idnits 2.17.1 draft-ietf-bess-service-chaining-05.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 (August 6, 2018) is 2084 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 5575 (Obsoleted by RFC 8955) ** Downref: Normative reference to an Informational RFC: RFC 7348 ** Downref: Normative reference to an Informational RFC: RFC 7665 Summary: 3 errors (**), 0 flaws (~~), 1 warning (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network Working Group R. Fernando 3 Internet-Draft Cisco Systems 4 Intended status: Standards Track S. Mackie 5 Expires: February 7, 2019 Juniper Networks 6 D. Rao 7 Cisco Systems 8 B. Rijsman 10 M. Napierala 11 ATT Labs 12 T. Morin 13 Orange 14 August 6, 2018 16 Service Function Chaining using Virtual Networks with BGP VPNs 17 draft-ietf-bess-service-chaining-05 19 Abstract 21 This document describes how service function chains (SFC) can be 22 applied to traffic flows using routing in a virtual (overlay) network 23 to steer traffic between service nodes. Chains can include services 24 running in routers, on physical appliances or in virtual machines. 25 Service chains have applicability at the subscriber edge, business 26 edge and in multi-tenant datacenters. The routing function into SFCs 27 and between service functions within an SFC can be performed by 28 physical devices (routers), be virtualized inside hypervisors, or run 29 as part of a host OS. 31 A BGP control plane for route distribution is used to create virtual 32 networks implemented using IP MPLS, VXLAN or other suitable 33 encapsulation, where the routes within the virtual networks cause 34 traffic to flow through a sequence of service nodes that apply packet 35 processing functions to the flows. 37 Two techniques are described: in one the service chain is implemented 38 as a sequence of distinct VPNs between sets of service nodes that 39 apply each service function; in the other, the routes within a VPN 40 are modified through the use of special route targets and modified 41 next-hop resolution to achieve the desired result. 43 In both techniques, service chains can be created by manual 44 configuration of routes and route targets in routing systems, or 45 through the use of a controller which contains a topological model of 46 the desired service chains. 48 This document also contains discussion of load balancing between 49 network functions, symmetric forward and reverse paths when stateful 50 services are involved, and use of classifiers to direct traffic into 51 a service chain. 53 Status of This Memo 55 This Internet-Draft is submitted in full conformance with the 56 provisions of BCP 78 and BCP 79. 58 Internet-Drafts are working documents of the Internet Engineering 59 Task Force (IETF). Note that other groups may also distribute 60 working documents as Internet-Drafts. The list of current Internet- 61 Drafts is at https://datatracker.ietf.org/drafts/current/. 63 Internet-Drafts are draft documents valid for a maximum of six months 64 and may be updated, replaced, or obsoleted by other documents at any 65 time. It is inappropriate to use Internet-Drafts as reference 66 material or to cite them other than as "work in progress." 68 This Internet-Draft will expire on February 7, 2019. 70 Copyright Notice 72 Copyright (c) 2018 IETF Trust and the persons identified as the 73 document authors. All rights reserved. 75 This document is subject to BCP 78 and the IETF Trust's Legal 76 Provisions Relating to IETF Documents 77 (https://trustee.ietf.org/license-info) in effect on the date of 78 publication of this document. Please review these documents 79 carefully, as they describe your rights and restrictions with respect 80 to this document. Code Components extracted from this document must 81 include Simplified BSD License text as described in Section 4.e of 82 the Trust Legal Provisions and are provided without warranty as 83 described in the Simplified BSD License. 85 Table of Contents 87 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 88 1.1. Terminology . . . . . . . . . . . . . . . . . . . . . . . 5 89 2. Service Function Chain Architecture Using Virtual Networking 7 90 2.1. High Level Architecture . . . . . . . . . . . . . . . . . 8 91 2.2. Service Function Chain Logical Model . . . . . . . . . . 10 92 2.3. Service Function Implemented in a Set of SF Instances . . 11 93 2.4. SF Instance Connections to VRFs . . . . . . . . . . . . . 12 94 2.4.1. SF Instance in Physical Appliance . . . . . . . . . . 12 95 2.4.2. SF Instance in a Virtualized Environment . . . . . . 13 97 2.5. Encapsulation Tunneling for Transport . . . . . . . . . . 15 98 2.6. SFC Creation Procedure . . . . . . . . . . . . . . . . . 15 99 2.6.1. SFC Provisioning Using Sequential VPNs . . . . . . . 16 100 2.6.2. Modified-Route SFC Creation . . . . . . . . . . . . . 17 101 2.6.3. Common SFC provisioning considerations . . . . . . . 19 102 2.7. Controller Function . . . . . . . . . . . . . . . . . . . 19 103 2.8. Variations on Setting Prefixes in an SFC . . . . . . . . 20 104 2.8.1. Using a Default Route . . . . . . . . . . . . . . . . 20 105 2.8.2. Using a Default Route and a Large Prefix . . . . . . 20 106 2.8.3. Disaggregated Gateway Routers . . . . . . . . . . . . 21 107 2.8.4. Optimizing VRF usage . . . . . . . . . . . . . . . . 22 108 2.8.5. Dynamic Entry and Exit Signaling . . . . . . . . . . 22 109 2.8.6. Dynamic Re-Advertisements in Intermediate Systems . . 22 110 2.9. Layer-2 Virtual Networks and Service Functions . . . . . 23 111 2.10. Header Transforming Service Functions . . . . . . . . . . 23 112 3. Load Balancing Along a Service Function Chain . . . . . . . . 24 113 3.1. SF Instances Connected to Separate VRFs . . . . . . . . . 24 114 3.2. SF Instances Connected to the Same VRF . . . . . . . . . 25 115 3.3. Combination of Egress and Ingress VRF Load Balancing . . 26 116 3.4. Forward and Reverse Flow Load Balancing . . . . . . . . . 27 117 3.4.1. Issues with Equal Cost Multi-Path Routing . . . . . . 27 118 3.4.2. Modified ECMP with Consistent Hash . . . . . . . . . 28 119 3.4.3. ECMP with Flow Table . . . . . . . . . . . . . . . . 28 120 3.4.4. Dealing with Different Hash Algorithms in an SFC . . 30 121 4. Steering into SFCs Using a Classifier . . . . . . . . . . . . 30 122 5. External Domain Co-ordination . . . . . . . . . . . . . . . . 32 123 6. Fine-grained steering using BGP Flow-Spec . . . . . . . . . . 33 124 7. Controller Federation . . . . . . . . . . . . . . . . . . . . 33 125 8. Coordination Between SF Instances and Controller using BGP . 33 126 9. BGP Extended Communities . . . . . . . . . . . . . . . . . . 34 127 9.1. Route-Target Record . . . . . . . . . . . . . . . . . . . 34 128 9.2. Consistent Hash Sort Order . . . . . . . . . . . . . . . 35 129 9.3. Load Balance Settings . . . . . . . . . . . . . . . . . . 35 130 10. Summary and Conclusion . . . . . . . . . . . . . . . . . . . 36 131 11. Security Considerations . . . . . . . . . . . . . . . . . . . 36 132 12. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 36 133 13. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 37 134 14. References . . . . . . . . . . . . . . . . . . . . . . . . . 37 135 14.1. Normative References . . . . . . . . . . . . . . . . . . 37 136 14.2. Informational References . . . . . . . . . . . . . . . . 38 137 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 39 139 1. Introduction 141 The purpose of networks is to allow computing systems to communicate 142 with each other. Requests are usually made from the client or 143 customer side of a network, and responses are generated by 144 applications residing in a datacenter. Over time, the network 145 between the client and the application has become more complex, and 146 traffic between the client and the application is acted on by 147 intermediate systems that apply network services. Some of these 148 activities, like firewall filtering, subscriber attachment and 149 network address translation are generally carried out in network 150 devices along the traffic path, while others are carried out by 151 dedicated appliances, such as media proxy and deep packet inspection 152 (DPI). Deployment of these in-network services is complex, time- 153 consuming and costly, since they require configuration of devices 154 with vendor-specific operating systems, sometimes with co-processing 155 cards, or deployment of physical devices in the network, which 156 requires cabling and configuration of the devices that they connect 157 to. Additionally, other devices in the network need to be configured 158 to ensure that traffic is correctly steered through the systems that 159 services are running on. 161 The current mode of operations does not easily allow common 162 operational processes to be applied to the lifecycle of services in 163 the network, or for steering of traffic through them. 165 The recent emergence of Network Functions Virtualization (NFV) 166 [NFVE2E] to provide a standard deployment model for network services 167 as software appliances, combined with Software Defined Networking 168 (SDN) for more dynamic traffic steering can provide foundational 169 elements that will allow network services to be deployed and managed 170 far more efficiently and with more agility than is possible today. 172 This document describes how the combination of several existing 173 technologies can be used to create chains of functions, while 174 preserving the requirements of scale, performance and reliability for 175 service provider networks. The technologies employed are: 177 o Traffic flow between service functions described by routing and 178 network policies rather than by static physical or logical 179 connectivity 181 o Packet header encapsulation in order to create virtual private 182 networks using network overlays 184 o VRFs on both physical devices and in hypervisors to implement 185 forwarding policies that are specific to each virtual network 187 o Optional use of a controller to calculate routes to be installed 188 in routing systems to form a service chain. The controller uses a 189 topological model that stores service function instance 190 connectivity to network devices and intended connectivity between 191 service functions. 