idnits 2.17.1 draft-ietf-sfc-architecture-10.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 : ---------------------------------------------------------------------------- ** The document seems to lack a both a reference to RFC 2119 and the recommended RFC 2119 boilerplate, even if it appears to use RFC 2119 keywords. RFC 2119 keyword, line 1193: '...e with SFC, they MUST ensure that outs...' RFC 2119 keyword, line 1195: '... border security MUST include any tunn...' RFC 2119 keyword, line 1197: '... then the tunnel MUST include addition...' RFC 2119 keyword, line 1235: '... Realizations of the architecture MUST...' RFC 2119 keyword, line 1249: '... mechanism MUST verify that appr...' (1 more instance...) Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (July 24, 2015) is 3189 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- No issues found here. Summary: 1 error (**), 0 flaws (~~), 1 warning (==), 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: Informational C. Pignataro, Ed. 5 Expires: January 25, 2016 Cisco 6 July 24, 2015 8 Service Function Chaining (SFC) Architecture 9 draft-ietf-sfc-architecture-10 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, with a focus on those to be standardized in the 18 IETF. This document does not propose solutions, protocols, or 19 extensions to existing protocols. 21 Status of This Memo 23 This Internet-Draft is submitted in full conformance with the 24 provisions of BCP 78 and BCP 79. 26 Internet-Drafts are working documents of the Internet Engineering 27 Task Force (IETF). Note that other groups may also distribute 28 working documents as Internet-Drafts. The list of current Internet- 29 Drafts is at http://datatracker.ietf.org/drafts/current/. 31 Internet-Drafts are draft documents valid for a maximum of six months 32 and may be updated, replaced, or obsoleted by other documents at any 33 time. It is inappropriate to use Internet-Drafts as reference 34 material or to cite them other than as "work in progress." 36 This Internet-Draft will expire on January 25, 2016. 38 Copyright Notice 40 Copyright (c) 2015 IETF Trust and the persons identified as the 41 document authors. All rights reserved. 43 This document is subject to BCP 78 and the IETF Trust's Legal 44 Provisions Relating to IETF Documents 45 (http://trustee.ietf.org/license-info) in effect on the date of 46 publication of this document. Please review these documents 47 carefully, as they describe your rights and restrictions with respect 48 to this document. Code Components extracted from this document must 49 include Simplified BSD License text as described in Section 4.e of 50 the Trust Legal Provisions and are provided without warranty as 51 described in the Simplified BSD License. 53 Table of Contents 55 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2 56 1.1. Scope . . . . . . . . . . . . . . . . . . . . . . . . . . 3 57 1.2. Assumptions . . . . . . . . . . . . . . . . . . . . . . . 4 58 1.3. Definition of Terms . . . . . . . . . . . . . . . . . . . 4 59 2. Architectural Concepts . . . . . . . . . . . . . . . . . . . 7 60 2.1. Service Function Chains . . . . . . . . . . . . . . . . . 7 61 2.2. Service Function Chain Symmetry . . . . . . . . . . . . . 8 62 2.3. Service Function Paths . . . . . . . . . . . . . . . . . 9 63 2.3.1. Service Function Chains, Service Function Paths, and 64 Rendered Service Path . . . . . . . . . . . . . . . . 10 65 3. Architecture Principles . . . . . . . . . . . . . . . . . . . 11 66 4. Core SFC Architecture Components . . . . . . . . . . . . . . 12 67 4.1. SFC Encapsulation . . . . . . . . . . . . . . . . . . . . 13 68 4.2. Service Function (SF) . . . . . . . . . . . . . . . . . . 14 69 4.3. Service Function Forwarder (SFF) . . . . . . . . . . . . 14 70 4.3.1. Transport Derived SFF . . . . . . . . . . . . . . . . 16 71 4.4. SFC-Enabled Domain . . . . . . . . . . . . . . . . . . . 16 72 4.5. Network Overlay and Network Components . . . . . . . . . 17 73 4.6. SFC Proxy . . . . . . . . . . . . . . . . . . . . . . . . 17 74 4.7. Classification . . . . . . . . . . . . . . . . . . . . . 18 75 4.8. Re-Classification and Branching . . . . . . . . . . . . . 18 76 4.9. Shared Metadata . . . . . . . . . . . . . . . . . . . . . 19 77 5. Additional Architectural Concepts . . . . . . . . . . . . . . 19 78 5.1. The Role of Policy . . . . . . . . . . . . . . . . . . . 19 79 5.2. SFC Control Plane . . . . . . . . . . . . . . . . . . . . 20 80 5.3. Resource Control . . . . . . . . . . . . . . . . . . . . 21 81 5.4. Infinite Loop Detection and Avoidance . . . . . . . . . . 22 82 5.5. Load Balancing Considerations . . . . . . . . . . . . . . 22 83 5.6. MTU and Fragmentation Considerations . . . . . . . . . . 23 84 5.7. SFC OAM . . . . . . . . . . . . . . . . . . . . . . . . . 24 85 5.8. Resilience and Redundancy . . . . . . . . . . . . . . . . 25 86 6. Security Considerations . . . . . . . . . . . . . . . . . . . 25 87 7. Contributors and Acknowledgments . . . . . . . . . . . . . . 28 88 8. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 29 89 9. Informative References . . . . . . . . . . . . . . . . . . . 29 90 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 30 92 1. Introduction 94 The delivery of end-to-end services often requires various service 95 functions. These include traditional network service functions such 96 as firewalls and traditional IP Network Address Translators (NATs), 97 as well as application-specific functions. The definition and 98 instantiation of an ordered set of service functions and subsequent 99 'steering' of traffic through them is termed Service Function 100 Chaining (SFC). 102 This document describes an architecture used for the creation and 103 ongoing maintenance of Service Function Chains (SFC) in a network. 104 It includes architectural concepts, principles, and components, with 105 a focus on those to be standardized in the IETF. Service function 106 chains enable composite services that are constructed from one or 107 more service functions. 109 An overview of the issues associated with the deployment of end-to- 110 end service function chains, abstract sets of service functions and 111 their ordering constraints that create a composite service and the 112 subsequent "steering" of traffic flows through said service 113 functions, is described in [RFC7498]. 115 The current service function deployment models are relatively static, 116 coupled to network topology and physical resources, greatly reducing 117 or eliminating the ability of an operator to introduce new services 118 or dynamically create service function chains. This architecture 119 presents a model addressing the problematic aspects of existing 120 service deployments, including topological independence and 121 configuration complexity. 123 1.1. Scope 125 This document defines the architecture for Service Function Chaining 126 (SFC) as standardized in the IETF. The SFC architecture is 127 predicated on topological independence from the underlying forwarding 128 topology. 130 In this architecture packets are classified on ingress for handling 131 by the required set of Service Functions (SFs) in the SFC-enabled 132 domain and are then forwarded through that set of functions for 133 processing by each function in turn. Packets may be re-classified as 134 a result of this processing. 136 The architecture described in this document is independent of the 137 planned usage of the network and deployment context and thus, for 138 example, is applicable to both fixed and mobile networks as well as 139 being useful in many Data Center applications. 141 The architecture described herein is assumed to be applicable to a 142 single network administrative domain. While it is possible for the 143 architectural principles and components to be applied to inter-domain 144 SFCs, these are left for future study. 146 1.2. Assumptions 148 The following assumptions are made: 150 o There is no standard definition or characterization applicable to 151 all SFs, and thus the architecture considers each SF as an opaque 152 processing element. 154 o There is no global or standard list of SFs enabled in a given 155 administrative domain. The set of SFs enabled in a given domain 156 is a function of the currently active services which may vary with 157 time and according to the networking environment. 