idnits 2.17.1 draft-rtg-dt-encap-02.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 an IANA Considerations section. (See Section 2.2 of https://www.ietf.org/id-info/checklist for how to handle the case when there are no actions for IANA.) ** 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 292: '...italized keyword MUST is used as defin...' RFC 2119 keyword, line 365: '...lancing procedure MUST choose the same...' Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (May 21, 2015) is 3256 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- == Unused Reference: 'I-D.xu-bier-encapsulation' is defined on line 1809, but no explicit reference was found in the text == Outdated reference: A later version (-19) exists of draft-ietf-tsvwg-rfc5405bis-02 ** Obsolete normative reference: RFC 2460 (Obsoleted by RFC 8200) ** Obsolete normative reference: RFC 5405 (Obsoleted by RFC 8085) ** Obsolete normative reference: RFC 6830 (Obsoleted by RFC 9300, RFC 9301) == Outdated reference: A later version (-05) exists of draft-farrelll-mpls-opportunistic-encrypt-04 == Outdated reference: A later version (-02) exists of draft-herbert-remotecsumoffload-01 == Outdated reference: A later version (-08) exists of draft-ietf-nvo3-arch-03 == Outdated reference: A later version (-11) exists of draft-ietf-sfc-architecture-08 == Outdated reference: A later version (-15) exists of draft-ietf-tsvwg-circuit-breaker-01 == Outdated reference: A later version (-19) exists of draft-ietf-tsvwg-gre-in-udp-encap-06 == Outdated reference: A later version (-12) exists of draft-saldana-tsvwg-simplemux-02 == Outdated reference: A later version (-04) exists of draft-wei-tsvwg-tunnel-congestion-feedback-03 == Outdated reference: A later version (-06) exists of draft-xu-bier-encapsulation-02 Summary: 5 errors (**), 0 flaws (~~), 12 warnings (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 RTGWG E. Nordmark (ed) 3 Internet-Draft Arista Networks 4 Intended status: Informational A. Tian 5 Expires: November 22, 2015 Ericsson Inc. 6 J. Gross 7 VMware 8 J. Hudson 9 Brocade Communications Systems, 10 Inc. 11 L. Kreeger 12 Cisco Systems, Inc. 13 P. Garg 14 Microsoft 15 P. Thaler 16 Broadcom Corporation 17 T. Herbert 18 Google 19 May 21, 2015 21 Encapsulation Considerations 22 draft-rtg-dt-encap-02 24 Abstract 26 The IETF Routing Area director has chartered a design team to look at 27 common issues for the different data plane encapsulations being 28 discussed in the NVO3 and SFC working groups and also in the BIER 29 BoF, and also to look at the relationship between such encapsulations 30 in the case that they might be used at the same time. The purpose of 31 this design team is to discover, discuss and document considerations 32 across the different encapsulations in the different WGs/BoFs so that 33 we can reduce the number of wheels that need to be reinvented in the 34 future. 36 Status of this Memo 38 This Internet-Draft is submitted in full conformance with the 39 provisions of BCP 78 and BCP 79. 41 Internet-Drafts are working documents of the Internet Engineering 42 Task Force (IETF). Note that other groups may also distribute 43 working documents as Internet-Drafts. The list of current Internet- 44 Drafts is at http://datatracker.ietf.org/drafts/current/. 46 Internet-Drafts are draft documents valid for a maximum of six months 47 and may be updated, replaced, or obsoleted by other documents at any 48 time. It is inappropriate to use Internet-Drafts as reference 49 material or to cite them other than as "work in progress." 51 This Internet-Draft will expire on November 22, 2015. 53 Copyright Notice 55 Copyright (c) 2015 IETF Trust and the persons identified as the 56 document authors. All rights reserved. 58 This document is subject to BCP 78 and the IETF Trust's Legal 59 Provisions Relating to IETF Documents 60 (http://trustee.ietf.org/license-info) in effect on the date of 61 publication of this document. Please review these documents 62 carefully, as they describe your rights and restrictions with respect 63 to this document. Code Components extracted from this document must 64 include Simplified BSD License text as described in Section 4.e of 65 the Trust Legal Provisions and are provided without warranty as 66 described in the Simplified BSD License. 68 Table of Contents 70 1. Design Team Charter . . . . . . . . . . . . . . . . . . . . . 4 71 2. Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 72 3. Common Issues . . . . . . . . . . . . . . . . . . . . . . . . 6 73 4. Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 74 5. Assumptions . . . . . . . . . . . . . . . . . . . . . . . . . 7 75 6. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 8 76 7. Entropy . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 77 8. Next-protocol indication . . . . . . . . . . . . . . . . . . . 9 78 9. MTU and Fragmentation . . . . . . . . . . . . . . . . . . . . 11 79 10. OAM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 80 11. Security Considerations . . . . . . . . . . . . . . . . . . . 14 81 11.1. Encapsulation-specific considerations . . . . . . . . . . 14 82 11.2. Virtual network isolation . . . . . . . . . . . . . . . . 16 83 11.3. Packet level security . . . . . . . . . . . . . . . . . . 17 84 11.4. In summary: . . . . . . . . . . . . . . . . . . . . . . . 17 85 12. QoS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 86 13. Congestion Considerations . . . . . . . . . . . . . . . . . . 18 87 14. Header Protection . . . . . . . . . . . . . . . . . . . . . . 20 88 15. Extensibility Considerations . . . . . . . . . . . . . . . . . 22 89 16. Layering Considerations . . . . . . . . . . . . . . . . . . . 25 90 17. Service model . . . . . . . . . . . . . . . . . . . . . . . . 26 91 18. Hardware Friendly . . . . . . . . . . . . . . . . . . . . . . 27 92 18.1. Considerations for NIC offload . . . . . . . . . . . . . 28 93 19. Middlebox Considerations . . . . . . . . . . . . . . . . . . . 32 94 20. Related Work . . . . . . . . . . . . . . . . . . . . . . . . . 33 95 21. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 34 96 22. Open Issues . . . . . . . . . . . . . . . . . . . . . . . . . 34 97 23. Change Log . . . . . . . . . . . . . . . . . . . . . . . . . . 35 98 24. References . . . . . . . . . . . . . . . . . . . . . . . . . . 35 99 24.1. Normative References . . . . . . . . . . . . . . . . . . 35 100 24.2. Informative References . . . . . . . . . . . . . . . . . 37 101 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 40 103 1. Design Team Charter 105 There have been multiple efforts over the years that have resulted in 106 new or modified data plane behaviors involving encapsulations. That 107 includes IETF efforts like MPLS, LISP, and TRILL but also industry 108 efforts like VXLAN and NVGRE. These collectively can be seen as a 109 source of insight into the properties that data planes need to meet. 110 The IETF is currently working on potentially new encapsulations in 111 NVO3 and SFC and considering working on BIER. In addition there is 112 work on tunneling in the INT area. 114 This is a short term design team chartered to collect and construct 115 useful advice to parties working on new or modified data plane 116 behaviors that include additional encapsulations. The goal is for 117 the group to document useful advice gathered from interacting with 118 ongoing efforts. An Internet Draft will be produced for IETF92 to 119 capture that advice, which will be discussed in RTGWG. 121 Data plane encapsulations face a set of common issues such as: 122 o How to provide entropy for ECMP 123 o Issues around packet size and fragmentation/reassembly 124 o OAM - what support is needed in an encapsulation format? 125 o Security and privacy. 126 o QoS 127 o Congestion Considerations 128 o IPv6 header protection (zero UDP checksum over IPv6 issue) 129 o Extensibility - e.g., for evolving OAM, security, and/or 130 congestion control 131 o Layering of multiple encapsulations e.g., SFC over NVO3 over BIER 132 The design team will provide advice on those issues. The intention 133 is that even where we have different encapsulations for different 134 purposes carrying different information, each such encapsulation 135 doesn't have to reinvent the wheel for the above common issues. 137 The design team will look across the routing area in particular at 138 SFC, NVO3 and BIER. It will not be involved in comparing or 139 analyzing any particular encapsulation formats proposed in those WGs 140 and BoFs but instead focus on common advice. 142 2. Overview 144 The references provide background information on NVO3, SFC, and BIER. 145 In particular, NVO3 is introduced in [RFC7364], [RFC7365], and 146 [I-D.ietf-nvo3-arch]. SFC is introduced in 147 [I-D.ietf-sfc-architecture] and [I-D.ietf-sfc-problem-statement]. 148 Finally, the information on BIER is in 149 [I-D.shepherd-bier-problem-statement], 151 [I-D.wijnands-bier-architecture], and 152 [I-D.wijnands-mpls-bier-encapsulation]. We assume the reader has 153 some basic familiarity with those proposed encapsulations. The 154 Related Work section points at some prior work that relates to the 155 encapsulation considerations in this document. 157 Encapsulation protocols typically have some unique information that 158 they need to carry. In some cases that information might be modified 159 along the path and in other cases it is constant. The in-flight 160 modifications has impacts on what it means to provide security for 161 the encapsulation headers. 162 o NVO3 carries a VNI Identifier edge to edge which is not modified. 163 There has been OAM discussions in the WG and it isn't clear 164 whether some of the OAM information might be modified in flight. 165 o SFC carries service meta-data which might be modified or 166 unmodified as the packets follow the service path. SFC talks of 167 some loop avoidance mechanism which is likely to result in 168 modifications for for each hop in the service chain even if the 169 meta-data is unmodified. 