193 o MPLS or other labeling to facilitate identification of the next 194 interface to send packets to in a service function chain 196 o BGP or BGP-style signaling to distribute routes in order to create 197 service function chains 199 o Distributed load balancing between service functions performed in 200 the VRFs that service function instance connect to. 202 Virtualized environments can be supported without necessarily running 203 BGP or MPLS natively. Messaging protocols such as NC/YANG, XMPP or 204 OpenFlow may be used to signal forwarding information. Encapsulation 205 mechanisms such as VXLAN or GRE may be used for overlay transport. 206 The term 'BGP-style', above, refers to this type of signaling. 208 Traffic can be directed into service function chains using IP routing 209 at each end of the service function chain, or be directed into the 210 chain by a classifier function that can determine which service chain 211 a traffic flow should pass through based on deep packet inspection 212 (DPI) and/or subscriber identity. 214 The techniques can support an evolution from services implemented in 215 physical devices attached to physical forwarding systems (routers) to 216 fully virtualized implementations as well as intermediate hybrid 217 implementations. 219 1.1. Terminology 221 This document uses the following acronyms and terms. 223 Terms Meaning 224 ----- ----------------------------------------------- 225 AS Autonomous System 226 ASBR Autonomous System Border Router 227 RR Route Reflector 228 RT Route Target 229 SDN Software Defined Network 230 VM Virtual Machine 231 VPN Virtual Private Network 232 VRF VPN Routing and Forwarding table [RFC4364] 234 Table 1 236 This document follows some of the terminology used in [RFC7665] and 237 adds some new terminology: 239 Network Service: An externally visible service offered by a network 240 operator; a service may consist of a single service function or a 241 composite built from several service functions executed in one or 242 more pre-determined sequences and delivered by software executing 243 in physical or virtual devices. 245 Classification: Customer/network/service policy used to identify and 246 select traffic flow(s) requiring certain outbound forwarding 247 actions, in particular, to direct specific traffic flows into the 248 ingress of a particular service function chain, or causing 249 branching within a service function chain. 251 Virtual Network: A logical overlay network built using virtual links 252 or packet encapsulation, over an existing network (the underlay). 254 Service Function Chain (SFC): A service function chain defines an 255 ordered set of service functions that must be applied to packets 256 and/or frames selected as a result of classification. An SFC may 257 be either a linear chain or a complex service graph with multiple 258 branches. The term 'Service Chain' is often used in place of 259 'Service Function Chain'. 261 SFC Set: The pair of SFCs through which the forward and reverse 262 directions of a given classified flow will pass. 264 Service Function (SF): A logical function that is applied to 265 packets. A service function can act at the network layer or other 266 OSI layers. A service function can be embedded in one or more 267 physical network elements, or can be implemented in one or more 268 software instances running on physical or virtual hosts. One or 269 multiple service functions can be embedded in the same network 270 element or run on the same host. Multiple instances of a service 271 function can be enabled in the same administrative domain. We 272 will also refer to 'Service Function' as, simply, 'Service' for 273 simplicity. 275 A non-exhaustive list of services includes: firewalls, DDOS 276 protection, anti-malware/ant-virus systems, WAN and application 277 acceleration, Deep Packet Inspection (DPI), server load balancers, 278 network address translation, HTTP Header Enrichment functions, 279 video optimization, TCP optimization, etc. 281 SF Instance: An instance of software that implements the packet 282 processing of a service function 284 SF Instance Set: A group of SF instances that, in parallel, 285 implement a service function in an SFC. 287 Routing System: A hardware or software system that performs layer 3 288 routing and/or forwarding functions. The term includes physical 289 routers as well as hypervisor or Host OS implementations of the 290 forwarding plane of a conventional router. 292 Gateway: A routing system attached to the source or destination 293 network that peers with the controller, or with the routing system 294 at one end of an SFC. A source network gateway directs traffic 295 from the source network into an SFC, while a destination network 296 gateway distributes traffic towards destinations. The routing 297 systems at each end of an SFC can themselves act as gateways and 298 in a bidirectional SF instance set, gateways can act in both 299 directions VRF: A subsystem within a routing system as defined in 300 [RFC4364] that contains private routing and forwarding tables and 301 has physical and/or logical interfaces associated with it. In the 302 case of hypervisor/Host OS implementations, the term refers only 303 to the forwarding function of a VRF, and this will be referred to 304 as a 'VPN forwarder.' 306 Ingress VRF: A VRF containing an ingress interface of a SF instance 308 Egress VRF: A VRF containing an egress interface of a SF instance 310 Note that in this document the terms 'ingress' and 'egress' are 311 used with respect to SF instances rather than the tunnels that 312 connect SF instances. This is different usage than in VPN 313 literature in general. 315 Entry VRF: A VRF through which traffic enters the SFC from the 316 source network. This VRF may be used to advertise the destination 317 network's routes to the source network. It could be placed on a 318 gateway router or be collocated with the first ingress VRF. 320 Exit VRF: A VRF through which traffic exits the SFC into the 321 destination network. This VRF contains the routes from the 322 destination network and could be located on a gateway router. 323 Alternatively, the egress VRF attached to the last SF instance may 324 also function as the exit VRF. 326 2. Service Function Chain Architecture Using Virtual Networking 328 The techniques described in this document use virtual networks to 329 implement service function chains. Service function chains can be 330 implemented on devices that support existing MPLS VPN and BGP 331 standards [RFC4364], [RFC4271], [RFC4760], as well as other 332 encapsulations, such as VXLAN [RFC7348]. Similarly, equivalent 333 control plane protocols such as BGP-EVPN with type-2 and type-5 route 334 types can also be used where supported [RFC8365]. The set of 335 techniques described in this document represent one implementation 336 approach to realize the SFC architecture described in [RFC7665]. 338 The following sections detail the building blocks of the SFC 339 architecture, and outline the processes of route installation and 340 subsequent route exchange to create an SFC. 342 2.1. High Level Architecture 344 Service function chains can be deployed with or without a classifier. 345 Use cases where SFCs may be deployed without a classifier include 346 multi-tenant data centers, private and public cloud and virtual CPE 347 for business services. Classifiers will primarily be used in mobile 348 and wireline subscriber edge use cases. Use of a classifier is 349 discussed in Section 4. 351 A high-level architecture diagram of an SFC without a classifier, 352 where traffic is routed into and out of the SFC, is shown in 353 Figure 1, below. An optional controller is shown that contains a 354 topological model of the SFC and which configures the network 355 resources to implement the SFC. 357 +-------------------------+ 358 |--- Data plane connection| 359 |=== Encapsulation tunnel | 360 | O VRF | 361 +-------------------------+ 363 Control +------------------------------------------------+ 364 Plane | Controller | 365 ....... +-+------------+----------+----------+---------+-+ 366 | | | | | 367 Service | +---+ | +---+ | +---+ | | 368 Plane | |SF1| | |SF2| | |SF3| | | 369 | +---+ | +---+ | +---+ | | 370 ....... / | | / | | / | | / / 371 +-----+ +--|-|--+ +--|-|--+ +--|-|--+ +-----+ 372 | | | | | | | | | | | | | | | | 373 Net-A-->---O==========O O========O O========O O=========O---->Net-B 374 | | | | | | | | | | 375 Data | R-A | | R-1 | | R-2 | | R-3 | | R-B | 376 Plane +-----+ +-------+ +-------+ +-------+ +-----+ 378 ^ ^ ^ ^ 379 | | | | 380 | Ingress Egress | 381 | VRF VRF | 382 SFC Entry SFC Exit 383 VRF VRF 385 High Level SFC Architecture 387 Figure 1 389 Traffic from Network-A destined for Network-B will pass through the 390 SFC composed of SF instances, SF1, SF2 and SF3. Routing system R-A 391 contains a VRF (shown as 'O' symbol) that is the SFC entry point. 392 This VRF will advertise a route to reach Network-B into Network-A 393 causing any traffic from a source in Network-A with a destination in 394 Network-B to arrive in this VRF. The forwarding table in the VRF in 395 R-A will direct traffic destined for Network-B into an encapsulation 396 tunnel with destination R-1 and a label that identifies the ingress 397 (left) interface of SF1 that R-1 should send the packets out on. The 398 packets are processed by service instance SF-1 and arrive in the 399 egress (right) VRF in R-1. The forwarding entries in the egress VRF 400 direct traffic to the next ingress VRF using encapsulation tunneling. 401 The process is repeated for each service instance in the SFC until 402 packets arrive at the SFC exit VRF (in R-B). This VRF is peered with 403 Network-B and routes packets towards their destinations in the user 404 data plane. In this example, routing systems R-A and R-B are gateway 405 routing systems. 407 In the example, each pair of ingress and egress VRFs are configured 408 in separate routing systems, but such pairs could be collocated in 409 the same routing system, and it is possible for the ingress and 410 egress VRFs for a given SF instance to be in different routing 411 systems. The SFC entry and exit VRFs can be collocated in the same 412 routing system, and the service instances can be local or remote from 413 either or both of the routing systems containing the entry and exit 414 VRFs, and from each other. It is also possible that the ingress and 415 egress VRFs are implemented using alternative mechanisms. 417 The controller is responsible for configuring the VRFs in each 418 routing system, installing the routes in each of the VRFs to 419 implement the SFC, and, in the case of virtualized services, may 420 instantiate the service instances. 