159 o There is no global or standard SF chaining logic. The ordered set 160 of SFs that needs to be applied to deliver a given service is 161 specific to each administrative entity. 163 o The chaining of SFs and the criteria to invoke them are specific 164 to each administrative entity that operates an SF-enabled domain. 166 o Several SF chaining policies can be simultaneously applied within 167 an administrative domain to meet various business requirements. 169 o The underlay is assumed to provide the necessary connectivity to 170 interconnect the Service Function Forwarders (SFFs, see 171 Section 1.3), but the architecture places no constraints on how 172 that connectivity is realized other than it have the required 173 bandwidth, latency, and jitter to support the SFC. 175 o No assumption is made on how Forwarding Information Bases (FIBs) 176 and Routing Information Bases (RIBs) of involved nodes are 177 populated. 179 o How to bind traffic to a given SF chain is policy-based. 181 1.3. Definition of Terms 183 Network Service: An offering provided by an operator that is 184 delivered using one or more service functions. This may also be 185 referred to as a composite service. The term "service" is used 186 to denote a "network service" in the context of this document. 188 Note: Beyond this document, the term "service" is overloaded 189 with varying definitions. For example, to some a service is an 190 offering composed of several elements within the operator's 191 network, whereas for others a service, or more specifically a 192 network service, is a discrete element such as a "firewall". 193 Traditionally, such services (in the latter sense) host a set of 194 service functions and have a network locator where the service 195 is hosted. 197 Classification: Locally instantiated matching of traffic flows 198 against policy for subsequent application of the required set of 199 network service functions. The policy may be customer/network/ 200 service specific. 202 Classifier: An element that performs Classification. 204 Service Function Chain (SFC): A service function chain defines an 205 ordered set of abstract service functions (SFs) and ordering 206 constraints that must be applied to packets and/or frames and/or 207 flows selected as a result of classification. An example of an 208 abstract service function is "a firewall". The implied order 209 may not be a linear progression as the architecture allows for 210 SFCs that copy to more than one branch, and also allows for 211 cases where there is flexibility in the order in which service 212 functions need to be applied. The term service chain is often 213 used as shorthand for service function chain. 215 Service Function (SF): A function that is responsible for specific 216 treatment of received packets. A Service Function can act at 217 various layers of a protocol stack (e.g., at the network layer 218 or other OSI layers). As a logical component, a Service 219 Function can be realized as a virtual element or be embedded in 220 a physical network element. One or more Service Functions can 221 be embedded in the same network element. Multiple occurrences 222 of the Service Function can exist in the same administrative 223 domain. 225 One or more Service Functions can be involved in the delivery of 226 added-value services. A non-exhaustive list of abstract Service 227 Functions includes: firewalls, WAN and application acceleration, 228 Deep Packet Inspection (DPI), LI (Lawful Intercept), server load 229 balancing, NAT44 [RFC3022], NAT64 [RFC6146], NPTv6 [RFC6296], 230 HOST_ID injection, HTTP Header Enrichment functions, TCP 231 optimizer. 233 An SF may be SFC encapsulation aware, that is it receives and 234 acts on information in the SFC encapsulation, or unaware, in 235 which case data forwarded to the SF does not contain the SFC 236 encapsulation. 238 Service Function Forwarder (SFF): A service function forwarder is 239 responsible for forwarding traffic to one or more connected 240 service functions according to information carried in the SFC 241 encapsulation, as well as handling traffic coming back from the 242 SF. Additionally, a service function forwarder is responsible 243 for delivering traffic to a classifier when needed and 244 supported, transporting traffic to another SFF (in the same or 245 different type of overlay), and terminating the SFP. 247 Metadata: provides the ability to exchange context information 248 between classifiers and SFs and among SFs. 250 Service Function Path (SFP): The Service Function Path is a 251 constrained specification of where packets assigned to a certain 252 service function path must go. While it may be so constrained 253 as to identify the exact locations, it can also be less 254 specific. The SFP provides a level of indirection between the 255 fully abstract notion of service chain as a sequence of abstract 256 service functions to be delivered, and the fully specified 257 notion of exactly which SFF/SFs the packet will visit when it 258 actually traverses the network. By allowing the control 259 components to specify this level of indirection, the operator 260 may control the degree of SFF/SF selection authority that is 261 delegated to the network. 263 SFC Encapsulation: The SFC Encapsulation provides at a minimum SFP 264 identification, and is used by the SFC-aware functions, such as 265 the SFF and SFC-aware SFs. The SFC Encapsulation is not used 266 for network packet forwarding. In addition to SFP 267 identification, the SFC encapsulation carries metadata including 268 data plane context information. 270 Rendered Service Path (RSP): Within an SFP, packets themselves are 271 of course transmitted from and to specific places in the 272 network, visiting a specific sequence of SFFs and SFs. This 273 sequence of actual visits by a packet to specific SFFs and SFs 274 in the network is known as the Rendered Service Path (RSP). 275 This definition is included here for use by later documents, 276 such as when solutions may need to discuss the actual sequence 277 of locations the packets visit. 279 SFC-enabled Domain: A network or region of a network that implements 280 SFC. An SFC-enabled Domain is limited to a single network 281 administrative domain. 283 SFC Proxy: Removes and inserts SFC Encapsulation on behalf of an 284 SFC-unaware service function. SFC proxies are logical elements. 286 2. Architectural Concepts 288 The following sections describe the foundational concepts of service 289 function chaining and the SFC architecture. 291 Service Function Chaining enables the creation of composite (network) 292 services that consist of an ordered set of Service Functions (SF) 293 that must be applied to packets and/or frames and/or flows selected 294 as a result of classification. Each SF is referenced using an 295 identifier that is unique within an SF-enabled domain. 297 Service Function Chaining is a concept that provides for more than 298 just the application of an ordered set of SFs to selected traffic; 299 rather, it describes a method for deploying SFs in a way that enables 300 dynamic ordering and topological independence of those SFs as well as 301 the exchange of metadata between participating entities. 303 2.1. Service Function Chains 305 In most networks, services are constructed as abstract sequences of 306 SFs that represent SFCs. At a high level, an SFC is an abstracted 307 view of a service that specifies the set of required SFs as well as 308 the order in which they must be executed. Graphs, as illustrated in 309 Figure 1, define an SFC, where each graph node represents the 310 required existence of at least one abstract SF. Such graph nodes 311 (SFs) can be part of zero, one, or many SFCs. A given graph node 312 (SF) can appear one time or multiple times in a given SFC. 314 SFCs can start from the origination point of the service function 315 graph (i.e., node 1 in Figure 1), or from any subsequent node in the 316 graph. As shown, SFs may therefore become branching nodes in the 317 graph, with those SFs selecting edges that move traffic to one or 318 more branches. The top and middle graphs depict such a case, where a 319 second classification event occurs after node 2, and a new graph is 320 selected (i.e., node 3 instead of node 6). The bottom graph 321 highlights the concept of a cycle, in which a given SF (e.g., node 7 322 in the depiction) can be visited more than once within a given 323 service chain. An SFC can have more than one terminus. 