170 o BIER carries a bitmap of egress ports to which a packet should be 171 delivered, and as the packet is forwarded down different paths 172 different bits are cleared in that bitmap. 174 Even if information isn't modified in flight there might be devices 175 that wish to inspect that information. For instance, one can 176 envision future NVO3 security devices which filter based on the 177 virtual network identifier. 179 The need for extensibility is different across the protocols 180 o NVO3 might need some extensions for OAM and security. 181 o SFC is all about carrying service meta-data along a path, and 182 different services might need different types and amount of meta- 183 data. 184 o BIER might need variable number of bits in their bitmaps, or other 185 future schemes to scale up to larger network. 186 The extensibility needs and constraints might be different when 187 considering hardware vs. software implementations of the 188 encapsulation headers. NIC hardware might have different constraints 189 than switch hardware. 191 As the IETF designs these encapsulations the different WGs solve the 192 issues for their own encapsulation. But there are likely to be 193 future cases when the different encapsulations are combined in the 194 same header. For instance, NVO3 might be a "transport" used to carry 195 SFC between the different hops in the service chain. 197 Most of the issues discussed in this document are not new. The IETF 198 and industry as specified and deployed many different encapsulation 199 or tunneling protocols over time, ranging from simple IP-in-IP and 200 GRE encapsulation, IPsec, pseudo-wires, session-based approached like 201 L2TP, and the use of MPLS control and data planes. IEEE 802 has also 202 defined layered encapsulation for Provider Backbone Bridges (PBB) and 203 IEEE 802.1Qbp (ECMP). This document tries to leverage what we 204 collectively have learned from that experience and summarize what 205 would be relevant for new encapsulations like NVO3, SFC, and BIER. 207 3. Common Issues 209 [This section is mostly a repeat of the charter but with a few 210 modifications and additions.] 212 Any new encapsulation protocol would need to address a large set of 213 issues that are not central to the new information that this protocol 214 intends to carry. The common issues explored in this document are: 215 o How to provide entropy for Equal Cost MultiPath (ECMP) routing 216 o Issues around packet size and fragmentation/reassembly 217 o Next header indication - each encapsulation might be able to carry 218 different payloads 219 o OAM - what support is needed in an encapsulation format? 220 o Security and privacy 221 o QoS 222 o Congestion Considerations 223 o Header protection 224 o Extensibility - e.g., for evolving OAM, security, and/or 225 congestion control 226 o Layering of multiple encapsulations e.g., SFC over NVO3 over BIER 227 o Importance of being friendly to hardware and software 228 implementations 230 The degree to which these common issues apply to a particular 231 encapsulation can differ based on the intended purpose of the 232 encapsulation. But it is useful to understand all of them before 233 determining which ones apply. 235 4. Scope 237 It is important to keep in mind what we are trying to cover and not 238 cover in this document and effort. This is 239 o A look across the three new encapsulations, while taking lots of 240 previous work into account 241 o Focus on the class of encapsulations that would run over IP/UDP. 242 That was done to avoid being distracted by the data-plane and 243 control-plane interaction, which is more significant for protocols 244 that are designed to run over "transports" that maintain session 245 or path state. 246 o We later expanded the scope somewhat to consider how the 247 encapsulations would play with MPLS "transport", which is 248 important because SFC and BIER seem to target being independent of 249 the underlying "transport" 251 However, this document and effort is NOT intended to: 252 o Design some new encapsulation header to rule them all 253 o Design yet another new NVO3 encapsulation header 254 o Try to select the best encapsulation header 255 o Evaluate any existing and proposed encapsulations 257 While the origin and focus of this document is the routing area and 258 in particular NVO3, SFC, and BIER, the considerations apply to other 259 encapsulations that are being defined in the IETF and elsewhere. 260 There seems to be an increase in the number of encapsulations being 261 defined to run over UDP, where there might already exist an 262 encapsulation over IP or Ethernet. Feedback on how these 263 considerations apply in those contexts is welcome. 265 5. Assumptions 267 The design center for the new encapsulations is a well-managed 268 network. That network can be a datacenter network (plus datacenter 269 interconnect) or a service provider network. Based on the existing 270 and proposed encapsulations in those environment it is reasonable to 271 make these assumptions: 272 o The MTU is carefully managed and configured. Hence an 273 encapsulation protocol can make the packets bigger without 274 resulting in a requirement for fragmentation and reassembly 275 between ingress and egress. (However, it might be useful to 276 detecting MTU misconfigurations.) 277 o In general an encapsulation needs some approach for congestion 278 management. But the assumptions are different than for arbitrary 279 Internet paths in that the underlay might be well-provisioned and 280 better policed at the edge, and due to multi-tenancy, the 281 congestion control in the endpoints might be even less trusted 282 than on the Internet at large. 284 The goal is to implement these encapsulations in hardware and 285 software hence we can't assume that the needs of either 286 implementation approach can trump the needs of the other. In 287 particular, around extensibility the needs and constraints might be 288 quite different. 290 6. Terminology 292 The capitalized keyword MUST is used as defined in 293 http://en.wikipedia.org/wiki/Julmust 295 TBD: Refer to existing documents for at least NVO3 and SFC 296 terminology. We use at least the VNI ID in this document. 298 7. Entropy 300 In many cases the encapsulation format needs to enable ECMP in 301 unmodified routers. Those routers might use different fields in TCP/ 302 UDP packets to do ECMP without a risk of reordering a flow. 304 The common way to do ECMP-enabled encapsulation over IP today is to 305 add a UDP header and to use UDP with the UDP source port carrying 306 entropy from the inner/original packet headers as in LISP [RFC6830]. 307 The total entropy consists of 14 bits in the UDP source port (using 308 the ephemeral port range) plus the outer IP addresses which seems to 309 be sufficient for entropy; using outer IPv6 headers would give the 310 option for more entropy should it be needed in the future. 312 In some environments it might be fine to use all 16 bits of the port 313 range. However, middleboxes might make assumptions about the system 314 ports or user ports. But they should not make any assumptions about 315 the ports in the Dynamic and/or Private Port range, which have the 316 two MSBs set to 11b. 318 The UDP source port might change over the lifetime of an encapsulated 319 flow, for instance for DoS mitigation or re-balancing load across 320 ECMP. 322 There is some interaction between entropy and OAM and extensibility 323 mechanism. It is desirable to be able to send OAM packets to follow 324 the same path as network packets. Hence OAM packets should use the 325 same entropy mechanism as data packets. While routers might use 326 information in addition the entropy field and outer IP header, they 327 can not use arbitrary parts of the encapsulation header since that 328 might result in OAM frames taking a different path. Likewise if 329 routers look past the encapsulation header they need to be aware of 330 the extensibility mechanism(s) in the encapsulation format to be able 331 to find the inner headers in the presence of extensions; OAM frames 332 might use some extensions e.g. for timestamps. 334 Architecturally the entropy and the next header field are really part 335 of enclosing delivery header. UDP with entropy goes hand-in-hand 336 with the outer IP header. Thus the UDP entropy is present for the 337 underlay IP routers the same way that an MPLS entropy label is 338 present for LSRs. The entropy above is all about providing entropy 339 for the outer delivery of the encapsulated packets. 341 It has been suggested that when IPv6 is used it would not be 342 necessary to add a UDP header for entropy, since the IPv6 flow label 343 can be used for entropy. (This assumes that there is an IP protocol 344 number for the encapsulation in addition to a UDP destination port 345 number since UDP would be used with IPv4 underlay. And any use of 346 UDP checksums would need to be replaced by an encaps-specific 347 checksum or secure hash.) While such an approach would save 8 bytes 348 of headers when the underlay is IPv6, it does assume that the 349 underlay routers use the flow label for ECMP, and it also would make 350 the IPv6 approach different than the IPv4 approach. Currently the 351 leaning is towards recommending using the UDP encapsulation for both 352 IPv4 and IPv6 underlay. The IPv6 flow label can be used for 353 additional entropy if need be. 355 Note that in the proposed BIER encapsulation 356 [I-D.wijnands-mpls-bier-encapsulation], there is an an 8-bit field 357 which specifies an entropy value that can be used for load balancing 358 purposes. This entropy is for the BIER forwarding decisions, which 359 is independent of any outer delivery ECMP between BIER routers. Thus 360 it is not part of the delivery ECMP discussed in this section. 361 [Note: For any given bit in BIER (that identifies an exit from the 362 BIER domain) there might be multiple immediate next hops. The 363 BIER entropy field is used to select that next hop as part of BIER 364 processing. The BIER forwarding process may do equal cost load 365 balancing, but the load balancing procedure MUST choose the same 366 path for any two packets have the same entropy value.] 