422 2.2. Service Function Chain Logical Model 424 A service function chain is a set of logically connected service 425 functions through which traffic can flow. Each egress interface of 426 one service function is logically connected to an ingress interface 427 of the next service function. 429 +------+ +------+ +------+ 430 Network-A-->| SF-1 |-->| SF-2 |-->| SF-3 |-->Network-B 431 +------+ +------+ +------+ 433 A Chain of Service Functions 435 Figure 2 437 In Figure 2, above, a service function chain has been created that 438 connects Network-A to Network-B, such that traffic from a host in 439 Network-A to a host in Network-B will traverse the service function 440 chain. 442 As defined in [RFC7665], a service function chain can be uni- 443 directional or bi-directional. In this document, in order to allow 444 for the possibility that the forward and reverse paths may not be 445 symmetrical, SFCs are defined as uni-directional, and the term 'SFC 446 set' is used to refer to a pair of forward and reverse direction SFCs 447 for some set of routed or classified traffic. 449 2.3. Service Function Implemented in a Set of SF Instances 451 A service function instance is a software system that acts on packets 452 that arrive on an ingress interface of that software system. Service 453 function instances may run on a physical appliance or in a virtual 454 machine. A service function instance may be transparent at layer 2 455 and/or layer 3, and may support branching across multiple egress 456 interfaces and may support aggregation across ingress interfaces. 457 For simplicity, the examples in this document have a single ingress 458 and a single egress interface. 460 Each service function in a chain can be implemented by a single 461 service function instance, or by a set of instances in order to 462 provide scale and resilience. 464 +------------------------------------------------------------------+ 465 | Logical Service Functions Connected in a Chain | 466 | | 467 | +--------+ +--------+ | 468 | Net-A--->| SF-1 |----------->| SF-2 |--->Net-B | 469 | +--------+ +--------+ | 470 | | 471 +------------------------------------------------------------------+ 472 | Service Function Instances Connected by Virtual Networks | 473 | ...... ...... | 474 | : : +------+ : : | 475 | : :-->|SFI-11|-->: : ...... | 476 | : : +------+ : : +------+ : : | 477 | : : : :-->|SFI-21|-->: : | 478 | : : +------+ : : +------+ : : | 479 | A->: VN-1 :-->|SFI-12|-->: VN-2 : : VN-3 :-->B | 480 | : : +------+ : : +------+ : : | 481 | : : : :-->|SFI-22|-->: : | 482 | : : +------+ : : +------+ : : | 483 | : :-->|SFI-13|-->: : '''''' | 484 | : : +------+ : : | 485 | '''''' '''''' | 486 +------------------------------------------------------------------+ 488 Service Functions Are Composed of SF Instances 489 Connected Via Virtual Networks 491 Figure 3 493 In Figure 3, service function SF-1 is implemented in three service 494 function instances, SFI-11, SFI-12, and SFI-13. Service function SF- 495 2 is implemented in two SF instances. The service function instances 496 are connected to the next service function in the chain using a 497 virtual network, VN-2. Additionally, a virtual network (VN-1) is 498 used to enter the SFC and another (VN-3) is used at the exit. 500 The logical connection between two service functions is implemented 501 using a virtual network that contains egress interfaces for instances 502 of one service function, and ingress interfaces of instances of the 503 next service function. Traffic is directed across the virtual 504 network between the two sets of service function instances using 505 layer 3 forwarding (e.g. an MPLS VPN) or layer 2 forwarding (e.g. a 506 VXLAN). 508 The virtual networks could be described as "directed half-mesh", in 509 that the egress interface of each SF instance of one service function 510 can reach any ingress interface of the SF instances of the connected 511 service function. 513 Details on how routing across virtual networks is achieved, and 514 requirements on load balancing across ingress interfaces are 515 discussed in later sections of this document. 517 2.4. SF Instance Connections to VRFs 519 SF instances can be deployed as software running on physical 520 appliances, or in virtual machines running on a hypervisor. These 521 two types are described in more detail in the following sections. 523 2.4.1. SF Instance in Physical Appliance 525 The case of a SF instance running on a physical appliance is shown in 526 Figure 4, below. 528 +---------------------------------+ 529 | | 530 | +-----------------------------+ | 531 | | Service Function Instance | | 532 | +-------^-------------|-------+ | 533 | | Host | | 534 +---------|-------------|---------+ 535 | | 536 +------ |-------------|-------+ 537 | | | | 538 | +----|----+ +-----v----+ | 539 ---------+ Ingress | | Egress +--------- 540 ---------> VRF | | VRF ----------> 541 ---------+ | | +--------- 542 | +---------+ +----------+ | 543 | Routing System | 544 +-----------------------------+ 546 Ingress and Egress VRFs for a Physical Routing System 547 and Physical SF Instance 549 Figure 4 551 The routing system is a physical device and the service function 552 instance is implemented as software running in a physical appliance 553 (host) connected to it. The connection between the physical device 554 and the routing system may use physical or logical interfaces. 555 Transport between VRFs on different routing systems that are 556 connected to other SF instances in an SFC is via encapsulation 557 tunnels, such as MPLS over GRE, or VXLAN. 559 2.4.2. SF Instance in a Virtualized Environment 561 In virtualized environments, a routing system with VRFs that act as 562 VPN forwarders is resident in the hypervisor/Host OS, and is co- 563 resident in the host with one or more SF instances that run in 564 virtual machines. The egress VPN forwarder performs tunnel 565 encapsulation to send packets to other physical or virtual routing 566 systems with attached SF instances to form an SFC. The tunneled 567 packets are sent through the physical interfaces of the host to the 568 other hosts or physical routers. This is illustrated in Figure 5, 569 below. 571 +-------------------------------------+ 572 | +-----------------------------+ | 573 | | Service Function Instance | | 574 | +-------^-------------|-------+ | 575 | | | | 576 | +---------|-------------|---------+ | 577 | | +-------|-------------|-------+ | | 578 | | | | | | | | 579 | | | +----|----+ +-----v----+ | | | 580 ------------+ Ingress | | Egress +----------- 581 ------------> VRF | | VRF ------------> 582 ------------+ | | +----------- 583 | | | +---------+ +----------+ | | | 584 | | | Routing System | | | 585 | | +-----------------------------+ | | 586 | | Hypervisor or Host OS | | 587 | +---------------------------------+ | 588 | Host | 589 +-------------------------------------+ 591 Ingress and Egress VRFs for a Virtual Routing System 592 and Virtualized SF Instance 594 Figure 5 596 When more than one instance of an SF is running on a hypervisor, they 597 can be connected to the same VRF for scale out of an SF within an 598 SFC. 600 The routing mechanisms in the VRFs into and between service function 601 instances, and the encapsulation tunneling between routing systems 602 are identical in the physical and virtual implementations of SFCs and 603 routing systems described in this document. Physical and virtual 604 service functions can be mixed as needed with different combinations 605 of physical and virtual routing systems, within a single service 606 chain. 608 The SF instances are attached to the routing systems via physical, 609 virtual or logical (e.g, 802.1q) interfaces, and are assumed to 610 perform basic L3 or L2 forwarding. 612 A single SF instance can be part of multiple service chains. In this 613 case, the SF instance will have dedicated interfaces (typically 614 logical) and forwarding contexts associated with each service chain. 616 2.5. Encapsulation Tunneling for Transport 618 Encapsulation tunneling is used to transport packets between SF 619 instances in the chain and, when a classifier is not used, from the 620 originating network into the SFC and from the SFC into the 621 destination network. 623 The tunnels can be MPLS over GRE [RFC4023], MPLS over UDP [RFC7510], 624 MPLS over MPLS [RFC3031], VXLAN [RFC7348][RFC7348], or another 625 suitable encapsulation method. 627 Tunneling capabilities may be enabled in each routing system as part 628 of a base configuration or may be configured by the controller. 629 Tunnel encapsulations may be programmed by the controller or signaled 630 using BGP. The encapsulation to be used for a given route is 631 signaled in BGP using the procedures described in 632 [idr-tunnel-encaps], i.e. typically relying on the BGP Tunnel 633 Encapsulation Extended Community. 635 2.6. SFC Creation Procedure 637 This section describes how service chains are created using two 638 methods: 640 o Sequential VPNs - where a conventional VPN is created between each 641 set of SF instances to create the links in the SFC 643 o Route Modification - where each routing system modifies advertised 644 routes that it receives, to realize the links in an SFC on the 645 basis of a special service topology RT and a route- policy that 646 describes the service chain logical topology 648 In both cases the controller, when present, is responsible for 649 creating ingress and egress VRFs, configuring the interfaces 650 connected to SF instances in each VRF, and allocating and configuring 651 import and export RTs for each VRF. Additionally, in the second 652 method, the controller also sends the route-policy containing the 653 service chain logical topology to each routing system. If a 654 controller is not used, these procedures will require to be performed 655 manually or through scripting, for instance. 657 The source and destination networks' prefixes can be configured in 658 the controller, or may be automatically learned through peering 659 between the controller and each network's gateway. This is further 660 described in Section 2.8.5 and Section 5. 662 The following sub-sections describe how RT configuration, local route 663 installation and route distribution occur in each of the methods. 665 It should be noted that depending on the capabilities of the routing 666 systems, a controller can use one or more techniques to realize 667 forwarding along the service chain, ranging from fully centralized to 668 fully distributed. The goal of describing the following two methods 669 is to illustrate the broad approaches and as a base for various 670 optimization options. 