325 ,-+-. ,---. ,---. ,---. 326 / \ / \ / \ / \ 327 ( 1 )+--->( 2 )+---->( 6 )+---->( 8 ) 328 \ / \ / \ / \ / 329 `---' `---' `---' `---' 331 ,-+-. ,---. ,---. ,---. ,---. 332 / \ / \ / \ / \ / \ 333 ( 1 )+--->( 2 )+---->( 3 )+---->( 7 )+---->( 9 ) 334 \ / \ / \ / \ / \ / 335 `---' `---' `---' `---' `---' 337 ,-+-. ,---. ,---. ,---. ,---. 338 / \ / \ / \ / \ / \ 339 ( 1 )+--->( 7 )+---->( 8 )+---->( 4 )+---->( 7 ) 340 \ / \ / \ / \ / \ / 341 `---' `---' `---' `---' `---' 343 Figure 1: Service Function Chain Graphs 345 The concepts of classification, re-classification, and branching are 346 covered in subsequent sections of this architecture (see Section 4.7 347 and Section 4.8). 349 2.2. Service Function Chain Symmetry 351 SFCs may be unidirectional or bidirectional. A unidirectional SFC 352 requires that traffic be forwarded through the ordered SFs in one 353 direction (SF1 -> SF2 -> SF3), whereas a bidirectional SFC requires a 354 symmetric path (SF1 -> SF2 -> SF3 and SF3 -> SF2 -> SF1), and in 355 which the SF instances are the same in opposite directions. A hybrid 356 SFC has attributes of both unidirectional and bidirectional SFCs; 357 that is to say some SFs require symmetric traffic, whereas other SFs 358 do not process reverse traffic or are independent of the 359 corresponding forward traffic. 361 SFCs may contain cycles; that is traffic may need to traverse one or 362 more SFs within an SFC more than once. Solutions will need to ensure 363 suitable disambiguation for such situations. 365 The architectural allowance that is made for SFPs that delegate 366 choice to the network for which SFs and/or SFFs a packet will visit 367 creates potential issues here. A solution that allows such 368 delegation needs to also describe how the solution ensures that those 369 service chains that require service function chain symmetry can 370 achieve that. 372 Further, there are state tradeoffs in symmetry. Symmetry may be 373 realized in several ways depending on the SFF and classifier 374 functionality. In some cases, "mirrored" classification (i.e., from 375 Source to Destination and from Destination to Source) policy may be 376 deployed, whereas in others shared state between classifiers may be 377 used to ensure that symmetric flows are correctly identified, then 378 steered along the required SFP. At a high level, there are various 379 common cases. In a non-exhaustive way, there can be for example: 381 o A single classifier (or a small number of classifiers), in which 382 case both incoming and outgoing flows could be recognized at the 383 same classifier, so the synchronization would be feasible by 384 internal mechanisms internal to the classifier. 386 o Stateful classifiers where several classifiers may be clustered 387 and share state. 389 o Fully distributed classifiers, where synchronization needs to be 390 provided through unspecified means. 392 o A classifier that learns state from the egress packets/flows that 393 is then used to provide state for the return packets/flow. 395 o Symmetry may also be provided by stateful forwarding logic in the 396 SFF in some implementations. 398 This is a non-comprehensive list of common cases. 400 2.3. Service Function Paths 402 A service function path (SFP) is a mechanism used by service chaining 403 to express the result of applying more granular policy and 404 operational constraints to the abstract requirements of a service 405 chain (SFC). This architecture does not mandate the degree of 406 specificity of the SFP. Architecturally, within the same SFC-enabled 407 domain, some SFPs may be fully specified, selecting exactly which SFF 408 and which SF are to be visited by packets using that SFP, while other 409 SFPs may be quite vague, deferring to the SFF the decisions about the 410 exact sequence of steps to be used to realize the SFC. The 411 specificity may be anywhere in between these extremes. 413 As an example of such an intermediate specificity, there may be two 414 SFPs associated with a given SFC, where one SFP specifies that any 415 order of SFF and SF may be used as long as it is within data center 416 1, and where the second SFP allows the same latitude, but only within 417 data center 2. 419 Thus, the policies and logic of SFP selection or creation (depending 420 upon the solution) produce what may be thought of as a constrained 421 version of the original SFC. Since multiple policies may apply to 422 different traffic that uses the same SFC, it also follows that there 423 may be multiple SFPs may be associated with a single SFC. 425 The architecture allows for the same SF to be reachable through 426 multiple SFFs. In these cases, some SFPs may constrain which SFF is 427 used to reach which SF, while some SFPs may leave that decision to 428 the SFF itself. 430 Further, the architecture allows for two or more SFs to be attached 431 to the same SFF, and possibly connected via internal means allowing 432 more effective communication. In these cases, some solutions or 433 deployments may choose to use some form of internal inter-process or 434 inter-VM messaging (communication behind the virtual switching 435 element) that is optimized for such an environment. This must be 436 coordinated with the SFF so that the service function forwarding can 437 properly perform its job. Implementation details of such mechanisms 438 are considered out of scope for this document, and can include a 439 spectrum of methods: for example situations including all next-hops 440 explicitly, others where a list of possible next-hops is provided and 441 the selection is local, or cases with just an identifier, where all 442 resolution is local. 444 This architecture also allows the same SF to be part of multiple 445 SFPs. 447 2.3.1. Service Function Chains, Service Function Paths, and Rendered 448 Service Path 450 As an example of this progressive refinement, consider a service 451 function chain (SFC) which states that packets using this chain 452 should be delivered to a firewall and a caching engine. 454 A Service Function Path (SFP) could refine this, considering that 455 this architecture does not mandate the degree of specificity an SFP 456 has to have. It might specify that the firewall and caching engine 457 are both to be in a specific Data Center (e.g., in DC1), or it might 458 specify exactly which instance of each firewall and chaching engine 459 is to be used. 461 The Rendered Service Path (RSP) is the actual sequence of SFFs and 462 SFs that the packets will actually visit. So if the SFP picked the 463 DC, the RSP would be more specific. 465 3. Architecture Principles 467 Service function chaining is predicated on several key architectural 468 principles: 470 1. Topological independence: no changes to the underlay network 471 forwarding topology - implicit, or explicit - are needed to 472 deploy and invoke SFs or SFCs. 474 2. Plane separation: dynamic realization of SFPs is separated from 475 packet handling operations (e.g., packet forwarding). 477 3. Classification: traffic that satisfies classification rules is 478 forwarded according to a specific SFP. For example, 479 classification can be as simple as an explicit forwarding entry 480 that forwards all traffic from one address into the SFP. 481 Multiple classification points are possible within an SFC (i.e., 482 forming a service graph) thus enabling changes/updates to the SFC 483 by SFs. 485 Classification can occur at varying degrees of granularity; for 486 example, classification can use a 5-tuple, a transport port or 487 set of ports, part of the packet payload, it can be the result of 488 high-level inspections, or it can come from external systems. 490 4. Shared Metadata: Metadata/context data can be shared amongst SFs 491 and classifiers, between SFs, and between external systems and 492 SFs (e.g., orchestration). 