368 In summary: 369 o The entropy is associated with the transport, that is an outer IP 370 header or MPLS. 371 o In the case of IP transport use >=14 bits of UDP source port, plus 372 outer IPv6 flowid for entropy. 374 8. Next-protocol indication 376 Next-protocol indications appear in three different context for 377 encapsulations. 379 Firstly, the transport delivery mechanism for the encapsulations we 380 discuss in this document need some way to indicate which 381 encapsulation header (or other payload) comes next in the packet. 382 Some encapsulations might be identified by a UDP port; others might 383 be identified by an Ethernet type or IP protocol number. Which 384 approach is used is a function of the preceding header the same way 385 as IPv4 is identified by both an Ethernet type and an IP protocol 386 number (for IP-in-IP). In some cases the header type is implicit in 387 some session (L2TP) or path (MPLS) setup. But this is largely beyond 388 the control of the encapsulation protocol. For instance, if there is 389 a requirement to carry the encapsulation after an Ethernet header, 390 then an Ethernet type is needed. If required to be carried after an 391 IP/UDP header, then a UDP port number is needed. For UDP port 392 numbers there are considerations for port number conservation 393 described in [I-D.ietf-tsvwg-port-use]. 395 It is worth mentioning that in the MPLS case of no implicit protocol 396 type many forwarding devices peek at the first nibble of the payload 397 to determine whether to apply IPv4 or IPv6 L3/L4 hashes for load 398 balancing [RFC7325]. That behavior places some constraints on other 399 payloads carried over MPLS and some protocol define an initial 400 control word in the payload with a value of zero in its first nibble 401 [RFC4385] to avoid confusion with IPv4 and IPv6 payload headers. 403 Secondly, the encapsulation needs to indicate the type of its 404 payload, which is in scope for the design of the encapsulation. We 405 have existing protocols which use Ethernet types (such as GRE). Here 406 each encapsulation header can potentially makes its own choices 407 between: 408 o Reuse Ethernet types - makes it easy to carry existing L2 and L3 409 protocols including IPv6, IPv6, and Ethernet. Disadvantages are 410 that it is a 16 bit number and we probably need far less than 100 411 values, and the number space is controlled by the IEEE 802 RAC 412 with its own allocation policies. 413 o Reuse IP protocol numbers - makes it easy to carry e.g., ESP in 414 addition to IP and Etnernet but brings in all existing protocol 415 numbers many of which would never be used directly on top of the 416 encapsulation protocol. IANA managed eight bit values, presumably 417 more difficult to get an assigned number than to get a transport 418 port assignment. 419 o Define their own next-protocol number space, which can use fewer 420 bits than an Ethernet type and give more flexibility, but at the 421 cost of administering that numbering space (presumably by the 422 IANA). 424 Thirdly, if the IETF ends up defining multiple encapsulations at 425 about the same time, and there is some chance that multiple such 426 encapsulations can be combined in the same packet, there is a 427 question whether it makes sense to use a common approach and 428 numbering space for the encapsulation across the different protocols. 429 A common approach might not be beneficial as long as there is only 430 one way to indicate e.g., SFC inside NVO3. 432 Many Internet protocols use fixed values (typically managed by the 433 IANA function) for their next-protocol field. That facilitates 434 interpretation of packets by middleboxes and e.g., for debugging 435 purposes, but might make the protocol evolution inflexible. Our 436 collective experience with MPLS shows an alternative where the label 437 can be viewed as an index to a table containing processing 438 instructions and the table content can be managed in different ways. 439 Encapsulations might want to consider the tradeoffs between such more 440 flexible versus more fixed approaches. 442 In summary: 443 o Would it be useful for the IETF come up with a common scheme for 444 encapsulation protocols? If not each encapsulation can define its 445 own scheme. 447 9. MTU and Fragmentation 449 A common approach today is to assume that the underlay have 450 sufficient MTU to carry the encapsulated packets without any 451 fragmentation and reassembly at the tunnel endpoints. That is 452 sufficient when the operator of the ingress and egress have full 453 control of the paths between those endpoints. And it makes for 454 simpler (hardware) implementations if fragmentation and reassembly 455 can be avoided. 457 However, even under that assumption it would be beneficial to be able 458 to detect when there is some misconfiguration causing packets to be 459 dropped due to MTU issues. One way to do this is to have the 460 encapsulator set the don't-fragment (DF) flag in the outer IPv4 461 header and receive and log any received ICMP "packet too big" (PTB) 462 errors. Note that no flag needs to be set in an outer IPv6 header 463 [RFC2460]. 465 Encapsulations could also define an optional tunnel fragmentation and 466 reassembly mechanism which would be useful in the case when the 467 operator doesn't have full control of the path, or when the protocol 468 gets deployed outside of its original intended context. Such a 469 mechanism would be required if the underlay might have a path MTU 470 which makes it impossible to carry at least 1518 bytes (if offering 471 Ethernet service), or at least 1280 (if offering IPv6 service). The 472 use of such a protocol mechanism could be triggered by receiving a 473 PTB. But such a mechanism might not be implemented by all 474 encapsulators and decapsulators. [Aerolink is one example of such a 475 protocol.] 477 Depending on the payload carried by the encapsulation there are some 478 additional possibilities: 480 o If payload is IPv4/6 then the underlay path MTU could be used to 481 report end-to-end path MTU. 482 o If the payload service is Ethernet/L2, then there is no such per 483 destination reporting mechanism. However, there is a LLDP TLV for 484 reporting max frame size; might be useful to report minimum to end 485 stations, but unmodified end stations would do nothing with that 486 TLV since they assume that the MTU is at least 1518. 488 In summary: 489 o In some deployments an encapsulation can assume well-managed MTU 490 hence no need for fragmentation and reassembly related to the 491 encapsulation. 492 o Even so, it makes sense for ingress to track any ICMP packet too 493 big addressed to ingress to be able to log any MTU 494 misconfigurations. 495 o Should an encapsulation protocol be depoyed outside of the 496 original context it might very well need support for fragmentation 497 and reassembly. 499 10. OAM 501 The OAM area is seeing active development in the IETF with 502 discussions (at least) in NVO3 and SFC working groups, plus the new 503 LIME WG looking at architecture and YANG models. 505 The design team has take a narrow view of OAM to explore the 506 potential OAM implications on the encapsulation format. 508 In terms of what we have heard from the various working groups there 509 seem to be needs to: 510 o Be able to send out-of-band OAM messages - that potentially should 511 follow the same path through the network as some flow of data 512 packets. 513 * Such OAM messages should not accidentally be decapsulated and 514 forwarded to the end stations. 515 * Be able to add OAM information to data packets that are 516 encapsulated. Discussions have been around 517 * Using a bit in the OAM to synchronize sampling of counters 518 between the encapsulator and decapsulator. 519 * Optional timestamps, sequence numbers, etc for more detailed 520 measurements between encapsulator and decapsulator. 521 o Usable for both proactive monitoring (akin to BFD) and reactive 522 checks (akin to traceroute to pin-point a failure) 524 To ensure that the OAM messages can follow the same path the OAM 525 messages need to get the same ECMP (and LAG hashing) results as a 526 given data flow. An encapsulator can choose between one of: 528 o Limit ECMP hashing to not look past the UDP header i.e. the 529 entropy needs to be in the source/destination IP and UDP ports 530 o Make OAM packets look the same as data packets i.e. the initial 531 part of the OAM payload has the inner Ethernet, IP, TCP/UDP 532 headers as a payload. (This approach was taken in TRILL out of 533 necessity since there is no UDP header.) Any OAM bit in the 534 encapsulation header must in any case be excluded from the 535 entropy. 537 There can be several ways to prevent OAM packets from accidentally 538 being forwarded to the end station using: 539 o A bit in the frame (as in TRILL) indicating OAM 540 o A next-protocol indication with a designated value for "none" or 541 "oam". 542 This assumes that the bit or next protocol, respectively, would not 543 affect entropy/ECMP in the underlay. However, the next-protocol 544 field might be used to provide differentiated treatement of packets 545 based on their payload; for instance a TCP vs. IPsec ESP payload 546 might be handled differently. Based on that observation it might be 547 undesirable to overload the next protocol with the OAM drop behavior, 548 resulting in a preference for having a bit to indicate that the 549 packet should be forwarded to the end station after decapsulation. 551 There has been suggestions that one (or more) marker bits in the 552 encaps header would be useful in order to delineate measurement 553 epochs on the encapsulator and decapsulator and use that to compare 554 counters to determine packet loss. 556 A result of the above is that OAM is likely to evolve and needs some 557 degree of extensibility from the encapsulation format; a bit or two 558 plus the ability to define additional larger extensions. 