672 Interoperability between a controller implementing one method and a 673 controller implementing a different method is achieved by relying on 674 the techniques described in section 5 and section 8, that describe 675 the use of BGP-style service chaining within domains that are 676 interconnected using standard BGP VPN route exchanges. 678 2.6.1. SFC Provisioning Using Sequential VPNs 680 The task of the controller in this method of SFC provisioning is to 681 create a set of VPNs that carry traffic to the destination network 682 through instances of each service function in turn. This is achieved 683 by allocating and configuring RTs such that the egress VRFs of one 684 set of SF instances import an RT that is an export RT for the ingress 685 VRFs of the next, logically connected, set of SF instances. 687 The process of SFC creation is as follows: 689 1. Controller creates a VRF in each routing system that is connected 690 to a service instance that will be used in the SFC 692 2. Controller configures each VRF to contain the logical interface 693 that connects to a SF instance. 695 3. Controller implements route target import and export policies in 696 the VRFs using the same route targets for the egress VRFs of a 697 service function and the ingress VRFs of the next logically 698 connected service function in the SFC. 700 4. Controller installs a static route in each ingress VRF whose next 701 hop is the interface that a SF instance is connected to. The 702 prefix for the route is the destination network to be reached by 703 passing through the SFC. The following sections describe 704 variations that can be used. 706 5. Routing systems advertise the static routes via BGP as VPN routes 707 with next hop being the IP address of the router, with an 708 encapsulation specified and a label that identifies the service 709 instance interface. 711 6. Routing systems containing VRFs with matching route targets 712 receive the updates. 714 7. Routes are installed in egress VRFs with matching import targets. 715 The egress VRFs of each SF instance will now contain VPN routes 716 to one or more routers containing ingress VRFs for SF instances 717 of the next service function in the SFC. 719 Routes to the destination network via the first set of SF instances 720 are advertised into the source network, and the egress VRFs of the 721 last SF instance set have routes into the destination network. 723 As discussed further in Section 3, egress VRFs can load balance 724 across the multiple next hops advertised from the next set of ingress 725 VRFs. 727 2.6.2. Modified-Route SFC Creation 729 In this method of SFC configuration, all the VRFs connected to SF 730 instances for a given SFC are configured with same import and export 731 RT, so they form a VPN-connected mesh between the SF instance 732 interfaces. This is termed the 'Service VPN'. A route is configured 733 or learnt in each VRF with destination being the IP address of a 734 connected SF instance via an interface configured in the VRF. The 735 interface may be a physical or logical interface. The routing system 736 that hosts such a VRF advertises a VPN route for each locally 737 connected SF instance, with a forwarding label that enables it to 738 forward incoming traffic from other routing systems to the connected 739 SF instance. The VPN routes may be advertised via an RR or the 740 controller, which sends these updates to all the other routing 741 systems that have VRFs with the service VPN RT. At this point all 742 the VRFs have a route to reach every SF instance. The same virtual 743 IP address may be used for each SF instance in a set, enabling load- 744 balancing among multiple SF instances in the set. 746 The controller builds a route-policy for the routing systems in the 747 VPN, that describes the logical topology of each service chain that 748 it belongs to. The route-policy contains entries in the form of a 749 tuple for each service chain: 751 {Service-topology-name, Service-topology-RT, Service-node- sequence} 753 where Service-node-sequence is simply an ordered list of the service 754 function interface IP addresses that are in the chain. 756 Every service function chain has a single unique service-topology-RT 757 that is allocated and provisioned on all participating routing 758 systems in the relevant VRFs. 760 The VRF in the routing system that connects to the destination 761 network (i.e. the exit VRF) is configured to attach the Service- 762 topology-RT to exported routes, and the VRF connected to the source 763 network (i.e. the entry VRF) will import routes using the Service- 764 topology-RT. The controller may also be used to originate the 765 Service-topology-RT attached routes. 767 The route-policy may be described in a variety of formats and 768 installed on the routing system using a suitable mechanism. For 769 instance, the policy may be defined in YANG and provisioned using 770 Netconf [RFC6241]. 772 Using Figure 1 for reference, when the gateway R-B advertises a VPN 773 route to Network-B, it attaches the Service-topology-RT. BGP route 774 updates are sent to all the routing systems in the service VPN. The 775 routing systems perform a modified set of actions for next-hop 776 resolution and route installation in the ingress VRFs compared to 777 normal BGP VPN behavior in routing systems, but no changes are 778 required in the operation of the BGP protocol itself. The 779 modification of behavior in the routing systems allows the automatic 780 and constrained flow of traffic through the service chain. 782 Each routing system in the service VPN will process the VPN route to 783 Network-B via R-B as follows: 785 1. If the routing system contains VRFs that import the Service- 786 topology-RT, continue, otherwise ignore the route. 788 2. The routing system identifies the position and role (ingress/ 789 egress) of each of its VRFs in the SFC by comparing the IP 790 address of the route in the VRF to the connected SF instance with 791 those in the Service-node- sequence in the route-policy. 792 Alternatively, the controller may provision the specific service 793 node IP to be used as the next-hop in each VRF, in the route- 794 policy for the VRF. 796 3. The routing system modifies the next-hop of the imported route 797 with the Service-topology-RT, to select the appropriate next-hop 798 as per the route-policy. It ignores the next-hop and label in 799 the received route. It resolves the selected next-hop in the 800 local VRF routing table. 802 4. 804 a. The imported route to Network-B in the ingress VRF is 805 modified to have a next-hop of the IP address of the 806 logically connected SF instance. 808 b. The imported route to Network-B in the egress VRF is modified 809 to have a next hop of the IP address of the next SF instance 810 in the SFC. 812 5. The egress VRFs for the last service function install the VPN 813 route via the gateway R-B unmodified. 815 Note that the modified routes are not re-advertised into the VPN by 816 the various intermediate routing systems in the SFC. 818 2.6.3. Common SFC provisioning considerations 820 In both the methods, for physical routers, the creation and 821 configuration of VRFs, interfaces and local static routes can be 822 performed programmatically using Netconf; and BGP route distribution 823 can use a route reflector (which may be part of the controller). In 824 the virtualized case, where a VPN forwarder is present, creation and 825 configuration of VRFs, interfaces and installation of routes may 826 instead be performed using a single protocol like XMPP, NC/YANG or an 827 equivalent programmatic interface. 829 Also in the virtualized case, the actual forwarding table entries to 830 be installed in the ingress and egress VRFs may be calculated by the 831 controller based on its internal knowledge of the required SFC 832 topology and the connectivity of SF instances to routing systems. In 833 this case, the routes may be directly installed in the forwarders 834 using the programmatic interface and no BGP route advertisement is 835 necessary, except when coordination with external domains (Section 5) 836 or federation between controller domains is employed (Section 7). 837 Note however that this is just one typical model for a virtual 838 forwarding based system. In general, physical and virtual routing 839 systems can be treated exactly the same if they have the same 840 capabilities. 842 In both the methods, the SF instance may also need to be set up 843 appropriately to forward traffic between it's input and output 844 interfaces, either via static, dynamic or policy-based routing. If 845 the service function is a transparent L2 service, then the static 846 route installed in the ingress VRF will have a next-hop of the IP 847 address of the routing system interface that the service instance is 848 attached to on its other interface. 850 2.7. Controller Function 852 The purpose of the controller is to manage instantiation of SFCs in 853 networks and datacenters. When an SFC is to be instantiated, a model 854 of the desired topology (service functions, number of instances, 855 connectivity) is built in the controller either via an API or GUI. 857 The controller then selects resources in the infrastructure that will 858 support the SFC and configures them. This can involve instantiation 859 of SF instances to implement each service function, the instantiation 860 of VRFs that will form virtual networks between SF instances, and 861 installation of routes to cause traffic to flow into and between SF 862 instances. It can also include provisioning the necessary static, 863 dynamic or policy based forwarding on the service function instance 864 to enable it to forward traffic. 866 For simplicity, in this document, the controller is assumed to 867 contain all the required features for management of SFCs. In actual 868 implementations, these features may be distributed among multiple 869 inter-connected systems. E.g. An overarching orchestrator might 870 manage the overall SFC model, sending instructions to a separate 871 virtual machine manager to instantiate service function instances, 872 and to a virtual network manager to set up the service chain 873 connections between them. 875 The controller can also perform necessary BGP signaling and route 876 distribution actions as described throughout this document. 878 2.8. Variations on Setting Prefixes in an SFC 880 The SFC Creation section above described the basic procedures for a 881 couple of SFC creation methods. This section describes some 882 techniques that can extend and provide optimizations on top of the 883 basic procedures. 