494 One use of metadata is to provide and share the result of 495 classification (that occurs within the SFC-enabled domain, or 496 external to it) along an SFP. For example, an external 497 repository might provide user/subscriber information to a service 498 chain classifier. This classifier could in turn impose that 499 information in the SFC encapsulation for delivery to the 500 requisite SFs. The SFs could in turn utilize the user/subscriber 501 information for local policy decisions. Metadata can also share 502 SF output along the SFP. 504 5. Service definition independence: The SFC architecture does not 505 depend on the details of SFs themselves. 507 6. Service function chain independence: The creation, modification, 508 or deletion of an SFC has no impact on other SFCs. The same is 509 true for SFPs. 511 7. Heterogeneous control/policy points: The architecture allows SFs 512 to use independent mechanisms (out of scope for this document) to 513 populate and resolve local policy and (if needed) local 514 classification criteria. 516 4. Core SFC Architecture Components 518 The SFC Architecture is built out of architectural building blocks 519 which are logical components; these logical components are 520 classifiers, service function forwarders (SFF), the service functions 521 themselves (SF), and SFC-proxies. While this architecture describes 522 functionally distinct logical components and promotes transport 523 indepencence, they could be realized and combined in various ways in 524 deployed products, and could be combined with an overlay. 526 They are interconnected using the SFC Encapsulation. This results in 527 a high level logical architecture of an SFC-enabled Domain which 528 comprises: 530 o . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 531 . +--------------+ +------------------~~~ 532 . | Service | SFC | Service +---+ +---+ 533 . |Classification| Encapsulation | Function |sf1|...|sfn| 534 +---->| Function |+---------------->| Path +---+ +---+ 535 . +--------------+ +------------------~~~ 536 . SFC-enabled Domain 537 o . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 539 Figure 2: Service Function Chain Architecture 541 The following sub-sections provide details on each logical component 542 that form the basis of the SFC architecture. A detailed overview of 543 how some of these architectural components interact is provided in 544 Figure 3: 546 +----------------+ +----------------+ 547 | SFC-aware | | SFC-unaware | 548 |Service Function| |Service Function| 549 +-------+--------+ +-------+--------+ 550 | | 551 SFC Encapsulation No SFC Encapsulation 552 | SFC | 553 +---------+ +----------------+ Encapsulation +---------+ 554 |SFC-Aware|-----------------+ \ +------------|SFC Proxy| 555 | SF | ... ----------+ \ \ / +---------+ 556 +---------+ \ \ \ / 557 +-------+--------+ 558 | SF Forwarder | 559 | (SFF) | 560 +-------+--------+ 561 | 562 SFC Encapsulation 563 | 564 ... SFC-enabled Domain ... 565 | 566 Network Overlay Transport 567 | 568 _,....._ 569 ,-' `-. 570 / `. 571 | Network | 572 `. / 573 `.__ __,-' 574 `'''' 576 Figure 3: SFC Architecture Components Post Initial Classification 578 Please note that the depiction in Figure 3 shows packets post initial 579 classification, and therefore including the SFC encapsulation. 580 Although not included in Figure 3, the classifier is an SFC 581 architectural component. 583 4.1. SFC Encapsulation 585 The SFC encapsulation enables service function path selection. It 586 also enables the sharing of metadata/context information when such 587 metadata exchange is required. 589 The SFC encapsulation carries explicit information used to identify 590 the SFP. However, the SFC encapsulation is not a transport 591 encapsulation itself: it is not used to forward packets within the 592 network fabric. If packets need to flow between separate physical 593 platforms, the SFC encapsulation therefore relies on an outer network 594 transport. Transit forwarders -- such as router and switches -- 595 forward SFC encapsulated packets based on the outer (non-SFC) 596 encapsulation. 598 One of the key architecture principles of SFC is that the SFC 599 encapsulation remain transport independent. As such any network 600 transport protocol may be used to carry the SFC encapsulated traffic. 602 4.2. Service Function (SF) 604 The concept of an SF evolves; rather than being viewed as a bump in 605 the wire, an SF becomes a resource within a specified administrative 606 domain that is available for consumption as part of a composite 607 service. SFs send/receive data to/from one or more SFFs. SFC-aware 608 SFs receive this traffic with the SFC encapsulation. 610 While the SFC architecture defines the concept and specifies some 611 characteristics of a new encapsulation - the SFC encapsulation - and 612 several logical components for the construction of SFCs, existing SF 613 implementations may not have the capabilities to act upon or fully 614 integrate with the new SFC encapsulation. In order to provide a 615 mechanism for such SFs to participate in the architecture, an SFC 616 proxy function is defined (see Section 4.6). The SFC proxy acts as a 617 gateway between the SFC encapsulation and SFC-unaware SFs. The 618 integration of SFC-unaware service functions is discussed in more 619 detail in the SFC proxy section. 621 This architecture allows an SF to be part of multiple SFPs and SFCs. 623 4.3. Service Function Forwarder (SFF) 625 The SFF is responsible for forwarding packets and/or frames received 626 from the network to one or more SFs associated with a given SFF using 627 information conveyed in the SFC encapsulation. Traffic from SFs 628 eventually returns to the same SFF, which is responsible for 629 injecting traffic back onto the network. Some SFs, such as 630 firewalls, could also consume a packet. 632 The collection of SFFs and associated SFs creates a service plane 633 overlay in which SFC-aware SFs, as well as SFC-unaware SFs reside. 634 Within this service plane, the SFF component connects different SFs 635 that form a service function path. 637 SFFs maintain the requisite SFP forwarding information. SFP 638 forwarding information is associated with a service path identifier 639 that is used to uniquely identify an SFP. The service forwarding 640 state enables an SFF to identify which SFs of a given SFP should be 641 applied, and in what order, as traffic flows through the associated 642 SFP. While there may appear to the SFF to be only one available way 643 to deliver the given SF, there may also be multiple choices allowed 644 by the constraints of the SFP. 646 If there are multiple choices, the SFF needs to preserve the property 647 that all packets of a given flow are handled the same way, since the 648 SF may well be stateful. Additionally, the SFF may preserve the 649 handling of packets based on other properties on top of a flow, such 650 as a subscriber, session, or application instance identification. 652 The SFF also has the information to allow it to forward packets to 653 the next SFF after applying local service functions. Again, while 654 there may be only a single choice available, the architecture allows 655 for multiple choices for the next SFF. As with SFs, the solution 656 needs to operate such that the behavior with regard to specific flows 657 (see the Rendered Service Path) is stable. The selection of 658 available SFs and next SFFs may be interwoven when an SFF supports 659 multiple distinct service functions and the same service function is 660 available at multiple SFFs. Solutions need to be clear about what is 661 allowed in these cases. 663 Even when the SFF supports and utilizes multiple choices, the 664 decision as to whether to use flow-specific mechanisms or coarser 665 grained means to ensure that the behavior of specific flows is stable 666 is a matter for specific solutions and specific implementations. 