560 An open question is how to handle error messages or other reports 561 relating to OAM. One can think if such reporting as being associated 562 with the encapsulation the same way ICMP is associated with IP. 563 Would it make sense for the IETF to develop a common Encapsulation 564 Error Reporting Protocol as part of OAM, which can be used for 565 different encapsulations? And if so, what are the technical 566 challenges. For instance, how to avoid it being filtered as ICMP 567 often is? 569 A potential additional consideration for OAM is the possible future 570 existence of gateways that "stitch" together different dataplane 571 encapsulations and might want to carry OAM end-to-end across the 572 different encapsulations. 574 In summary: 576 o It makes sense to reserve a bit for "drop after decapsulation" for 577 OAM out-of-band. 578 o An encapsulation needs sufficient extensibility for OAM (such as 579 bits, timestamps, sequence numbers). That might be motivated by 580 in-band OAM but it would make sense to leverage the same 581 extensions for out-of band OAM. 582 o OAM places some constraints on use of entropy in forwarding 583 devices. 584 o Should IETF look into error reporting that is independent of the 585 specific encapsulation? 587 11. Security Considerations 589 Different encapsulation use cases will have different requirements 590 around security. For instance, when encapsulation is used to build 591 overlay networks for network virtualization, isolation between 592 virtual networks may be paramount. BIER support of multicast may 593 entail different security requirements than encapsulation for 594 unicast. 596 In real deployment, the security of the underlying network may be 597 considered for determining the level of security needed in the 598 encapsulation layer. However for the purposes of this discussion, we 599 assume that network security is out of scope and that the underlying 600 network does not itself provide adequate or as least uniform security 601 mechanisms for encapsulation. 603 There are at least three considerations for security: 604 o Anti-spoofing/virtual network isolation 605 o Interaction with packet level security such as IPsec or DTLS 606 o Privacy (e.g., VNI ID confidentially for NVO3) 608 This section uses a VNI ID in NVO3 as an example. A SFC or BIER 609 encapsulation is likely to have fields with similar security and 610 privacy requirements. 612 11.1. Encapsulation-specific considerations 614 Some of these considerations appear for a new encapsulation, and 615 others are more specific to network virtualization in datacenters. 616 o New attack vectors: 617 * DDOS on specific queued/paths by attempting to reproduce the 618 5-tuple hash for targeted connections. 619 * Entropy in outer 5-tuple may be too little or predictable. 620 * Leakage of identifying information in the encapsulation header 621 for an encrypted payload. 623 * Vulnerabilities of using global values in fields like VNI ID. 624 o Trusted versus untrusted tenants in network virtualization: 625 * The criticality of virtual network isolation depends on whether 626 tenants are trusted or untrusted. In the most extreme cases, 627 tenants might not only be untrusted but may be considered 628 hostile. 629 * For a trusted set of users (e.g. a private cloud) it may be 630 sufficient to have just a virtual network identifier to provide 631 isolation. Packets inadvertently crossing virtual networks 632 should be dropped similar to a TCP packet with a corrupted port 633 being received on the wrong connection. 634 * In the presence of untrusted users (e.g. a public cloud) the 635 virtual network identifier must be adequately protected against 636 corruption and verified for integrity. This case may warrant 637 keyed integrity. 638 o Different forms of isolation: 639 * Isolation could be blocking all traffic between tenants (or 640 except as allowed by some firewall) 641 * Could also be about performance isolation i.e. one tenant can 642 overload the network in a way that affects other tenants 643 * Physical isolation of traffic for different tenants in network 644 may be required, as well as required restrictions that tenants 645 may have on where their packets may be routed. 646 o New attack vectors from untrusted tenants: 647 * Third party VMs with untrusted tenants allows internally borne 648 attacks within data centers 649 * Hostile VMs inside the system may exist (e.g. public cloud) 650 * Internally launched DDOS 651 * Passive snooping for mis-delivered packets 652 * Mitigate damage and detection in event that a VM is able to 653 circumvent isolation mechanisms 654 o Tenant-provider relationship: 655 * Tenant might not trust provider, hypervisors, network 656 * Provider likely will need to provide SLA or a least a statement 657 on security 658 * Tenant may implement their own additional layers of security 659 * Regulation and certification consuderations 660 o Trend towards tighter security: 661 * Tenants' data in network increases in volume and value, attacks 662 become more sophisticated 663 * Large DCs already encrypt everything on disk 664 * DCs likely to encrypt inter-DC traffic at this point, use TLS 665 to Internet. 666 * Encryption within DC is becoming more commonplace, becomes 667 ubiquitous when cost is low enough. 668 * Cost/performance considerations. Cost of support for strong 669 security has made strong network security in DCs prohibitive. 671 * Are there lessons from MacSec? 673 11.2. Virtual network isolation 675 The first requirement is isolation between virtual networks. Packets 676 sent in one virtual network should never be illegitimately received 677 by a node in another virtual network. Isolation should be protected 678 in the presence of malicious attacks or inadvertent packet 679 corruption. 681 The second requirement is sender authentication. Sender identity is 682 authenticated to prevent anti-spoofing. Even if an attacker has 683 access to the packets in the network, they cannot send packets into a 684 virtual network. This may have two possibilities: 685 o Pairwise sender authentication. Any two communicating hosts 686 negotiate a shared key. 687 o Group authentication. A group of hosts share a key (this may be 688 more appropriate for multicast of encapsulation). 690 Possible security solutions: 691 o Security cookie: This is similar to L2TP cookie mechanism 692 [RFC3931]. A shared plain text cookie is shared between 693 encapsulator and decapsulator. A receiver validates a packet by 694 evaluating if the cookie is correct for the virtual network and 695 address of a sender. Validation function is F(cookie, VNI ID, 696 source addr). If cookie matches, accept packet, else drop. Since 697 cookie is plain text this method does not protect against an 698 eavesdropping. Cookies are set and may be rotated out of band. 699 o Secure hash: This is a stronger mechanism than simple cookies that 700 borrows from IPsec and PPP authentication methods. In this model 701 security field contains a secure hash of some fields in the packet 702 using a shared key. Hash function may be something like H(key, 703 VNI ID, addrs, salt). The salt ensures the hash is not the same 704 for every packet, and if it includes a sequence number may also 705 protect against replay attacks. 707 In any use of a shared key, periodic re-keying should be allowed. 708 This could include use of techniques like generation numbers, key 709 windows, etc. See [I-D.farrelll-mpls-opportunistic-encrypt] for an 710 example application. 712 We might see firewalls that are aware of the encapsulation and can 713 provide some defense in depth combined with the above example anti- 714 spoofing approaches. An example would be an NVO3-aware firewall 715 being able to check the VNI ID. 717 Separately and in addition to such filtering, there might be a desire 718 to completely block an encapsulation protocol at certain places in 719 the network, e.g., at the edge of a datacenter. Using a fixed 720 standard UDP destination port number for each encapsulation protocol 721 would facilitate such blocking. 723 11.3. Packet level security 725 An encapsulated packet may itself be encapsulated in IPsec (e.g. 726 ESP). This should be straightforward and in fact is what would 727 happen today in security gateways. In this case, there is no special 728 consideration for the fact that packet is encapsulated, however since 729 the encapsulation layer headers are included (part of encrypted data 730 for instance) we lose visibility in the network of the encapsulation. 732 The more interesting case is when security is applied to the 733 encapsulation payload. This will keep the encapsulation headers in 734 the outer header visible to the network (for instance in nvo3 we may 735 way to firewall based on VNI ID even if the payload is encrypted). 736 One possibility is to apply DTLS to the encapsulation payload. In 737 this model the protocol stack may be something like IP|UDP|Encap| 738 DTLS|encrypted_payload. The encapsulation and security should be 739 done together at an encapsulator and resolved at the decapsulator. 740 Since the encapsulation header is outside of the security coverage, 741 this may itself require security (like described above). 743 In both of the above the security associations (SAs) may be between 744 physical hosts, so for instance in nvo3 we can have packets of 745 different virtual networks using the same SA-- this should not be an 746 issue since it is the VNI ID that ensures isolation (which needs to 747 be secured also). 749 11.4. In summary: 751 o Encapsulations need extensibility mechanisms to be able to add 752 security features like cookies and secure hashes protecting the 753 encapsulation header. 754 o NVO3 proably has specific higher requirements relating to 755 isolation for network virtualization, which is in scope for the 756 NVO3 WG/ 757 o Our collective IETF experience is that succesful protocols get 758 deployed outside of the original intended context, hence the 759 initial assumptions about the threat model might become invalid. 