885 2.8.1. Using a Default Route 887 In the methods described above, it can be noted that only the gateway 888 routing systems need the specific network prefixes to steer traffic 889 in and out of the SFC. The intermediate systems can direct traffic 890 in the ingress and egress VRFs by using only a default route. Hence, 891 it is possible to avoid installing the network prefixes in the 892 intermediate systems. This can be done by splitting the SFC into two 893 sections - one linking the entry and exit VRFs and the other 894 including the intermediate systems. For instance, this may be 895 achieved by using two different Service-topology-RTs in the second 896 method. 898 2.8.2. Using a Default Route and a Large Prefix 900 In the configuration methods described above, the network prefixes 901 for each network (Network-A and Network-B in the example above) 902 connected to the SFC are used in the routes that direct traffic 903 through the SFC. This creates an operational linkage between the 904 implementation of the SFC and the insertion of the SFC into a 905 network. 907 For instance, subscriber network prefixes will normally be segmented 908 across subscriber attachment points such as broadband or mobile 909 gateways. This means that each SFC would have to be configured with 910 the subscriber network prefixes whose traffic it is handling. 912 In a variation of the SFC configuration method described above, the 913 prefixes used in each direction can be such that they include all 914 possible addresses at each side of the SFC. For example, in 915 Figure 1, the prefix for Network-A could include all subscriber IP 916 addresses and the prefix for Network-B could be the default route, 917 0/0. 919 Using this technique, the same routes can be installed in all 920 instances of an SFC that serve different groups of subscribers in 921 different geographic locations. 923 The routes forwarding traffic into a SF instance and to the next SF 924 instance are installed when an SFC is initially built, and each time 925 a SF instance is connected into the SFC, but there is no requirement 926 for VRFs to be reconfigured when traffic from different networks pass 927 through the service chain, so long as their prefix is included in the 928 prefixes in the VRFs along the SFC. 930 In this variation, it is assumed that no subscriber-originated 931 traffic will enter the SFC destined for an IP address also in the 932 subscriber network address range. This will not be a restriction in 933 many cases. 935 2.8.3. Disaggregated Gateway Routers 937 As a slight variation of the above, a network prefix may be 938 disaggregated and spread out among various gateway routers, for 939 instance, in the case of virtual machines in a data-center. In order 940 to reduce the scaling requirements on the routing systems along the 941 SFC, the SFC can again be split into two sections as described above. 942 In addition, the last egress VRF may act as the exit VRF and install 943 the destination network's disaggregated routes. If the destination 944 network's prefixes can be aggregated, for instance into a subnet 945 prefix, then the aggregate prefix may be advertised and installed in 946 the entry VRF. 948 2.8.4. Optimizing VRF usage 950 It may be desirable to avoid using distinct ingress and egress VRFs 951 for the service instances in order to make more efficient use of VRF 952 resources, especially on physical routing systems. The ingress VRF 953 and egress VRF may be treated as conceptual entities and the 954 forwarding realized using one or more options described in this 955 section, combined with the methods described earlier. 957 For instance, the next-hop forwarding label described earlier serves 958 the purpose of directing traffic received from other routing systems 959 directly towards an attached service instance. On the other hand, if 960 the encapsulation mechanism or the device in use requires an IP 961 lookup for incoming packets from other routing systems, then the 962 specific network prefixes may be installed in the intermediate 963 service VRFs to direct traffic towards the attached service 964 instances. 966 Similarly, a per-interface policy-based-routing rule applied to an 967 access interface can serve to direct traffic coming in from attached 968 service instances towards the next SF set. 970 2.8.5. Dynamic Entry and Exit Signaling 972 When either of the methods of the previous sections are employed, the 973 prefixes of the attached networks at each end of an SFC can be 974 signaled into the corresponding VRFs dynamically. This requires that 975 a BGP session is configured either from the network device at each 976 end of the SFC into each network or from the controller. 978 If dynamic signaling is performed, and a bidirectional SFC set is 979 configured, and the gateways to the networks connected via the SFC 980 exchange routes, steps must be taken to ensure that routes to both 981 networks do not get advertised from both ends of the SFC set by re- 982 origination. This can be achieved if a new BGP Extended Community 983 [RFC4360] is implemented to control re-origination. When a route is 984 re- originated, the RTs of the re-originated routes are appended to 985 the new RT-Record Extended Community, and if the RT for the route 986 already exists in the Extended Community, the route is not re- 987 originated (see Section 9.1). 989 2.8.6. Dynamic Re-Advertisements in Intermediate Systems 991 The intermediate routing systems attached to the service instances 992 may also use the dynamic signaling technique from the previous 993 section to re-advertise received routes up the chain. In this case, 994 the ingress and egress VRFs are combined into one; and a local route- 995 policy ensures the re-advertised routes are associated with labels 996 that direct incoming traffic directly to the attached service 997 instances on that routing system. 999 2.9. Layer-2 Virtual Networks and Service Functions 1001 There are SFs that operate at layer-2, in a transparent mode, and 1002 forward traffic based on the MAC DA. When such a SF is present in 1003 the SFC, the procedures at the routing system are modified slightly. 1004 In this case, the IP address associated with the SF instance (and 1005 used as the next-hop of routes in the above procedures) is actually 1006 the one assigned to the routing system interface attached to the 1007 other end of the SF instance, or it could be a virtual IP address 1008 logically associated with the service function with a next-hop of the 1009 other routing system interface. The routing system interface uses 1010 distinct interface MAC addresses. This allows the current scheme to 1011 be supported, while allowing the transparent service function to work 1012 using its existing behavior. 1014 A SFC may be also be set up between end systems or network segments 1015 within the same Layer-2 bridged network. In this case, applying the 1016 procedures described earlier, the segments or groups of end systems 1017 are placed in distinct Layer-2 virtual networks, which are then then 1018 inter-connected via a sequence of intermediate Layer-2 virtual 1019 networks that form the links in the SFC. Each virtual network maps 1020 to a pair of ingress and egress MAC VRFs on the routing systems to 1021 which the SF instances are attached. The routing systems at the ends 1022 of the SFC will advertise the locally learnt or installed MAC entries 1023 using BGP-EVPN type-2 routes, which will get installed in the MAC 1024 VRFs at the other end. The intermediate systems may use default MAC 1025 routes installed in the ingress and egress MAC VRFs, or the other 1026 variations described earlier in this document. 1028 2.10. Header Transforming Service Functions 1030 If a service function performs an action that changes the source 1031 address in the packet header (e.g., NAT), the routes that were 1032 installed as described above may not support reverse flow traffic. 1034 The solution to this is for the controller modify the routes in the 1035 reverse direction to direct traffic into instances of the 1036 transforming service function. The original routes with a source 1037 prefix (Network-A in Figure 2) are replaced with a route that has a 1038 prefix that includes all the possible addresses that the source 1039 address could be mapped to. In the case of network address 1040 translation, this would correspond to the NAT pool. 1042 3. Load Balancing Along a Service Function Chain 1044 One of the key concepts driving NFV [NFVE2E]is the idea that each 1045 service function along an SFC can be separately scaled by changing 1046 the number of service function instances that implement it. This 1047 requires that load balancing be performed before entry into each 1048 service function. In this architecture, load balancing is performed 1049 in either or both of egress and ingress VRFs depending on the type of 1050 load balancing being performed, and if more than one service instance 1051 is connected to the same ingress VRF. 1053 3.1. SF Instances Connected to Separate VRFs 1055 If SF instances implementing a service in an SFC are each connected 1056 to separate VRFs(e.g. instances are connected to different routers or 1057 are running on different hosts), load balancing is performed in the 1058 egress VRFs of the previous service, or in the VRF that is the entry 1059 to the SFC. The controller distributes BGP multi-path routes to the 1060 egress VRFs. The destination prefix of each route is the ultimate 1061 destination network, or its representative aggregate or default. The 1062 next-hops in the ECMP set are BGP next-hops of the service instances 1063 attached to ingress VRFs of the next service in the SFC. The load 1064 balancing corresponds to BGP Multipath, which requires that the route 1065 distinguishers for each route are distinct in order to recognize that 1066 distinct paths should be used. Hence, each VRF in a distributed, SFC 1067 environment should have a unique route distinguisher. 