668 The SFF component has the following primary responsibilities: 670 1. SFP forwarding : Traffic arrives at an SFF from the network. The 671 SFF determines the appropriate SF the traffic should be forwarded 672 to via information contained in the SFC encapsulation. Post-SF, 673 the traffic is returned to the SFF, and, if needed, is forwarded 674 to another SF associated with that SFF. If there is another non- 675 local (i.e., different SFF) hop in the SFP, the SFF further 676 encapsulates the traffic in the appropriate network transport 677 protocol and delivers it to the network for delivery to the next 678 SFF along the path. Related to this forwarding responsibility, 679 an SFF should be able to interact with metadata. 681 2. Terminating SFPs : An SFC is completely executed when traffic has 682 traversed all required SFs in a chain. When traffic arrives at 683 the SFF after the last SF has finished processing it, the final 684 SFF knows from the service forwarding state that the SFC is 685 complete. The SFF removes the SFC encapsulation and delivers the 686 packet back to the network for forwarding. 688 3. Maintaining flow state: In some cases, the SFF may be stateful. 689 It creates flows and stores flow-centric information. This state 690 information may be used for a range of SFP-related tasks such as 691 ensuring consistent treatment of all packets in a given flow, 692 ensuring symmetry or for state-aware SFC Proxy functionality (see 693 Section 4.8). 695 4.3.1. Transport Derived SFF 697 Service function forwarding, as described above, directly depends 698 upon the use of the service path information contained in the SFC 699 encapsulation. However, existing implementations may not be able to 700 act on the SFC encapsulation. These platforms may opt to use 701 existing transport information if it can be arranged to provide 702 explicit service path information. 704 This results in the same architectural behavior and meaning for 705 service function forwarding and service function paths. It is the 706 responsibility of the control components to ensure that the transport 707 path executed in such a case is fully aligned with the path 708 identified by the information in the service chaining encapsulation. 710 4.4. SFC-Enabled Domain 712 Specific features may need to be enforced at the boundaries of an 713 SFC-enabled domain, for example to avoid leaking SFC information. 714 Using the term node to refer generically to an entity that is 715 performing a set of functions, in this context, an SFC Boundary Node 716 denotes a node that connects one SFC-enabled domain to a node either 717 located in another SFC-enabled domain or in a domain that is SFC- 718 unaware. 720 An SFC Boundary node can act as egress or ingress. An SFC Egress 721 Node denotes a SFC Boundary Node that handles traffic leaving the 722 SFC-enabled domain the Egress Node belongs to. Such a node is 723 required to remove any information specific to the SFC Domain, 724 typically the SFC Encapsulation. Further, from a privacy 725 perspective, an SFC Egress Node is required to ensure that any 726 sensitive information added as part of SFC gets removed. In this 727 context, information may be sensitive due to network concerns or end- 728 customer concerns. An SFC Ingress Node denotes an SFC Boundary Node 729 that handles traffic entering the SFC-enabled domain. In most 730 solutions and deployments this will need to include a classifier, and 731 will be responsible for adding the SFC encapsulation to the packet. 733 An SFC Proxy and corresponding SFC-unaware Service Function (see 734 Figure 3) are inside the SFC-enabled domain. 736 4.5. Network Overlay and Network Components 738 Underneath the SFF there are components responsible for performing 739 the transport (overlay) forwarding. They do not consult the SFC 740 encapsulation or inner payload for performing this forwarding. They 741 only consult the outer-transport encapsulation for the transport 742 (overlay) forwarding. 744 4.6. SFC Proxy 746 In order for the SFC architecture to support SFC-unaware SFs (e.g., 747 legacy service functions) a logical SFC proxy function may be used. 748 This function sits between an SFF and one or more SFs to which the 749 SFF is directing traffic (see Figure 3). 751 The proxy accepts packets from the SFF on behalf of the SF. It 752 removes the SFC encapsulation, and then uses a local attachment 753 circuit to deliver packets to SFC unaware SFs. It also receives 754 packets back from the SF, reapplies the SFC encapsulation, and 755 returns them to the SFF for processing along the service function 756 path. 758 Thus, from the point of view of the SFF, the SFC proxy appears to be 759 part of an SFC aware SF. 761 Communication details between the SFF and the SFC Proxy are the same 762 as those between the SFF and an SFC aware SF. The details of that 763 are not part of this architecture. The details of the communication 764 methods over the local attachment circuit between the SFC proxy and 765 the SFC-unaware SF are dependent upon the specific behaviors and 766 capabilities of that SFC-unaware SF, and thus are also out of scope 767 for this architecture. 769 Specifically, for traffic received from the SFF intended for the SF 770 the proxy is representing, the SFC proxy: 772 o Removes the SFC encapsulation from SFC encapsulated packets. 774 o Identifies the required SF to be applied based on available 775 information including that carried in the SFC encapsulation. 777 o Selects the appropriate outbound local attachment circuit through 778 which the next SF for this SFP is reachable. This is derived from 779 the identification of the SF carried in the SFC encapsulation, and 780 may include local techniques. Examples of a local attachment 781 circuit include, but are not limited to, VLAN, IP-in-IP, L2TPv3, 782 GRE, VXLAN. 784 o Forwards the original payload via the selected local attachment 785 circuit to the appropriate SF. 787 When traffic is returned from the SF: 789 o Applies the required SFC encapsulation. The determination of the 790 encapsulation details may be inferred by the local attachment 791 circuit through which the packet and/or frame was received, or via 792 packet classification, or other local policy. In some cases, 793 packet ordering or modification by the SF may necessitate 794 additional classification in order to re-apply the correct SFC 795 encapsulation. 797 o Delivers the packet with the SFC Encapsulation to the SFF, as 798 would happen with packets returned from an SFC-aware SF. 800 4.7. Classification 802 Traffic from the network that satisfies classification criteria is 803 directed into an SFP and forwarded to the requisite service 804 function(s). Classification is handled by a service classification 805 function; initial classification occurs at the ingress to the SFC 806 domain. The granularity of the initial classification is determined 807 by the capabilities of the classifier and the requirements of the SFC 808 policy. For instance, classification might be relatively coarse: all 809 packets from this port are subject to SFC policy X and directed into 810 SFP A, or quite granular: all packets matching this 5-tuple are 811 subject to SFC policy Y and directed into SFP B. 813 As a consequence of the classification decision, the appropriate SFC 814 encapsulation is imposed on the data, and a suitable SFP is selected 815 or created. Classification results in attaching the traffic to a 816 specific SFP. 818 4.8. Re-Classification and Branching 820 The SFC architecture supports re-classification (or non-initial 821 classification) as well. As packets traverse an SFP, re- 822 classification may occur - typically performed by a classification 823 function co-resident with a service function. Reclassification may 824 result in the selection of a new SFP, an update of the associated 825 metadata, or both. This is referred to as "branching". 827 For example, an initial classification results in the selection of 828 SFP A: DPI_1 --> SLB_8. However, when the DPI service function is 829 executed, attack traffic is detected at the application layer. DPI_1 830 re-classifies the traffic as attack and alters the service path to 831 SFP B, to include a firewall for policy enforcement: dropping the 832 traffic: DPI_1 --> FW_4. Subsequent to FW_4, surviving traffic would 833 be returned to the original SFF. In this simple example, the DPI 834 service function re-classifies the traffic based on local application 835 layer classification capabilities (that were not available during the 836 initial classification step). 