760 That needs to be considered in the standardization of new 761 encapsulations. 763 12. QoS 765 In the Internet architecture we support QoS using the Differentiated 766 Services Code Points (DSCP) in the formerly named Type-of-Service 767 field in the IPv4 header, and in the Traffic-Class field in the IPv6 768 header. The ToS and TC fields also contain the two ECN bits. 770 We have existing specifications how to process those bits. See 771 [RFC2983] for diffserv handling, which specifies how the received 772 DSCP value is used to set the DSCP value in an outer IP header when 773 encapsulating. (There are also existing specifications how DSCP can 774 be mapped to layer2 priorities.) 776 Those specifications apply whether or not there is some intervening 777 headers (e.g., for NVO3 or SFC) between the inner and outer IP 778 headers. Thus the encapsulation considerations in this area are 779 mainly about applying the framework in [RFC2983]. 781 Note that the DSCP and ECN bits are not the only part of an inner 782 packet that might potentially affect the outer packet. For example, 783 [RFC2473] specifies handling of inner IPv6 hop-by-hop options that 784 effectively result in copying some options to the outer header. It 785 is simpler to not have future encapsulations depend on such copying 786 behavior. 788 There are some other considerations specific to doing OAM for 789 encapsulations. If OAM messages are used to measure latency, it 790 would make sense to treat them the same as data payloads. Thus they 791 need to have the same outer DSCP value as the data packets which they 792 wish to measure. 794 Due to OAM there are constraints on middleboxes in general. If 795 middleboxes inspect the packet past the outer IP+UDP and 796 encapsulation header and look for inner IP and TCP/UDP headers, that 797 might violate the assumption that OAM packets will be handled the 798 same as regular data packets. That issue is broader than just QoS - 799 applies to firewall filters etc. 801 In summary: 802 o Leverage the existing approach in [RFC2983] for DSCP handling. 804 13. Congestion Considerations 806 Additional encapsulation headers does not introduce anything new for 807 Explicit Congestion Notification. It is just like IP-in-IP and IPsec 808 tunnels which is specified in [RFC6040] in terms of how the ECN bits 809 in the inner and outer header are handled when encapsulating and 810 decapsulating packets. Thus new encapsulations can more or less 811 include that by reference. 813 There are additional considerations around carrying non-congestion 814 controlled traffic. These details have been worked out in 815 [I-D.ietf-mpls-in-udp]. As specified in [RFC5405]: "IP-based traffic 816 is generally assumed to be congestion-controlled, i.e., it is assumed 817 that the transport protocols generating IP-based traffic at the 818 sender already employ mechanisms that are sufficient to address 819 congestion on the path Consequently, a tunnel carrying IP-based 820 traffic should already interact appropriately with other traffic 821 sharing the path, and specific congestion control mechanisms for the 822 tunnel are not necessary". Those considerations are being captured 823 in [I-D.ietf-tsvwg-rfc5405bis]. 825 For this reason, where an encapsulation method is used to carry IP 826 traffic that is known to be congestion controlled, the UDP tunnels 827 does not create an additional need for congestion control. Internet 828 IP traffic is generally assumed to be congestion-controlled. 829 Similarly, in general Layer 3 VPNs are carrying IP traffic that is 830 similarly assumed to be congestion controlled. 832 However, some of the encapsulations (at least NVO3) will be able to 833 carry arbitrary Layer 2 packets to provide an L2 service, in which 834 case one can not assume that the traffic is congestion controlled. 836 One could handle this by adding some congestion control support to 837 the encapsulation header (one instance of which would end up looking 838 like DCCP). However, if the underlay is well-provisioned and managed 839 as opposed to being arbitrary Internet path, it might be sufficient 840 to have a slower reaction to congestion induced by that traffic. 841 There is work underway on a notion of "circuit breakers" for this 842 purpose. See See [I-D.ietf-tsvwg-circuit-breaker]. Encapsulations 843 which carry arbitrary Layer 2 packets want to consider that ongoing 844 work. 846 If the underlay is provisioned in such a way that it can guarantee 847 sufficient capacity for non-congestion controlled Layer 2 traffic, 848 then such circuit breakers might not be needed. 850 Two other considerations appear in the context of these 851 encapsulations as applied to overlay networks: 852 o Protect against malicious end stations 853 o Ensure fairness and/or measure resource usage across multiple 854 tenants 855 Those issues are really orthogonal to the encapsulation, in that they 856 are present even when no new encapsulation header is in use. 857 However, the application of the new encapsulations are likely to be 858 in environments where those issues are becoming more important. 859 Hence it makes sense to consider them. 861 One could make the encapsulation header be extensible to that it can 862 carry sufficient information to be able to measure resource usage, 863 delays, and congestion. The suggestions in the OAM section about a 864 single bit for counter synchronization, and optional timestamps 865 and/or sequence numbers, could be part of such an approach. There 866 might also be additional congestion-control extensions to be carried 867 in the encapsulation. Overall this results in a consideration to be 868 able to have sufficient extensibility in the encapsulation to be 869 handle to handle potential future developments in this space. 871 Coarse measurements are likely to suffice, at least for circuit- 872 breaker-like purposes, see [I-D.wei-tsvwg-tunnel-congestion-feedback] 873 and [I-D.briscoe-conex-data-centre] for examples on active work in 874 this area via use of ECN. [RFC6040] Appendix C is also relevant. 875 The outer ECN bits seem sufficient (at least when everything uses 876 ECN) to do this course measurements. Needs some more study for the 877 case when there are also drops; might need to exchange counters 878 between ingress and egress to handle drops. 880 Circuit breakers are not sufficient to make a network with different 881 congestion control when the goal is to provide a predictable service 882 to different tenants. The fallback would be to rate limit different 883 traffic. 885 In summary: 886 o Leverage the existing approach in [RFC6040] for ECN handling. 887 o If the encapsulation can carry non-IP, hence non-congestion 888 controlled traffic, then leverage the approach in 889 [I-D.ietf-mpls-in-udp]. 890 o "Watch this space" for circuit breakers. 892 14. Header Protection 894 Many UDP based encapsulations such as VXLAN [RFC7348] either 895 discourage or explicitly disallow the use of UDP checksums. The 896 reason is that the UDP checksum covers the entire payload of the 897 packet and switching ASICs are typically optimized to look at only a 898 small set of headers as the packet passes through the switch. In 899 these case, computing a checksum over the packet is very expensive. 900 (Software endpoints and the NICs used with them generally do not have 901 the same issue as they need to look at the entire packet anyways.) 903 The lack a header checksum creates the possibility that bit errors 904 can be introduced into any information carried by the new headers. 905 Specifically, in the case of IPv6, the assumption is that a transport 906 layer checksum - UDP in this case - will protect the IP addresses 907 through the inclusion of a pseudoheader in the calculation. This is 908 different from IPv4 on which many of these encapsulation protocols 909 are initially deployed which contains its own header checksum. In 910 addition to IP addresses, the encapsulation header often contains its 911 own information which is used for addressing packets or other high 912 value network functions. Without a checksum, this information is 913 potentially vulnerable - an issue regardless of whether the packet is 914 carried over IPv4 or IPv6. 916 Several protocols cite [RFC6935] and [RFC6936] as an exemption to the 917 IPv6 checksum requirements. However, these are intended to be 918 tailored to a fairly narrow set of circumstances - primarily relying 919 on sparseness of the address space to detect invalid values and well 920 managed networks - and are not a one size fits all solution. In 921 these cases, an analysis should be performed of the intended 922 environment, including the probability of errors being introduced and 923 the use of ECC memory in routing equipment. 925 Conceptually, the ideal solution to this problem is a checksum that 926 covers only the newly added headers of interest. There is little 927 value in the portion of the UDP checksum that covers the encapsulated 928 packet because that would generally be protected by other checksums 929 and this is the expensive portion to compute. In fact, this solution 930 already exists in the form of UDP-Lite and UDP based encapsulations 931 could be easily ported to run on top of it. Unfortunately, the main 932 value in using UDP as part of the encapsulation header is that it is 933 recognized by already deployed equipment for the purposes of ECMP, 934 RSS, and middlebox operations. As UDP-Lite uses a different protocol 935 number than UDP and it is not widely implemented in middleboxes, this 936 value is lost. A possible solution is to incorporate the same 937 partial-checksum concept as UDP-Lite or other header checksum 938 protection into the encapsulation header and continue using UDP as 939 the outer protocol. One potential challenge with this approach is 940 the use of NAT or other form of translation on the outer header will 941 result in an invalid checksum as the translator will not know to 942 update the encapsulation header. 