1069 +------+ +-------------------------+ 1070 O----|SFI-11|---O |--- Data plane connection| 1071 // +------+ \\ |=== Encapsulation tunnel | 1072 // \\ | O VRF | 1073 // \\ | * Load balancer | 1074 // \\ +-------------------------+ 1075 // +------+ \\ 1076 Net-A-->O*====O---|SFI-12|---O====O-->Net-B 1077 \\ +------+ // 1078 \\ // 1079 \\ // 1080 \\ // 1081 \\ +------+ // 1082 O----|SFI-13|---O 1083 +------+ 1085 Egress VRF Load Balancing across SF Instances 1086 Connected to Different VRFs 1088 Figure 6 1090 In the diagram, above, a service function is implemented in three 1091 service instances each connected to separate VRFs. Traffic from 1092 Network-A arrives at VRF at the start of the SFC, and is load 1093 balanced across the service instances using a set of ECMP routes with 1094 next hops being the addresses of the routing systems containing the 1095 ingress VRFs and with labels that identify the ingress interfaces of 1096 the service instances. 1098 In the case that the bandwidth of the links between the load balancer 1099 and the ingress VRFs are unequal, or that the bandwidth capacity of 1100 the service function instances are unequal, this can be signalled in 1101 the routes for each ingress VRF using the extended community 1102 described in [draft-ietf-idr-link-bandwidth] and procedures from 1103 [draft-malhotra-bess-evpn-unequal-lb] could be followed. 1105 3.2. SF Instances Connected to the Same VRF 1107 When SF instances implementing a service in an SFC are connected to 1108 the same ingress VRF, load balancing is performed in the ingress VRF 1109 across the service instances connected to it. The controller will 1110 install routes in the ingress VRF to the destination network with the 1111 interfaces connected to each service instance as next hops. The 1112 ingress VRF will then use ECMP to load balance across the service 1113 instances. 1115 +------+ +-------------------------+ 1116 |SFI-11| |--- Data plane connection| 1117 +------+ |=== Encapsulation tunnel | 1118 / \ | O VRF | 1119 / \ | * Load balancer | 1120 / \ +-------------------------+ 1121 / +------+ \ 1122 Net-A-->O====O*---|SFI-12|----O====O-->Net-B 1123 \ +------+ / 1124 \ / 1125 \ / 1126 \ / 1127 +------+ 1128 |SFI-13| 1129 +------+ 1131 Ingress VRF Load Balancing across SF Instances 1132 Connected to the Same VRF 1134 Figure 7 1136 In the diagram, above, a service is implemented by three service 1137 instances that are connected to the same ingress and egress VRFs. 1139 The ingress VRF load balances across the ingress interfaces using 1140 ECMP, and the egress traffic is aggregated in the egress VRF. 1142 If forwarding labels that identify each SFI ingress interface are 1143 used, and if the routes to each SF instance are advertised with 1144 different route distinguishers, then it is possible to perform ECMP 1145 load balancing at the routing instance at the beginning of the 1146 encapsulation tunnel (which could be the egress VRF of the previous 1147 SF in the SFC). 1149 3.3. Combination of Egress and Ingress VRF Load Balancing 1151 In Figure 8, below, an example SFC is shown where load balancing is 1152 performed in both ingress and egress VRFs. 1154 +-------------------------+ 1155 |--- Data plane connection| 1156 +------+ |=== Encapsulation tunnel | 1157 |SFI-11| | O VRF | 1158 +------+ | * Load balancer | 1159 / \ +-------------------------+ 1160 / \ 1161 / +------+ \ +------+ 1162 O*---|SFI-12|---O*====O---|SFI-21|---O 1163 // +------+ \\ // +------+ \\ 1164 // \\// \\ 1165 // \\ \\ 1166 // //\\ \\ 1167 // +------+ // \\ +------+ \\ 1168 Net-A-->O*====O----|SFI-13|---O*====O---|SFI-22|---O====O-->Net-B 1169 +------+ +------+ 1170 ^ ^ ^ ^ ^ ^ 1171 | | | | | | 1172 | Ingress Egress | | | 1173 | Ingress Egress | 1174 SFC Entry SFC Exit 1176 Load Balancing across SF Instances 1178 Figure 8 1180 In Figure 8, above, an SFC is composed of two services implemented by 1181 three service instances and two service instances, respectively. The 1182 service instances SFI-11 and SFI-12 are connected to the same ingress 1183 and egress VRFs, and all the other service instances are connected to 1184 separate VRFs. 1186 Traffic entering the SFC from Network-A is load balanced across the 1187 ingress VRFs of the first service function by the chain entry VRF, 1188 and then load balanced again across the ingress interfaces of SFI-11 1189 and SFI-12 by the shared ingress VRF. Note that use of standard ECMP 1190 will lead to an uneven distribution of traffic between the three 1191 service instances (25% to SFI-11, 25% to SFI-12, and 50% to SFI-13). 1192 This issue can be mitigated through the use of BGP link bandwidth 1193 extended community [draft-ietf-idr-link-bandwidth] and use of 1194 procedures described in [draft-malhotra-bess-evpn-unequal-lb]. As 1195 described in the previous section, if a next-hop forwarding label is 1196 used, another way to mitigate this effect would be to advertise 1197 routes to each SF instance connected to a VRF with a different route 1198 distinguisher. 1200 After traffic passes through the first set of service instances, it 1201 is load balanced in each of the egress VRFs of the first set of 1202 service instances across the ingress VRFs of the next set of service 1203 instances. 1205 3.4. Forward and Reverse Flow Load Balancing 1207 This section discusses requirements in load balancing for forward and 1208 reverse paths when stateful service functions are deployed. 1210 3.4.1. Issues with Equal Cost Multi-Path Routing 1212 As discussed in the previous sections, load balancing in the forward 1213 SFC in the above example can automatically occur with standard BGP, 1214 if multiple equal cost routes to Network-B are installed into all the 1215 ingress VRFs, and each route directs traffic through a different 1216 service function instance in the next set. The multiple BGP routes 1217 in the routing table will translate to Equal Cost Multi-Path in the 1218 forwarding table. The hash used in the load balancing algorithm (per 1219 packet, per flow or per prefix) is implementation specific. 1221 If a service function is stateful, it is required that forward flows 1222 and reverse flows always pass through the same service function 1223 instance. Standard ECMP does not provide this capability, since the 1224 hash calculation will see different input data for the same flow in 1225 the forward and reverse directions (since the source and destination 1226 fields are reversed). 1228 Additionally, if the number of SF instances changes, either 1229 increasing to expand capacity, or decreases (planned, or due to a SF 1230 instance failure), the hash table in ECMP is recalculated, and most 1231 flows will be directed to a different SF instance and user sessions 1232 will be disrupted. 1234 There are a number of ways to satisfy the requirements of symmetric 1235 forward/reverse paths for flows and minimal disruption when SF 1236 instances are added to or removed from a set. Two techniques that 1237 can be employed are described in the following sections. 1239 3.4.2. Modified ECMP with Consistent Hash 1241 Symmetric forwarding into each side of an SF instance set can be 1242 achieved with a small modification to ECMP if the packet headers are 1243 preserved after passing through the SF instance set and assuming that 1244 the same hash function, same hash salt and same ordering association 1245 of hash buckets to ECMP routes is used in both directions. Each 1246 packet's 5-tuple data is used to calculate which hash bucket, and 1247 therefore which service instance, that the packet will be sent to, 1248 but the source and destination IP address and port information are 1249 swapped in the calculation in the reverse direction. This method 1250 only requires that the list of available service function instances 1251 is consistently maintained in load balance tables in all the routing 1252 systems rather than maintaining flow tables. This requirement can be 1253 met by the use of a distinct VPN route for each instance. 1255 In the SFC architecture described in this document, when SF instances 1256 are added or removed, the controller is required to install (or 1257 remove) routes to the SF instances. The controller could configure 1258 the load balancing function in VRFs that connect to each added (or 1259 removed) SF instance as part of the same network transaction as route 1260 updates to ensure that the load balancer configuration is 1261 synchronized with the set of SF instances. 1263 The consistent ordering among ECMP routes in the routing systems 1264 could be achieved through configuration of the routing systems by the 1265 controller using, for instance, Netconf; or when the routes are 1266 signaled using BGP by the controller or a routing system, the order 1267 for a given instance can be sent in a new 'Consistent Hash Sort 1268 Order' BGP Extended Community (defined in Section 9.2). 1270 The effect of rehashing when SF instances are added or removed can be 1271 minimized, or even eliminated using variations of the technique of 1272 consistent hashing [consistent-hash]. Details are outside the scope 1273 of this document. 1275 3.4.3. ECMP with Flow Table 1277 A second refinement that can ensure forward/reverse flow consistency, 1278 and also provides stability when the number of SF instances changes 1279 ('flow-stickiness'), is the use of dynamically configured IP flow 1280 tables in the VRFs. In this technique, flow tables are used to 1281 ensure that existing flows are unaffected if the number of ECMP 1282 routes changes, and that forward and reverse traffic passes through 1283 the same SF instance in each set of SF instances implementing a 1284 service function. 1286 The flow tables are set up as follows: 1288 1. User traffic with a new 5-tuple enters an egress VRF from a 1289 connected SF instance. 1291 2. The VRF calculates the ECMP hash across available routes (i.e., 1292 ECMP group) to the ingress interfaces of the SF instances in the 1293 next SF instance set. The consistent hash technique described in 1294 section 3.4.2 must be used here and in subsequent steps. 1296 3. The VRF creates a new flow entry for the 5-tuple of the new 1297 traffic with the next-hop being the chosen downstream ECMP group 1298 member (determined in the step 2. above). All subsequent packets 1299 for the same flow will be forwarded using flow lookup and, hence, 1300 will use the same next-hop. 1302 4. The encapsulated packet arrives in the routing system that hosts 1303 the ingress VRF for the selected SF instance. 1305 5. The ingress VRF of the next service instance determines if the 1306 packet came from a routing system that is in an ECMP group in the 1307 reverse direction(i.e., from this ingress VRF back to the 1308 previous set of SF instances). 1310 6. If an ECMP group is found, the ingress VRF creates a flow entry 1311 for the reversed 5-tuple with next-hop of the tunnel on which 1312 traffic arrived. This is for the traffic in the reverse 1313 direction. 