838 When traffic arrives after being steered through an SFC-unaware SF, 839 the SFC Proxy must perform re-classification of traffic to determine 840 the SFP. The SFC Proxy is concerned with re-attaching information 841 for SFC-unaware SFs, and a stateful SFC Proxy simplifies such 842 classification to a flow lookup. 844 4.9. Shared Metadata 846 Sharing metadata allows the network to provide network-derived 847 information to the SFs, SF-to-SF information exchange and the sharing 848 of service-derived information to the network. Some SFCs may not 849 require metadata exchange. SFC infrastructure enables the exchange 850 of this shared data along the SFP. The shared metadata serves 851 several possible roles within the SFC architecture: 853 o Allows elements that typically operate as ships in the night to 854 exchange information. 856 o Encodes information about the network and/or data for post- 857 service forwarding. 859 o Creates an identifier used for policy binding by SFs. 861 Context information can be derived in several ways: 863 o External sources 865 o Network node classification 867 o Service function classification 869 5. Additional Architectural Concepts 871 There are a number of issues which solutions need to address, and 872 which the architecture informs but does not determine. This section 873 lays out some of those concepts. 875 5.1. The Role of Policy 877 Much of the behavior of service chains is driven by operator and per- 878 customer policy. This architecture is structured to isolate the 879 policy interactions from the data plane and control logic. 881 Specifically, it is assumed that the service chaining control plane 882 creates the service paths. The service chaining data plane is used 883 to deliver the classified packets along the service chains to the 884 intended service functions. 886 Policy, in contrast, interacts with the system in other places. 887 Policies and policy engines may monitor service functions to decide 888 if additional (or fewer) instances of services are needed. When 889 applicable, those decisions may in turn result in interactions that 890 direct the control logic to change the SFP placement or packet 891 classification rules. 893 Similarly, operator service policy, often managed by operational or 894 business support systems (OSS or BSS), will frequently determine what 895 service functions are available. Operator service policies also 896 determine which sequences of functions are valid and are to be used 897 or made available. 899 The offering of service chains to customers, and the selection of 900 which service chain a customer wishes to use, are driven by a 901 combination of operator and customer policies using appropriate 902 portals in conjunction with the OSS and BSS tools. These selections 903 then drive the service chaining control logic, which in turn 904 establishes the appropriate packet classification rules. 906 5.2. SFC Control Plane 908 The SFC Control Plane is part of the overall SFC architecture, and 909 this section describes its high-level functions. However, the 910 detailed definition of the SFC Control Plane is outside the scope of 911 this document. 913 The SFC control plane is responsible for constructing SFPs, 914 translating SFCs to forwarding paths and propagating path information 915 to participating nodes to achieve requisite forwarding behavior to 916 construct the service overlay. For instance, an SFC construction may 917 be static; selecting exactly which SFFs and which SFs from those SFFs 918 are to be used, or it may be dynamic, allowing the network to perform 919 some or all of the choices of SFF or SF to use to deliver the 920 selected service chain within the constraints represented by the 921 service path. 923 In the SFC architecture, SFs are resources; the control plane manages 924 and communicates their capabilities, availability and location in 925 fashions suitable for the transport and SFC operations in use. The 926 control plane is also responsible for the creation of the context 927 (see below). The control plane may be distributed (using new or 928 existing control plane protocols), or be centralized, or a 929 combination of the two. 931 The SFC control plane provides the following functionality: 933 1. An SFC-enabled domain wide view of all available service function 934 resources as well as the network locators through which they are 935 reachable. 937 2. Uses SFC policy to construct service function chains, and 938 associated service function paths. 940 3. Selection of specific SFs for a requested SFC, either statically 941 (using specific SFs) or dynamically (using service explicit SFs 942 at the time of delivering traffic to them). 944 4. Provides requisite SFC data plane information to the SFC 945 architecture components, most notably the SFF. 947 5. Provide the metadata and usage information classifiers need so 948 that they in turn can provide this metadata for appropriate 949 packets in the data plane. 951 6. When needed, provide information including policy information to 952 other SFC elements to be able to properly interpret metadata. 954 5.3. Resource Control 956 The SFC system may be responsible for managing all resources 957 necessary for the SFC components to function. This includes network 958 constraints used to plan and choose network path(s) between service 959 function forwarders, network communication paths between service 960 function forwarders and their attached service functions, 961 characteristics of the nodes themselves such as memory, number of 962 virtual interfaces, routes, and instantiation, configuration, and 963 deletion of SFs. 965 The SFC system will also be required to reflect policy decisions 966 about resource control, as expressed by other components in the 967 system. 969 While all of these aspects are part of the overall system, they are 970 beyond the scope of this architecture. 972 5.4. Infinite Loop Detection and Avoidance 974 This SFC architecture is predicated on topological independence from 975 the underlying forwarding topology. Consequently, a service topology 976 is created by Service Function Paths or by the local decisions of the 977 Service Function Forwarders based on the constraints expressed in the 978 SFP. Due to the overlay constraints, the packet-forwarding path may 979 need to visit the same SFF multiple times, and in some less common 980 cases may even need to visit the same SF more than once. The Service 981 Chaining solution needs to permit these limited and policy-compliant 982 loops. At the same time, the solutions must ensure that indefinite 983 and unbounded loops cannot be formed, as such would consume unbounded 984 resources without delivering any value. 986 In other words, this architecture requires the solution to prevent 987 infinite Service Function Loops, even when Service Functions may be 988 invoked multiple times in the same SFP. 990 5.5. Load Balancing Considerations 992 Supporting function elasticity and high-availability should not 993 overly complicate SFC or lead to unnecessary scalability problems. 995 In the simplest case, where there is only a single function in the 996 SFP (the next hop is either the destination address of the flow or 997 the appropriate next hop to that destination), one could argue that 998 there may be no need for SFC. 1000 In the cases where the classifier is separate from the single 1001 function or a function at the terminal address may need sub-prefix 1002 (e.g., finer grained address information) or per-subscriber metadata, 1003 a single SFP exists (i.e., the metadata changes but the SFP does 1004 not), regardless of the number of potential terminal addresses for 1005 the flow. This is the case of the simple load balancer. See 1006 Figure 4. 1008 +---+ +---++--->web server 1009 source+-->|sff|+-->|sf1|+--->web server 1010 +---+ +---++--->web server 1012 Figure 4: Simple Load Balancing 1014 By extrapolation, in the case where intermediary functions within a 1015 chain had similar "elastic" behaviors, we do not need separate chains 1016 to account for this behavior - as long as the traffic coalesces to a 1017 common next-hop after the point of elasticity. 