944 The method chosen to protect headers is often related to the security 945 needs of the encapsulation mechanism. On one hand, the impact of a 946 poorly protected header is not limited to only data corruption but 947 can also introduce a security vulnerability in the form of 948 misdirected packets to an unauthorized recipient. Conversely, high 949 security protocols that already include a secure hash over the 950 valuable portion of the header (such as by encrypting the entire IP 951 packet using IPsec, or some secure hash of the encap header) do not 952 require additional checksum protection as the hash provides stronger 953 assurance than a simple checksum. 955 If the sender has included a checksum, then the receiver should 956 verify that checksum or, if incapable, drop the packet. The 957 assumption is that configuration and/or control-plane capability 958 exchanges can be used when different receiver have different checksum 959 validation capabilities. 961 In summary: 962 o Encapsulations need extensibility to be able to add checksum/CRC 963 for the encapsulation header itself. 964 o When the encapsulation has a checksum/CRC, include the IPv6 965 pseudo-header in it. 966 o The checksum/CRC can potentially be avoided when cryptographic 967 protection is applied to to the encapsulation. 969 15. Extensibility Considerations 971 Protocol extensibility is the concept that a networking protocol may 972 be extended to include new use cases or functionality that were not 973 part of the original protocol specification. Extensibility may be 974 used to add security, control, management, or performance features to 975 a protocol. A solution may allow private extensions for 976 customization or experimentation. 978 Extending a protocol often implies that a protocol header must carry 979 new information. There are two usual methods to accomplish this: 980 1. Define or redefine the meaning of existing fields in a protocol 981 header. 982 2. Add new (optional) fields to the protocol header. 983 It is also possible to create a new protocol version, but this is 984 more associated with defining a protocol than extending it (IPv6 985 being a successor to IPv4 is an example of protocol versioning). 987 In some cases it might be more appropriate to define a new inner 988 protocol which can carry the new functionality instead of extending 989 the outer protocol. Examples where this works well is in the IP/ 990 transport split, where the earlier architecture had a single NCP 991 protocol which carried both the hop-by-hop semantics which are now in 992 IP, and the end-to-end semantics which are now in TCP. Such a split 993 is effective when different nodes need to act upon the different 994 information. Applying this for general protocol extensibility 995 through nesting is not well understood, and does result in longer 996 header chains. Furthermore, our experience with IPv6 extension 997 headers [RFC2460] in middleboxes indicates that the approach does not 998 help with middlebox traversal. 1000 Many protocol definitions include some number of reserved fields or 1001 bits which can be used for future extension. VXLAN is an example of 1002 a protocol that includes reserved bits which are subsequently being 1003 allocated for new purposes. Another technique employed is to 1004 repurpose existing header fields with new meanings. A classic 1005 example of this is the definition of DSCP code point which redefines 1006 the ToS field originally specified in IPv4. When a field is 1007 redefined, some mechanism may be needed to ensure that all interested 1008 parties agree on the meaning of the field. The techniques of 1009 defining meaning for reserved bits or redefining existing fields have 1010 the advantage that a protocol header can be kept a fixed length. The 1011 disadvantage is that the extensibility is limited. For instance, the 1012 number reserved bits in a fixed protocol header is limited. For 1013 standard protocols the decision to commit to a definition for a field 1014 can be wrenching since it is difficult to retract later. Also, it is 1015 difficult to predict a priori how many reserved fields or bits to put 1016 into a protocol header to satisfy the extensions create over the 1017 lifetime of the protocol. 1019 Extending a protocol header with new fields can be done in several 1020 ways. 1021 o TLVs are a very popular method used in such protocols as IP and 1022 TCP. Depending on the type field size and structure, TLVs can 1023 offer a virtually unlimited range of extensions. A disadvantage 1024 of TLVs is that processing them can be verbose, quite complicated, 1025 several validations must often be done for each TLV, and there is 1026 no deterministic ordering for a list of TLVs. TCP serves as an 1027 example of a protocol where TLVs have been successfully used (i.e. 1028 required for protocol operation). IP is an example of a protocol 1029 that allows TLVs but are rarely used in practice (router fast 1030 paths usually that assume no IP options). Note that TCP TLVs are 1031 implemented in software as well as (NIC) hardware handling various 1032 forms of TCP offload. 1033 o Extension headers are closely related to TLVs. These also carry 1034 type/value information, but instead of being a list of TLVs within 1035 a single protocol header, each one is in its own protocol header. 1036 IPv6 extension headers and SFC NSH are examples of this technique. 1037 Similar to TLVs these offer a wide range of extensibility, but 1038 have similarly complex processing. Another difference with TLVs 1039 is that each extension header is idempotent. This is beneficial 1040 in cases where a protocol implements a push/pop model for header 1041 elements like service chaining, but makes it more difficult group 1042 correlated information within one protocol header. 1043 o A particular form of extension headers are the tags used by IEEE 1044 802 protocols. Those are similar to e.g., IPv6 extension headers 1045 but with the key difference that each tag is a fixed length header 1046 where the length is implicit in the tag value. Thus as long as a 1047 receiver can be programmed with a tag value to length map, it can 1048 skip those new tags. 1050 o Flag-fields are a non-TLV like method of extending a protocol 1051 header. The basic idea is that the header contains a set of 1052 flags, where each set flags corresponds to optional field that is 1053 present in the header. GRE is an example of a protocol that 1054 employs this mechanism. The fields are present in the header in 1055 the order of the flags, and the length of each field is fixed. 1056 Flag-fields are simpler to process compared to TLVs, having fewer 1057 validations and the order of the optional fields is deterministic. 1058 A disadvantage is that range of possible extensions with flag- 1059 fields is smaller than TLVs. 1061 The requirements for receiving unknown or unimplemented extensible 1062 elements in an encapsulation protocol (flags, TLVs, optional fields) 1063 need to be specified. There are two parties to consider, middle 1064 boxes and terminal endpoints of encapsulation (at the decapsulator). 1066 A protocol may allow or expect nodes in a path to modify fields in an 1067 encapsulation (example use of this is BIER). In this case, the 1068 middleboxes should follow the same requirements as nodes terminating 1069 the encapsulation. In the case that middle boxes do not modify the 1070 encapsulation, we can assume that they may still inspect any fields 1071 of the encapsulation. Missing or unknown fields should be accepted 1072 per protocol specification, however it is permissible for a site to 1073 implement a local policy otherwise (e.g. a firewall may drop packets 1074 with unknown options). 1076 For handling unknown options at terminal nodes, there are two 1077 possibilities: drop packet or accept while ignoring the unknown 1078 options. Many Internet protocols specify that reserved flags must be 1079 set to zero on transmission and ignored on reception. L2TP is 1080 example data protocol that has such flags. GRE is a notable 1081 exception to this rule, reserved flag bits 1-5 cannot be ignored 1082 [RFC2890]. For TCP and IPv4, implementations must ignore optional 1083 TLVs with unknown type; however in IPv6 if a packet contains an 1084 unknown extension header (unrecognized next header type) the packet 1085 must be dropped with an ICMP error message returned. The IPv6 1086 options themselves (encoded inside the destinations options or hop- 1087 by-hop options extension header) have more flexibility. There bits 1088 in the option code are used to instruct the receiver whether to 1089 ignore, silently drop, or drop and send error if the option is 1090 unknown. Some protocols define a "mandatory bit" that can is set 1091 with TLVs to indicate that an option must not be ignored. 1092 Conceptually, optional data elements can only be ignored if they are 1093 idempotent and do not alter how the rest of the packet is parsed or 1094 processed. 1096 Depending on what type of protocol evolution one can predict, it 1097 might make sense to have an way for a sender to express that the 1098 packet should be dropped by a terminal node which does not understand 1099 the new information. In other cases it would make sense to have the 1100 receiver silently ignore the new info. The former can be expressed 1101 by having a version field in the encapsulation, or a notion of 1102 "mandatory bit" as discussed above. 1104 A security mechanism which use some form secure hash over the 1105 encapsulation header would need to be able to know which extensions 1106 can be changed in flight. 1108 In summary: 1109 o Encapsulations need the ability to be extended to handle e.g., the 1110 OAM or security aspects discussed in this document. 