1315 7. If multiple SF instances are connected to the ingress VRF, the 1316 ECMP consistent hash is used to choose which one to send the 1317 traffic into. 1319 8. A forward flow table entry is created for the traffic's 5-tuple 1320 with next hop of the interface of the SF instance chosen in the 1321 previous step. 1323 9. The packet is sent into the selected SF instance. 1325 The above method ensures that forward and reverse flows pass through 1326 the same SF instances, and that if the number of ECMP routes changes 1327 when SF instances are added or removed, all existing flows will 1328 continue to flow through the same SF instances, but new flows will 1329 use the new ECMP hash. The only flows affected will be those that 1330 were passing through an SF instance that was removed, and those will 1331 be spread among the remaining SF instances using the updated ECMP 1332 hash. 1334 If the consistent hash algorithm is used in both directions, then 1335 only the forwarding flow entries would be required, and would be 1336 built independently in each direction. If distinct VPN routes with 1337 next-hop forwarding labels are used, then only the flow table in step 1338 3 is sufficient to provide flow stickiness. 1340 3.4.4. Dealing with Different Hash Algorithms in an SFC 1342 In some cases, there will be two or more hash algorithms in 1343 forwarders along an SFC. E.g. when a physical router is at the entry 1344 and exit of the chain, and virtual forwarders are used within the 1345 chain. Forward and reverse flows will mostly not pass through the 1346 same SF instances of the first SF, and the SFC will not operate as 1347 intended if the first SF is stateful. It may be impractical, or 1348 prohibitively expensive to implement the flow table-based methods 1349 described above to achieve flow stability and symmetry. This issue 1350 can be mitigated by ensuring that the first SF is not stateful, or by 1351 placing a null SF between the physical router and the first actual SF 1352 in the SFC. This ensures that the hash method on both sides of 1353 stateful service instances is the same, and the SFC will operate with 1354 flow stability and symmetry if the methods described above are 1355 employed. 1357 4. Steering into SFCs Using a Classifier 1359 In many applications of SFCs, a classifier will be used to direct 1360 traffic into SFCs. The classifier inspects the first or first few 1361 packets in a flow to determine which SFC the flow should be sent 1362 into. The decision criteria can be based on just the IP 5-tuple of 1363 the header (i.e filter-based forwarding), or could involve analysis 1364 of the payload of packets using deep packet inspection. Integration 1365 with a subscriber management system such as PCRF or AAA may be 1366 required in order to identify which SFC to send traffic to based on 1367 subscriber policy. 1369 An example logical architecture is shown in Figure 9, below where a 1370 classifier is external to a physical router that is hosting the VRFs 1371 that form the ends of two SFC sets. In the case of filter-based 1372 forwarding, classification could occur in a VRF on the router. 1374 +----------+ 1375 | PCRF/AAA | 1376 +-----+----+ 1377 : 1378 : 1379 Subscriber +-----+------+ 1380 Traffic----->| Classifier | 1381 +------------+ 1382 | | 1383 +-------|---|------------------------+ 1384 | | | Router | 1385 | | | | 1386 | O O X--------->Internet 1387 | | | / \ | 1388 | | | O O | 1389 +-------|---|----------------|---|---+ 1390 | | +---+ +---+ | | 1391 | +--+ U +---+ V +-+ | 1392 | +---+ +---+ | 1393 | | 1394 | +---+ +---+ +---+ | 1395 +--+ X +---+ Y +---+ Z +-+ 1396 +---+ +---+ +---+ 1398 Subscriber/Application-Aware Steering with a Classifier 1400 Figure 9 1402 In the diagram, the classifier receives subscriber traffic and sends 1403 the traffic out of one of two logical interfaces, depending on 1404 classification criteria. The logical interfaces of the classifier 1405 are connected to VRFs in a router that are entries to two SFCs (shown 1406 as O in the diagram). 1408 In this scenario, the entry VRF for each chain does not advertise the 1409 destination network prefixes and the modified method of setting 1410 prefixes, described in Section 2.8.2 can be employed. Also, the exit 1411 VRF for each SFC does not peer with a gateway or proxy node in the 1412 destination network and packets are forwarded using IP lookup in the 1413 main routing table or in a VRF that the exit traffic from the SFCs is 1414 directed into (shown as X in the diagram). A flow table may be 1415 required to ensure that reverse traffic is sent into the correct SFC. 1417 An alternative would be where the classifier is itself a distributed, 1418 virtualized service function, but with multiple egress interfaces. 1419 In that case, each virtual classifier instance could be attached to a 1420 set of VRFs that connect to different SFCs. Each chain 1421 entry VRF would load balance across the first SF instance set in its 1422 SFC. The reverse flow table mechanism described in Section 3.4.3 1423 could be employed to ensure that flows return to the originating 1424 classifier instance which may maintain subscriber context and perform 1425 charging and accounting. 1427 5. External Domain Co-ordination 1429 It is likely that SFCs will be managed as a separate administrative 1430 domain from the networks that they receive traffic from, and send 1431 traffic to. If the connected networks use BGP for route 1432 distribution, the controller in the SFC domain can join the network 1433 domains by creating BGP peering sessions with routing systems or 1434 route reflectors in those network domains to exchange VPN routes, or 1435 with local border routers that peer with the external domains. While 1436 a controller can modify route targets for the VRFs within its SFC 1437 domain, it is likely to not have any control over the external 1438 networks with which it is peering. Hence, the design does not assume 1439 that the RTs of external network domains can be modified by the 1440 controller. It may however learn those RTs and use them in it's 1441 modified route advertisements. 1443 In order to steer traffic from external network domains into an SFC, 1444 the controller will advertise a destination network's prefixes into 1445 the peering source network domain with a BGP next-hop and label 1446 associated with the SFC entry point that may be on a routing system 1447 attached to the first SF instance. This advertisement may be over 1448 regular MP-BGP/VPN peering which assumes existing standard VPN 1449 routing/forwarding behavior on the network domain's routers (PEs/ 1450 ASBRs). The controller can learn routes to networks in external 1451 domains at the egress of an SFC and advertise routes to those network 1452 into other external domains using the first ingress routing instance 1453 as the next hop thus allowing dynamic steering through re- 1454 origination of routes. 1456 An operational benefit of this approach is that the SFC topology 1457 within a domain need not be exposed to other domains. Additionally, 1458 using non-specific routes inside an SFC, as described in 1459 Section 2.8.1, means that new networks can be attached to a SFC 1460 without needing to configure prefixes inside the chain. 1462 The controller will typically remove the destination network's RTs 1463 and replace them with the RTs of the source network while advertising 1464 the modified routes. Alternatively, an external domain may be 1465 provisioned with an additional export-only RT and an import- only RT 1466 that the controller can use. 1468 6. Fine-grained steering using BGP Flow-Spec 1470 When steering traffic from an external network domain into an SFC 1471 based on attributes of the packet flow, BGP Flow-spec can be used as 1472 a signaling option. 1474 In this case, the controller can advertise one or more flow-spec 1475 routes into the entry VRF with the appropriate Service-topology-RT 1476 for the SFC. Alternatively, it can use the procedures described in 1477 [RFC5575] or [flowspec-redirect-ip] on the gateway router to redirect 1478 traffic towards the first SF. 1480 If it is desired to steer specific flows from a network domain's 1481 existing routers, the controller can advertise the above flow-spec 1482 routes to the network domain's border routers or route reflectors. 1484 7. Controller Federation 1486 When SFCs are distributed geographically, or in very large-scale 1487 environments, there may be multiple SFC controllers present and they 1488 may variously employ both of the SFC creation methods described in 1489 Section 2.6. If there is a requirement for SFCs to span controller 1490 domains there may be a requirement to exchange information between 1491 controllers. Again, a BGP session between controllers can be used to 1492 exchange route information as described in the previous sections and 1493 allow such domain spanning SFCs to be created. 1495 8. Coordination Between SF Instances and Controller using BGP 1497 In many cases, the configuration of SF instance determines its 1498 network behavior. E.g. when NAT pools are set up, or when an SSL 1499 gateway is configured with a set of enterprise IP addresses to use. 1500 In these cases, the addresses that will be used by the SFs need to be 1501 known in the networks connecting to them in order that traffic can be 1502 properly routed. When SFCs are involved, this means that the 1503 controller has to be notified when such configuration changes are 1504 made in SF instances. Sometimes, the changes will be made by end- 1505 customers and it is desirable the controller adjust the SFC routing 1506 configuration automatically when the change is made, and without 1507 customers needing to notify the service provider via a portal, for 1508 instance, or requiring development of integration modules linking the 1509 SF instances and the controller. 1511 One option for automatic notification for SFs that support BGP is for 1512 the connected forwarding system (physical or virtual SFF) to also 1513 support BGP, and for SF instances to be configured to peer with the 1514 SFF. When changes are made to the configuration of a SF instance, 1515 that for example, the SF will accept packets from a particular 1516 network prefix on one of its interfaces, the SF instance will send a 1517 BGP route update to the SFF it is connected to and which it has a BGP 1518 session with. The controller can then adjust the routes along SFCs 1519 to ensure that packets with destinations in the new prefix reach the 1520 reconfigured SF instance. 