1019 In Figure 5, we have a chain of five service functions between the 1020 traffic source and its destination. 1022 +---+ +---+ +---+ +---+ +---+ +---+ 1023 |sf2| |sf2| |sf3| |sf3| |sf4| |sf4| 1024 +---+ +---+ +---+ +---+ +---+ +---+ 1025 | | | | | | 1026 +-----+-----+ +-----+-----+ 1027 | | 1028 + + 1029 +---+ +---+ +---+ +---+ +---+ 1030 source+-->|sff|+-->|sff|+--->|sff|+--->|sff|+-->|sff|+-->destination 1031 +---+ +---+ +---+ +---+ +---+ 1032 + + + 1033 | | | 1034 +---+ +---+ +---+ 1035 |sf1| |sf3| |sf5| 1036 +---+ +---+ +---+ 1038 Figure 5: Load Balancing 1040 This would be represented as one service function path: 1041 sf1->sf2->sf3->sf4->sf5. The SFF is a logical element, which may be 1042 made up of one or multiple components. In this architecture, the SFF 1043 may handle load distribution based on policy. 1045 It can also be seen in the above that the same service function may 1046 be reachable through multiple SFFs, as discussed earlier. The 1047 selection of which SFF to use to reach SF3 may be made by the control 1048 logic in defining the SFP, or may be left to the SFFs themselves, 1049 depending upon policy, solution, and deployment constraints. In the 1050 latter case, it needs to be assured that exactly one SFF takes 1051 responsibility to steer traffic through SF3. 1053 5.6. MTU and Fragmentation Considerations 1055 This architecture prescribes additional information being added to 1056 packets to identify service function paths and often to represent 1057 metadata. It also envisions adding transport information to carry 1058 packets along service function paths, at least between service 1059 function forwarders. This added information increases the size of 1060 the packet to be carried by service chaining. Such additions could 1061 potentially increase the packet size beyond the MTU supported on some 1062 or all of the media used in the service chaining domain. 1064 Such packet size increases can thus cause operational MTU problems. 1065 Requiring fragmentation and reassembly in an SFF would be a major 1066 processing increase, and might be impossible with some transports. 1068 Expecting service functions to deal with packets fragmented by the 1069 SFC function might be onerous even when such fragmentation was 1070 possible. Thus, at the very least, solutions need to pay attention 1071 to the size cost of their approach. There may be alternative or 1072 additional means available, although any solution needs to consider 1073 the tradeoffs. 1075 These considerations apply to any generic architecture that increases 1076 the header size. There are also more specific MTU considerations: 1077 Effects on Path MTU Discovery (PMTUD) as well as deployment 1078 considerations. Deployments within a single administrative control 1079 or even a single Data Center complex can afford more flexibility in 1080 dealing with larger packets, and deploying existing mitigations that 1081 decrease the likelihood of fragmentation or discard. 1083 5.7. SFC OAM 1085 Operations, Administration, and Maintenance (OAM) tools are an 1086 integral part of the architecture. These serve various purposes, 1087 including fault detection and isolation, and performance management. 1088 For example, there are many advantages of SFP liveness detection, 1089 including status reporting, support for resiliency operations and 1090 policies, and an enhanced ability to balance load. 1092 Service Function Paths create a services topology, and OAM performs 1093 various functions within this service layer. Furthermore, SFC OAM 1094 follows the same architectural principles of SFC in general. For 1095 example, topological independence (including the ability to run OAM 1096 over various overlay technologies) and classification-based policy. 1098 We can subdivide the SFC OAM architecture in two parts: 1100 o In-band: OAM packets follow the same path and share fate with user 1101 packets, within the service topology. For this, they also follow 1102 the architectural principle of consistent policy identifiers, and 1103 use the same path IDs as the service chain data packets. Load 1104 balancing and SFC encapsulation with packet forwarding are 1105 particularly important here. 1107 o Out-of-band: reporting beyond the actual data plane. An 1108 additional layer beyond the data-plane OAM allows for additional 1109 alerting and measurements. 1111 This architecture prescribes end-to-end SFP OAM functions, which 1112 implies SFF understanding of whether an in-band packet is an OAM or 1113 user packet. However, service function validation is outside of the 1114 scope of this architecture, and application-level OAM is not what 1115 this architecture prescribes. 1117 Some of the detailed functions performed by SFC OAM include fault 1118 detection and isolation in a Service Function Path or a Service 1119 Function, verification that connectivity using SFPs is both effective 1120 and directing packets to the intended service functions, service path 1121 tracing, diagnostic and fault isolation, alarm reporting, performance 1122 measurement, locking and testing of service functions, validation 1123 with the control plane (see Section 5.2), and also allow for vendor- 1124 specific as well as experimental functions. SFC should leverage, and 1125 if needed extend relevant existing OAM mechanisms. 1127 5.8. Resilience and Redundancy 1129 As a practical operational requirement, any service chaining solution 1130 needs to be able to respond effectively, and usually very quickly, to 1131 failure conditions. These may be failures of connectivity in the 1132 network between SFFs, failures of SFFs, or failures of SFs. Per-SF 1133 state, as for example stateful-firewall state, is the responsibility 1134 of the SF, and not addressed by this architecture. 1136 Multiple techniques are available to address this issue. Solutions 1137 can describe both what they require and what they allow to address 1138 failure. Solutions can make use of flexible specificity of service 1139 function paths, if the SFF can be given enough information in a 1140 timely fashion to do this. Solutions can also make use of MAC or IP 1141 level redundancy mechanisms such as VRRP. Also, particularly for SF 1142 failures, load balancers co-located with the SFF or as part of the 1143 service function delivery mechanism can provide such robustness. 1145 Similarly, operational requirements imply resilience in the face of 1146 load changes. While mechanisms for managing (e.g., monitoring, 1147 instantiating, loading images, providing configuration to service 1148 function chaining control, deleting, etc.) virtual machines are out 1149 of scope for this architecture, solutions can and are aided by 1150 describing how they can make use of scaling mechanisms. 1152 6. Security Considerations 1154 The architecture described here is different from the current model, 1155 and moving to the new model could lead to different security 1156 arrangements and modeling. In the SFC architecture, a relatively 1157 static topologically-dependent deployment model is replaced with the 1158 chaining of sets of service functions. This can change the flow of 1159 data through the network, and the security and privacy considerations 1160 of the protocol and deployment will need to be reevaluated in light 1161 of the new model. 1163 Security considerations apply to the realization of this 1164 architecture, in particular to the documents that will define 1165 protocols. Such realization ought to provide means to protect 1166 against security and privacy attacks in the areas hereby described. 1168 Building from the categorization of [RFC7498], we can largely divide 1169 the security considerations in four areas: 1171 Service Overlay: Underneath the Service Function Forwarders, the 1172 components that are responsible for performing the transport 1173 forwarding consult the outer-transport encapsulation for 1174 underlay forwarding. Used transport mechanisms should satisfy 1175 the security requirements of the specific SFC deployment. These 1176 requirements typically include varying degrees of traffic 1177 separation, protection against different attacks (e.g., 1178 spoofing, man-in-the-middle, brute-force, or insertion attacks), 1179 and can also include authenticity and integrity checking, and/or 1180 confidentiality provisions, for both the network overlay 1181 transport and traffic it encapsulates. 