1111 o Practical experience seems to tell us that extensibility 1112 mechanisms which are not in use on day one might result in 1113 immediate ossification by lack of implementation support. In some 1114 cases that has occurred in routers and in other cases in 1115 middleboxes. Hence devising ways where the extensibility 1116 mechanisms are in use seems important. 1118 16. Layering Considerations 1120 One can envision that SFC might use NVO3 as a delivery/transport 1121 mechanism. With more imagination that in turn might be delivered 1122 using BIER. Thus it is useful to think about what things look like 1123 when we have BIER+NVO3+SFC+payload. Also, if NVO3 is widely deployed 1124 there might be cases of NVO3 nesting where a customer uses NVO3 to 1125 provide network virtualization e.g., across departments. That 1126 customer uses a service provider which happens to use NVO3 to provide 1127 transport for their customers.Thus NVO3 in NVO3 might happen. 1129 A key question we set out to answer is what the packets might look 1130 like in such a case, and in particular whether we would end up with 1131 multiple UDP headers for entropy. 1133 Based on the discussion in the Entropy section, the entropy is 1134 associated with the outer delivery IP header. Thus if there are 1135 multiple IP headers there would be a UDP header for each one of the 1136 IP headers. But SFC does not require its own IP header. So a case 1137 of NVO3+SFC would be IP+UDP+NVO3+SFC. A nested NVO3 encapsulation 1138 would have independent IP+UDP headers. 1140 The layering also has some implications for middleboxes. 1141 o A device on the path between the ingress and egress is allowed to 1142 transparently inspect all layers of the protocol stack and drop or 1143 forward, but not transparently modify anything but the layer in 1144 which they operate. What this means is that an IP router is 1145 allowed modify the outer IP ttl and ECN bits, but not the 1146 encapsulation header or inner headers and payload. And a BIER 1147 router is allowed to modify the BIER header. 1148 o Alternatively such a device can become visible at a higher layer. 1149 E.g., a middlebox could become an decapsulate + function + 1150 encapsulate which means it will generate a new encapsulation 1151 header. 1153 The design team asked itself some additional questions: 1154 o Would it make sense to have a common encapsulation base header 1155 (for OAM, security?, etc) and then followed by the specific 1156 information for NVO3, SFC, BIER? Given that there are separate 1157 proposals and the set of information needing to be carried 1158 differs, and the extensibility needs might be different, it would 1159 be difficult and not that useful to have a common base header. 1160 o With a base header in place, one could view the different 1161 functions (NVO3, SFC, and BIER) as different extensions to that 1162 base header resulting in encodings which are more space optimal by 1163 not repeating the same base header. The base header would only be 1164 repeated when there is an additional IP (and hence UDP) header. 1165 That could mean a single length field (to skip to get to the 1166 payload after all the encapsulation headers). That might be 1167 technically feasible, but it would create a lot of dependencies 1168 between different WGs making it harder to make progress. Compare 1169 with the potential savings in packet size. 1171 17. Service model 1173 The IP service is lossy and subject to reordering. In order to avoid 1174 a performance impact on transports like TCP the handling of packets 1175 is designed to avoid reordering packets that are in the same 1176 transport flow (which is typically identified by the 5-tuple). But 1177 across such flows the receiver can see different ordering for a given 1178 sender. That is the case for a unicast vs. a multicast flow from the 1179 same sender. 1181 There is a general tussle between the desire for high capacity 1182 utilization across a multipath network and the import on packet 1183 ordering within the same flow (which results in lower transport 1184 protocol performance). That isn't affected by the introduction of an 1185 encapsulation. However, the encapsulation comes with some entropy, 1186 and there might be cases where folks want to change that in response 1187 to overload or failures. For instance, might want to change UDP 1188 source port to try different ECMP route. Such changes can result in 1189 packet reordering within a flow, hence would need to be done 1190 infrequently and with care e.g., by identifying packet trains. 1192 There might be some applications/services which are not able to 1193 handle reordering across flows. The IETF has defined pseudo-wires 1194 [RFC3985] which provides the ability to ensure ordering (implemented 1195 using sequence numbers and/or timestamps). 1197 Architectural such services would make sense, but as a separate layer 1198 on top of an encapsulation protocol. They could be deployed between 1199 ingress and egress of a tunnel which uses some encaps. Potentially 1200 the tunnel control points at the ingress and egress could become a 1201 platform for fixing suboptimal behavior elsewhere in the network. 1202 That would clearly be undesirable in the general case. However, 1203 handling encapsulation of non-IP traffic hence non-congestion- 1204 controlled traffic is likely to be required, which implies some 1205 fairness and/or QoS policing on the ingress and egress devices. 1207 But the tunnels could potentially do more like increase reliability 1208 (retransmissions, FEC) or load spreading using e.g. MP-TCP between 1209 ingress and egress. 1211 18. Hardware Friendly 1213 Hosts, switches and routers often leverage capabilities in the 1214 hardware to accelerate packet encapsulation, decapsulation and 1215 forwarding. 1217 Some design considerations in encapsulation that leverage these 1218 hardware capabilities may result in more efficiently packet 1219 processing and higher overall protocol throughput. 1221 While "hardware friendliness" can be viewed as unnecessary 1222 considerations for a design, part of the motivation for considering 1223 this is ease of deployment; being able to leverage existing NIC and 1224 switch chips for at least a useful subset of the functionality that 1225 the new encapsulation provides. The other part is the ease of 1226 implementing new NICs and switch/router chips that support the 1227 encapsulation at ever increasing line rates. 1229 [disclaimer] There are many different types of hardware in any given 1230 network, each maybe better at some tasks while worse at others. We 1231 would still recommend protocol designers to examine the specific 1232 hardware that are likely to be used in their networks and make 1233 decisions on a case by case basis. 1235 Some considerations are: 1236 o Keep the encap header small. Switches and routers usually only 1237 read the first small number of bytes into the fast memory for 1238 quick processing and easy manipulation. The bulk of the packets 1239 are usually stored in slow memory. A big encap header may not fit 1240 and additional read from the slow memory will hurt the overall 1241 performance and throughput. 1242 o Put important information at the beginning of the encapsulation 1243 header. The reasoning is similar as explained in the previous 1244 point. If important information are located at the beginning of 1245 the encapsulation header, the packet may be processed with smaller 1246 number of bytes to be read into the fast memory and improve 1247 performance. 1248 o Avoid full packet checksums in the encapsulation if possible. 1249 Encapsulations should instead consider adding their own checksum 1250 which covers the encapsulation header and any IPv6 pseudo-header. 1251 The motivation is that most of the switch/router hardware make 1252 switching/forwarding decisions by reading and examining only the 1253 first certain number of bytes in the packet. Most of the body of 1254 the packet do not need to be processed normally. If we are 1255 concerned of preventing packet to be misdelivered due to memory 1256 errors, consider only perform header checksums. Note that NIC 1257 chips can typically already do full packet checksums for TCP/UDP, 1258 while adding a header checksum might require adding some hardware 1259 support. 1260 o Place important information at fixed offset in the encapsulation 1261 header. Packet processing hardware may be capable of parallel 1262 processing. If important information can be found at fixed 1263 offset, different part of the encapsulation header may be 1264 processed by different hardware units in parallel (for example 1265 multiple table lookups may be launched in parallel). It is easier 1266 for hardware to handle optional information when the information, 1267 if present, can be found in ideally one place, but in general, in 1268 as few places as possible. That facilitates parallel processing. 1269 TLV encoding with unconstrained order typically does not have that 1270 property. 1271 o Limit the number of header combinations. In many cases the 1272 hardware can explore different combinations of headers in 1273 parallel, however there is some added cost for this. 1275 18.1. Considerations for NIC offload 1277 This section provides guidelines to provide support of common 1278 offloads for encapsulation in Network Interface Cards (NICs). 1279 Offload mechanisms are techniques that are implemented separately 1280 from the normal protocol implementation of a host networking stack 1281 and are intended to optimize or speed up protocol processing. 1282 Hardware offload is performed within a NIC device on behalf of a 1283 host. 1285 There are three basic offload techniques of interest: 1287 o Receive multi queue 1288 o Checksum offload 1289 o Segmentation offload 1291 18.1.1. Receive multi-queue 1293 Contemporary NICs support multiple receive descriptor queues (multi- 1294 queue). Multi-queue enables load balancing of network processing for 1295 a NIC across multiple CPUs. On packet reception, a NIC must select 1296 the appropriate queue for host processing. Receive Side Scaling 1297 (RSS) is a common method which uses the flow hash for a packet to 1298 index an indirection table where each entry stores a queue number. 