of flexible specificity of service 1040 function paths, if the SFF can be given enough information in a 1041 timely fashion to do this. Solutions can also make use of MAC or IP 1042 level redundancy mechanisms such as VRRP. Also, particularly for SF 1043 failures, load balancers co-located with the SFF or as part of the 1044 service function delivery mechanism can provide such robustness. 1046 Similarly, operational requirements imply resilience in the face of 1047 load changes. While mechanisms for managing (e.g., monitoring, 1048 instantiating, loading images, providing configuration to service 1049 function chaining control, deleting, etc.) virtual machines are out 1050 of scope for this architecture, solutions can and are aided by 1051 describing how they can make use of scaling mechanisms. 1053 6. Security Considerations 1055 This document does not define a new protocol and therefore creates no 1056 new security issues. 1058 Security considerations apply to the realization of this 1059 architecture. Such realization ought to provide means to protect the 1060 SFC-enabled domain and its borders against various forms of attacks, 1061 including DDoS attacks. Further, SFC OAM Functions need to not 1062 negatively affect the security considerations of an SFC-enabled 1063 domain. Additionally, all entities (software or hardware) 1064 interacting with the service chaining mechanisms need to provide 1065 means of security against malformed, poorly configured (deliberate or 1066 not) protocol constructs and loops. These considerations are largely 1067 the same as those in any network, particularly an overlay network. 1069 7. Contributors and Acknowledgments 1071 The editors would like to thank Sam Aldrin, Linda Dunbar, Alla 1072 Goldner, Ken Gray, Shunsuke Homma, Dave Hood, Nagendra Kumar, Andrew 1073 Malis, Kengo Naito, Ron Parker, Xiaohu Xu, and Lucy Yong for a 1074 thorough review and useful comments. 1076 The initial version of this "Service Function Chaining (SFC) 1077 Architecture" document is the result of merging two previous 1078 documents, and this section lists the aggregate of authors, editors, 1079 contributors and acknowledged participants, all who provided 1080 important ideas and text that fed into this architecture. 1082 [I-D.boucadair-sfc-framework]: 1084 Authors: 1086 Mohamed Boucadair 1087 Christian Jacquenet 1088 Ron Parker 1089 Diego R. Lopez 1090 Jim Guichard 1091 Carlos Pignataro 1093 Contributors: 1095 Parviz Yegani 1096 Paul Quinn 1097 Linda Dunbar 1099 Acknowledgements: 1101 Many thanks to D. Abgrall, D. Minodier, Y. Le Goff, D. 1102 Cheng, R. White, and B. Chatras for their review and 1103 comments. 1105 [I-D.quinn-sfc-arch]: 1107 Authors: 1109 Paul Quinn (editor) 1110 Joel Halpern (editor) 1112 Contributors: 1114 Puneet Agarwal 1115 Andre Beliveau 1116 Kevin Glavin 1117 Ken Gray 1118 Jim Guichard 1119 Surendra Kumar 1120 Darrel Lewis 1121 Nic Leymann 1122 Rajeev Manur 1123 Thomas Nadeau 1124 Carlos Pignataro 1125 Michael Smith 1126 Navindra Yadav 1128 Acknowledgements: 1130 The authors would like to thank David Ward, Abhijit Patra, 1131 Nagaraj Bagepalli, Darrel Lewis, Ron Parker, Lucy Yong and 1132 Christian Jacquenet for their review and comments. 1134 8. IANA Considerations 1136 This document creates no new requirements on IANA namespaces 1137 [RFC5226]. 1139 9. References 1141 9.1. Normative References 1143 [RFC5226] Narten, T. and H. Alvestrand, "Guidelines for Writing an 1144 IANA Considerations Section in RFCs", BCP 26, RFC 5226, 1145 May 2008. 1147 9.2. Informative References 1149 [I-D.boucadair-sfc-framework] 1150 Boucadair, M., Jacquenet, C., Parker, R., Lopez, D., 1151 Guichard, J., and C. Pignataro, "Service Function 1152 Chaining: Framework & Architecture", draft-boucadair-sfc- 1153 framework-02 (work in progress), February 2014. 1155 [I-D.ietf-sfc-problem-statement] 1156 Quinn, P. and T. Nadeau, "Service Function Chaining 1157 Problem Statement", draft-ietf-sfc-problem-statement-10 1158 (work in progress), August 2014. 1160 [I-D.quinn-sfc-arch] 1161 Quinn, P. and J. Halpern, "Service Function Chaining (SFC) 1162 Architecture", draft-quinn-sfc-arch-05 (work in progress), 1163 May 2014. 1165 [RFC3022] Srisuresh, P. and K. Egevang, "Traditional IP Network 1166 Address Translator (Traditional NAT)", RFC 3022, January 1167 2001. 1169 [RFC6146] Bagnulo, M., Matthews, P., and I. van Beijnum, "Stateful 1170 NAT64: Network Address and Protocol Translation from IPv6 1171 Clients to IPv4 Servers", RFC 6146, April 2011. 1173 [RFC6296] Wasserman, M. and F. Baker, "IPv6-to-IPv6 Network Prefix 1174 Translation", RFC 6296, June 2011. 1176 Authors' Addresses 1178 Joel Halpern (editor) 1179 Ericsson 1181 Email: jmh@joelhalpern.com 1182 Carlos Pignataro (editor) 1183 Cisco Systems, Inc. 1185 Email: cpignata@cisco.com