1522 BGP could also be used to signal from the controller to a SF instance 1523 that certain traffic should be sent out from a particular interface. 1524 This could be used to direct suspect traffic to a security scrubbing 1525 center,for example. 1527 Note that the SFF need not support a BGP stack itself; it can proxy 1528 BGP messages to the controller which will support such a stack. 1530 9. BGP Extended Communities 1532 9.1. Route-Target Record 1534 Route-Target Record (RT-Record)is defined as a transitive BGP 1535 Extended Community, that contains an Route-Target value representing 1536 one of the RTs that the route has been attached with previously, and 1537 which may no longer be attached to the route on subsequent re- 1538 advertisements (see Section 2.8.5). 1540 A Sub-Type code 0x13 is assigned in the three BGP Extended Community 1541 types - Two-Octet AS-Specific 0x00, IPv4-Address-Specific 0x01 and 1542 Four-Octet AS-Specific 0x02. A Sub-Type code 0x0013 is also assigned 1543 in the BGP Transitive IPv6 Address-Specific Extended Community. 1545 The Extended Community is encoded as follows: 1547 0 1 2 3 1548 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 1549 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1550 | 0x00,0x01,0x02| Sub-Type=0x13 | Route-Target Value | 1551 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1552 | Route-Target Value contd. | 1553 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1555 The Type field of the BGP Route-Target Extended Community is copied 1556 into the Type field of the RT Record Extended Community. 1558 The Value field (Global Administrator and Local Administrator) of the 1559 Route-Target Extended Community is copied into the Route-Target Value 1560 field of the RT Record Extended Community. 1562 When comparing a RT-Record to a Route-Target, only the Type and the 1563 Route-Target value fields are used in the comparison. The sub-type 1564 field is masked out. 1566 When a speaker re-originates a route that contains one or more RTs, 1567 it must add each of these RTs as RT Record extended communities in 1568 the re-originated route. 1570 A speaker must not re-originate a route with an RT, if this RT is 1571 already present as an RT Record extended community. 1573 9.2. Consistent Hash Sort Order 1575 Consistent Hash Sort Order is an optional transitive Opaque BGP 1576 Extended Community of type 0x14, defined as follows: 1578 Type Field : The value of the high-order octet is 0x03 (transitive 1579 opaque). The value of the low-order octet is assigned 1580 as 0x14 by IANA from the Transitive Opaque Extended 1581 Community Sub-Types registry. 1583 Value Field : The value field contains a Sort Order sub-field that 1584 indicates the relative order of this route among the 1585 ECMP set for the prefix, to be sorted in increasing 1586 order. It is a 32-bit unsigned integer. The field is 1587 encoded as shown below: 1589 +------------------------------+ 1590 | Sort Order (4 octets) | 1591 +------------------------------+ 1592 | Reserved (2 octets) | 1593 +------------------------------+ 1595 9.3. Load Balance Settings 1597 Consistent Hash Sort Order is an optional transitive Opaque BGP 1598 Extended Community of type 0x14, defined as follows: 1600 Type Field : The value of the high-order octet is 0x03 (transitive 1601 opaque). The value of the low-order octet is assigned 1602 as 0xaa by IANA from the Transitive Opaque Extended 1603 Community Sub-Types registry. 1605 Value Field : The value field contains flags that indicate which 1606 values in an IP packet's 5-tuple should be used as 1607 inputs to the ECMP hash algorithm. The field is 1608 encoded as shown below: 1610 * 0 1 2 3 1611 * 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 1612 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1613 * | Type 0x03 | Sub-Type 0xaa |s d c p P R R R|R R R R R R R R| 1614 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1615 * | Reserved |B R R R R R R R| Reserved | Reserved | 1616 * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1617 * 1618 * Type: 0x03 Opaque 1619 * SubType: 0xAA LoadBalance attribute information (TBA) 1620 * s: Use l3_source_address ECMP Load-balancing 1621 * d: Use l3_destination_address ECMP Load-balancing 1622 * c: Use l4_protocol ECMP Load-balancing 1623 * p: Use l4_source_port ECMP Load-balancing 1624 * P: Use l4_destination_port ECMP Load-balancing 1625 * B: Use source_bias (instead of ECMP load-balancing) 1626 * R: Reserved 1628 10. Summary and Conclusion 1630 The architecture for service function chains described in this 1631 document uses virtual networks implemented as overlays in order to 1632 create service function chains. The virtual networks use standards- 1633 based encapsulation tunneling, such as MPLS over GRE/UDP or VXLAN, to 1634 transport packets into an SFC and between service function instances 1635 without routing in the user address space. Two methods of installing 1636 routes to form service chains are described. 1638 In environments with physical routers, a controller may operate in 1639 tandem with existing BGP route reflectors, and would contain the SFC 1640 topology model, and the ability to install the local static interface 1641 routes to SF instances. In a virtualized environment, the controller 1642 can emulate route refection internally and simply install required 1643 routes directly without advertisements occurring. 1645 11. Security Considerations 1647 The security considerations for SFCs are broadly similar to those 1648 concerning the data, control and management planes of any device 1649 placed in a network. Details are out of scope for this document. 1651 12. IANA Considerations 1653 The new BGP Extended Communities in are assigned types as defined 1654 above in the IANA registry for extended communities. 1656 13. Acknowledgments 1658 The authors would like to thank D. Daino, D.R. Lopez, D. Bernier, 1659 W. Haeffner, A. Farrel, L. Fang, and N. So, for their 1660 contributions to the earlier drafts. The authors would also like to 1661 thank the following individuals for their review and feedback on the 1662 original proposals: E. Rosen, J. Guchard, P. Quinn, P. Bosch, D. 1663 Ward, A. Ganesan, N. Seth, G. Pildush and N. Bitar. The authors 1664 also thank Wim Henderickx for his useful suggestions on several 1665 aspects of the draft. 1667 14. References 1669 14.1. Normative References 1671 [RFC3031] Rosen, E., Viswanathan, A., and R. Callon, "Multiprotocol 1672 Label Switching Architecture", RFC 3031, 1673 DOI 10.17487/RFC3031, January 2001, 1674 . 1676 [RFC4023] Worster, T., Rekhter, Y., and E. Rosen, Ed., 1677 "Encapsulating MPLS in IP or Generic Routing Encapsulation 1678 (GRE)", RFC 4023, DOI 10.17487/RFC4023, March 2005, 1679 . 1681 [RFC4271] Rekhter, Y., Ed., Li, T., Ed., and S. Hares, Ed., "A 1682 Border Gateway Protocol 4 (BGP-4)", RFC 4271, 1683 DOI 10.17487/RFC4271, January 2006, 1684 . 1686 [RFC4360] Sangli, S., Tappan, D., and Y. Rekhter, "BGP Extended 1687 Communities Attribute", RFC 4360, DOI 10.17487/RFC4360, 1688 February 2006, . 1690 [RFC4364] Rosen, E. and Y. Rekhter, "BGP/MPLS IP Virtual Private 1691 Networks (VPNs)", RFC 4364, DOI 10.17487/RFC4364, February 1692 2006, . 1694 [RFC4760] Bates, T., Chandra, R., Katz, D., and Y. Rekhter, 1695 "Multiprotocol Extensions for BGP-4", RFC 4760, 1696 DOI 10.17487/RFC4760, January 2007, 1697 . 1699 [RFC5575] Marques, P., Sheth, N., Raszuk, R., Greene, B., Mauch, J., 1700 and D. McPherson, "Dissemination of Flow Specification 1701 Rules", RFC 5575, DOI 10.17487/RFC5575, August 2009, 1702 . 1704 [RFC6241] Enns, R., Ed., Bjorklund, M., Ed., Schoenwaelder, J., Ed., 1705 and A. Bierman, Ed., "Network Configuration Protocol 1706 (NETCONF)", RFC 6241, DOI 10.17487/RFC6241, June 2011, 1707 . 1709 [RFC7348] Mahalingam, M., Dutt, D., Duda, K., Agarwal, P., Kreeger, 1710 L., Sridhar, T., Bursell, M., and C. Wright, "Virtual 1711 eXtensible Local Area Network (VXLAN): A Framework for 1712 Overlaying Virtualized Layer 2 Networks over Layer 3 1713 Networks", RFC 7348, DOI 10.17487/RFC7348, August 2014, 1714 . 1716 [RFC7510] Xu, X., Sheth, N., Yong, L., Callon, R., and D. Black, 1717 "Encapsulating MPLS in UDP", RFC 7510, 1718 DOI 10.17487/RFC7510, April 2015, 1719 . 1721 [RFC7665] Halpern, J., Ed. and C. Pignataro, Ed., "Service Function 1722 Chaining (SFC) Architecture", RFC 7665, 1723 DOI 10.17487/RFC7665, October 2015, 1724 . 1726 [RFC8365] Sajassi, A., Ed., Drake, J., Ed., Bitar, N., Shekhar, R., 1727 Uttaro, J., and W. Henderickx, "A Network Virtualization 1728 Overlay Solution Using Ethernet VPN (EVPN)", RFC 8365, 1729 DOI 10.17487/RFC8365, March 2018, 1730 . 1732 14.2. Informational References 1734 [consistent-hash] 1735 Karger, D., Lehman, E., Leighton, T., Panigrahy, R., 1736 Levine, M., and D. Lewin, ""Consistent Hashing and Random 1737 Trees: Distributed Caching Protocols for Relieving Hot 1738 Spots on the World Wide Web"", 1997, . 1742 [draft-ietf-idr-link-bandwidth] 1743 Mohapatra, P. and R. Fernando, ""BGP Link Bandwidth 1744 Extended Community"", March 2018. 1746 [draft-malhotra-bess-evpn-unequal-lb] 1747 Malhotra, N., Sajassi, A., Rabadan, J., Drake, J., 1748 Lingala, A., and S. Thoria, ""Weighted Multi-Path 1749 Procedures for EVPN All-Active Multi-Homing"", June 2018. 1751 [flowspec-redirect-ip] 1752 Uttaro, J., Haas, J., Texier, M., Karch, A., Sreekanth, 1753 A., Ray, S., Simpson, A., and W. Henderickx, ""BGP Flow- 1754 Spec Redirect to IP Action"", February 2015. 1756 [idr-tunnel-encaps] 1757 Rosen, E., Patel, K., and G. van de Velde, ""The BGP 1758 Tunnel Encapsulation Attribute"", February 2018. 1760 [NFVE2E] ETSI, ""Network Functions Virtualisation (NFV): 1761 Architectural Framework"", 2013. 1763 Authors' Addresses 1765 Rex Fernando 1766 Cisco Systems 1767 170 W. Tasman Drive 1768 San Jose, CA 95134 1770 Email: rex@cisco.com 1772 Stuart Mackie 1773 Juniper Networks 1774 1133 Innovation Way 1775 Sunnyvale, CA 94089 1777 Email: wsmackie@juniper.net 1779 Dhananjaya Rao 1780 Cisco Systems 1781 170 W. Tasman Drive 1782 San Jose, CA 95134 1784 Email: dhrao@cisco.com 1786 Bruno Rijsman 1788 Email: brunorijsman@gmail.com 1789 Maria Napierala 1790 ATT Labs 1791 200 Laurel Avenue 1792 Middletown, NJ 07748 1794 Email: mnapierala@att.com 1796 Thomas Morin 1797 Orange 1798 2, Avenue Pierre Marzin 1799 Lannion, France 22307 1801 Email: thomas.morin@orange.com