1183 Boundaries: Specific requirements may need to be enforced at the 1184 boundaries of an SFC-enabled domain. These include, for 1185 example, to avoid leaking SFC information, and to protect its 1186 borders against various forms of attacks. If untrusted parties 1187 can inject packets which will be treated as being properly 1188 classified for service chaining, there are a large range of 1189 attacks which can be mounted against the resulting system. 1190 Depending upon deployment details, these likely include spoofing 1191 packets from users and creating DDoS and reflection attacks of 1192 various kinds. Thus, when a transport mechanisms are selected 1193 for use with SFC, they MUST ensure that outside parties can not 1194 inject SFC packets which will be accepted for processing into 1195 the domain. This border security MUST include any tunnels to 1196 other domains. If those tunnels are to be used for SFC without 1197 reclassification, then the tunnel MUST include additional 1198 techniques to ensure the integrity and validity of such packets. 1200 Classification: Classification is used at the ingress edge of an 1201 SFC-enabled domain. Policy for this classification is done 1202 using a plurality of methods. Whatever method is used needs to 1203 consider a range of security issues. These include appropriate 1204 authentication and authorization of classification policy, 1205 potential confidentiality issues of that policy, protection 1206 against corruption, and proper application of policy with needed 1207 segregation of application. This includes proper controls on 1208 the policies which drive the application of the SFC 1209 Encapsulation and associated metadata to packets. Similar 1210 issues need to be addressed if classification is performed 1211 within a service chaining domain, i.e., re-classification. 1213 SFC Encapsulation: The SFC Encapsulation provides at a minimum SFP 1214 identification, and carries metadata. An operator may consider 1215 the SFC Metadata as sensitive. From a privacy perspective, a 1216 user may be concerned about the operator revealing data about 1217 (and not belonging to) the customer. Therefore, solutions 1218 should consider whether there is a risk of sensitive information 1219 slipping out of the operators control. Issues of information 1220 exposure should also consider flow analysis. Further, when a 1221 specific metadata element is defined, it should be carefully 1222 considered whether origin authentication is needed for it. 1224 A classifier may have privileged access to information about a 1225 packet or inside a packet (see Section 3, bullet 4, and 1226 Section 4.9) that is then communicated in the metadata. The 1227 threat of leaking this private data needs to be mitigated 1228 [RFC6973]. As one example, if private data is represented by an 1229 identifier, then a new identifier can be allocated, such that 1230 the mapping from the private data to the new identifier is not 1231 broadly shared. 1233 Some metadata added to and carried in SFC packets is sensitive 1234 for various reasons, including potentially revealing personally 1235 identifying information. Realizations of the architecture MUST 1236 protect to ensure that such information is handled with suitable 1237 care and precautions against inappropriate dissemination of the 1238 information. This can have implications to the data plane, the 1239 control plane, or both. Data plane protocol definitions for SFC 1240 can include suitable provision for protect such information for 1241 use when handling sensitive information, with packet or SFP 1242 granularity. Equally, the control mechanisms use with SFC can 1243 have provisions to determine that such mechanisms are available, 1244 and to ensure that they are used when needed. Inability to do 1245 so needs to result in error indications to appropriate 1246 management systems. In particular, when the control systems 1247 know that sensitive information may potentially be added to 1248 packets at certain points on certain service chains, the control 1249 mechanism MUST verify that appropriate protective treatment of 1250 NSH information is available from the point where the 1251 information is added to the point where it will be removed. If 1252 such mechanisms are unavailable, error notifications SHOULD be 1253 generated. 1255 Additionally, SFC OAM Functions need to not negatively affect the 1256 security considerations of an SFC-enabled domain. 1258 Finally, all entities (software or hardware) interacting with the 1259 service chaining mechanisms need to provide means of security against 1260 malformed, poorly configured (deliberate or not) protocol constructs 1261 and loops. These considerations are largely the same as those in any 1262 network, particularly an overlay network. 1264 7. Contributors and Acknowledgments 1266 The editors would like to thank Sam Aldrin, Alia Atlas, Nicolas 1267 Bouthors, Stewart Bryant, Linda Dunbar, Alla Goldner, Ken Gray, Barry 1268 Greene, Anil Gunturu, David Harrington, Shunsuke Homma, Dave Hood, 1269 Nagendra Kumar, Hongyu Li, Andrew Malis, Guy Meador III, Kengo Naito, 1270 Thomas Narten, Ron Parker, Reinaldo Penno, Naiming Shen, Xiaohu Xu, 1271 and Lucy Yong for a thorough review and useful comments. 1273 The initial version of this "Service Function Chaining (SFC) 1274 Architecture" document is the result of merging two previous 1275 documents, and this section lists the aggregate of authors, editors, 1276 contributors and acknowledged participants, all who provided 1277 important ideas and text that fed into this architecture. 1279 [I-D.boucadair-sfc-framework]: 1281 Authors: 1283 Mohamed Boucadair 1284 Christian Jacquenet 1285 Ron Parker 1286 Diego R. Lopez 1287 Jim Guichard 1288 Carlos Pignataro 1290 Contributors: 1292 Parviz Yegani 1293 Paul Quinn 1294 Linda Dunbar 1296 Acknowledgements: 1298 Many thanks to D. Abgrall, D. Minodier, Y. Le Goff, D. 1299 Cheng, R. White, and B. Chatras for their review and 1300 comments. 1302 [I-D.quinn-sfc-arch]: 1304 Authors: 1306 Paul Quinn (editor) 1307 Joel Halpern (editor) 1309 Contributors: 1311 Puneet Agarwal 1312 Andre Beliveau 1313 Kevin Glavin 1314 Ken Gray 1315 Jim Guichard 1316 Surendra Kumar 1317 Darrel Lewis 1318 Nic Leymann 1319 Rajeev Manur 1320 Thomas Nadeau 1321 Carlos Pignataro 1322 Michael Smith 1323 Navindra Yadav 1325 Acknowledgements: 1327 The authors would like to thank David Ward, Abhijit Patra, 1328 Nagaraj Bagepalli, Darrel Lewis, Ron Parker, Lucy Yong and 1329 Christian Jacquenet for their review and comments. 1331 8. IANA Considerations 1333 [RFC Editor: please remove this section prior to publication.] 1335 This document has no IANA actions. 1337 9. Informative References 1339 [I-D.boucadair-sfc-framework] 1340 Boucadair, M., Jacquenet, C., Parker, R., Lopez, D., 1341 Guichard, J., and C. Pignataro, "Service Function 1342 Chaining: Framework & Architecture", draft-boucadair-sfc- 1343 framework-02 (work in progress), February 2014. 1345 [I-D.quinn-sfc-arch] 1346 Quinn, P. and J. Halpern, "Service Function Chaining (SFC) 1347 Architecture", draft-quinn-sfc-arch-05 (work in progress), 1348 May 2014. 1350 [RFC3022] Srisuresh, P. and K. Egevang, "Traditional IP Network 1351 Address Translator (Traditional NAT)", RFC 3022, January 1352 2001. 1354 [RFC6146] Bagnulo, M., Matthews, P., and I. van Beijnum, "Stateful 1355 NAT64: Network Address and Protocol Translation from IPv6 1356 Clients to IPv4 Servers", RFC 6146, April 2011. 1358 [RFC6296] Wasserman, M. and F. Baker, "IPv6-to-IPv6 Network Prefix 1359 Translation", RFC 6296, June 2011. 1361 [RFC6973] Cooper, A., Tschofenig, H., Aboba, B., Peterson, J., 1362 Morris, J., Hansen, M., and R. Smith, "Privacy 1363 Considerations for Internet Protocols", RFC 6973, July 1364 2013. 1366 [RFC7498] Quinn, P. and T. Nadeau, "Problem Statement for Service 1367 Function Chaining", RFC 7498, April 2015. 1369 Authors' Addresses 1371 Joel Halpern (editor) 1372 Ericsson 1374 Email: jmh@joelhalpern.com 1376 Carlos Pignataro (editor) 1377 Cisco Systems, Inc. 1379 Email: cpignata@cisco.com