1300 UDP encapsulation, where the source port is used for entropy, should 1301 be compatible with multi-queue NICs that support five-tuple hash 1302 calculation for UDP/IP packets as input to RSS. The source port 1303 ensures classification of the encapsulated flow even in the case that 1304 the outer source and destination addresses are the same for all flows 1305 (e.g. all flows are going over a single tunnel). 1307 18.1.2. Checksum offload 1309 Many NICs provide capabilities to calculate standard ones complement 1310 payload checksum for packets in transmit or receive. When using 1311 encapsulation over UDP there are at least two checksums that may be 1312 of interest: the encapsulated packet's transport checksum, and the 1313 UDP checksum in the outer header. 1315 18.1.2.1. Transmit checksum offload 1317 NICs may provide a protocol agnostic method to offload transmit 1318 checksum (NETIF_F_HW_CSUM in Linux parlance) that can be used with 1319 UDP encapsulation. In this method the host provides checksum related 1320 parameters in a transmit descriptor for a packet. These parameters 1321 include the starting offset of data to checksum, the length of data 1322 to checksum, and the offset in the packet where the computed checksum 1323 is to be written. The host initializes the checksum field to pseudo 1324 header checksum. In the case of encapsulated packet, the checksum 1325 for an encapsulated transport layer packet, a TCP packet for 1326 instance, can be offloaded by setting the appropriate checksum 1327 parameters. 1329 NICs typically can offload only one transmit checksum per packet, so 1330 simultaneously offloading both an inner transport packet's checksum 1331 and the outer UDP checksum is likely not possible. In this case 1332 setting UDP checksum to zero (per above discussion) and offloading 1333 the inner transport packet checksum might be acceptable. 1335 There is a proposal in [I-D.herbert-remotecsumoffload] to leverage 1336 NIC checksum offload when an encapsulator is co-resident with a host. 1338 18.1.2.2. Receive checksum offload 1340 Protocol encapsulation is compatible with NICs that perform a 1341 protocol agnostic receive checksum (CHECKSUM_COMPLETE in Linux 1342 parlance). In this technique, a NIC computes a ones complement 1343 checksum over all (or some predefined portion) of a packet. The 1344 computed value is provided to the host stack in the packet's receive 1345 descriptor. The host driver can use this checksum to "patch up" and 1346 validate any inner packet transport checksum, as well as the outer 1347 UDP checksum if it is non-zero. 1349 Many legacy NICs don't provide checksum-complete but instead provide 1350 an indication that a checksum has been verified (CHECKSUM_UNNECESSARY 1351 in Linux). Usually, such validation is only done for simple TCP/IP 1352 or UDP/IP packets. If a NIC indicates that a UDP checksum is valid, 1353 the checksum-complete value for the UDP packet is the "not" of the 1354 pseudo header checksum. In this way, checksum-unnecessary can be 1355 converted to checksum-complete. So if the NIC provides checksum- 1356 unnecessary for the outer UDP header in an encapsulation, checksum 1357 conversion can be done so that the checksum-complete value is derived 1358 and can be used by the stack to validate an checksums in the 1359 encapsulated packet. 1361 18.1.3. Segmentation offload 1363 Segmentation offload refers to techniques that attempt to reduce CPU 1364 utilization on hosts by having the transport layers of the stack 1365 operate on large packets. In transmit segmentation offload, a 1366 transport layer creates large packets greater than MTU size (Maximum 1367 Transmission Unit). It is only at much lower point in the stack, or 1368 possibly the NIC, that these large packets are broken up into MTU 1369 sized packet for transmission on the wire. Similarly, in receive 1370 segmentation offload, small packets are coalesced into large, greater 1371 than MTU size packets at a point low in the stack receive path or 1372 possibly in a device. The effect of segmentation offload is that the 1373 number of packets that need to be processed in various layers of the 1374 stack is reduced, and hence CPU utilization is reduced. 1376 18.1.3.1. Transmit Segmentation Offload 1378 Transmit Segmentation Offload (TSO) is a NIC feature where a host 1379 provides a large (larger than MTU size) TCP packet to the NIC, which 1380 in turn splits the packet into separate segments and transmits each 1381 one. This is useful to reduce CPU load on the host. 1383 The process of TSO can be generalized as: 1384 o Split the TCP payload into segments which allow packets with size 1385 less than or equal to MTU. 1386 o For each created segment: 1387 1. Replicate the TCP header and all preceding headers of the 1388 original packet. 1389 2. Set payload length fields in any headers to reflect the length 1390 of the segment. 1391 3. Set TCP sequence number to correctly reflect the offset of the 1392 TCP data in the stream. 1393 4. Recompute and set any checksums that either cover the payload 1394 of the packet or cover header which was changed by setting a 1395 payload length. 1397 Following this general process, TSO can be extended to support TCP 1398 encapsulation UDP. For each segment the Ethernet, outer IP, UDP 1399 header, encapsulation header, inner IP header if tunneling, and TCP 1400 headers are replicated. Any packet length header fields need to be 1401 set properly (including the length in the outer UDP header), and 1402 checksums need to be set correctly (including the outer UDP checksum 1403 if being used). 1405 To facilitate TSO with encapsulation it is recommended that optional 1406 fields should not contain values that must be updated on a per 1407 segment basis-- for example an encapsulation header should not 1408 include checksums, lengths, or sequence numbers that refer to the 1409 payload. If the encapsulation header does not contain such fields 1410 then the TSO engine only needs to copy the bits in the encapsulation 1411 header when creating each segment and does not need to parse the 1412 encapsulation header. 1414 18.1.3.2. Large Receive Offload 1416 Large Receive Offload (LRO) is a NIC feature where packets of a TCP 1417 connection are reassembled, or coalesced, in the NIC and delivered to 1418 the host as one large packet. This feature can reduce CPU 1419 utilization in the host. 1421 LRO requires significant protocol awareness to be implemented 1422 correctly and is difficult to generalize. Packets in the same flow 1423 need to be unambiguously identified. In the presence of tunnels or 1424 network virtualization, this may require more than a five-tuple match 1425 (for instance packets for flows in two different virtual networks may 1426 have identical five-tuples). Additionally, a NIC needs to perform 1427 validation over packets that are being coalesced, and needs to 1428 fabricate a single meaningful header from all the coalesced packets. 1430 The conservative approach to supporting LRO for encapsulation would 1431 be to assign packets to the same flow only if they have identical 1432 five-tuple and were encapsulated the same way. That is the outer IP 1433 addresses, the outer UDP ports, encapsulated protocol, encapsulation 1434 headers, and inner five tuple are all identical. 1436 18.1.3.3. In summary: 1438 In summary, for NIC offload: 1439 o The considerations for using full UDP checksums are different for 1440 NIC offload than for implementations in forwarding devices like 1441 routers and switches. 1442 o Be judicious about encapsulations that change fields on a per- 1443 packet basis, since such behavior might make it hard to use TSO. 1445 19. Middlebox Considerations 1447 This document has touched upon middleboxes in different section. The 1448 reason for this is as encapsulations get widely deployed one would 1449 expect different forms of middleboxes might become aware of the 1450 encapsulation protocol just as middleboxes have been made aware of 1451 other protocols where there are business and deployment 1452 opportunities. Such middleboxes are likely to do more than just drop 1453 packets based on the UDP port number used by an encapsulation 1454 protocol. 1456 We note that various forms of encapsulation gateways that stitch one 1457 encapsulation protocol together with another form of protocol could 1458 have similar effects. 1460 An example of a middlebox that could see some use would be an NVO3- 1461 aware firewall that would filter on the VNI IDs to provide some 1462 defense in depth inside or across NVO3 datacenters. 1464 A question for the IETF is whether we should document what to do or 1465 what not to do in such middleboxes. This document touches on areas 1466 of OAM and ECMP as it relates to middleboxes and it might make sense 1467 to document how encapsulation-aware middleboxes should avoid 1468 unintended consequences in those (and perhaps other) areas. 1470 In summary: 1471 o We are likely to see middleboxes that at least parse the headers 1472 for succesful new encapsulations. 1473 o Should the IETF document considerations for what not to do in such 1474 middleboxes? 1476 20. Related Work 1478 The IETF and industry has defined encapsulations for a long time, 1479 with examples like GRE [RFC2890], VXLAN [RFC7348], and NVGRE 1480 [I-D.sridharan-virtualization-nvgre] being able to carry arbitrary 1481 Ethernet payloads, and various forms of IP-in-IP and IPsec 1482 encapsulations that can carry IP packets. As part of NVO3 there has 1483 been additional proposals like Geneve [I-D.gross-geneve] and GUE 1484 [I-D.herbert-gue] which look at more extensibility. NSH 1485 [I-D.quinn-sfc-nsh] is an example of an encapsulation that tries to 1486 provide extensibility mechanisms which target both hardware and 1487 software implementations. 1489 There is also a large body of work around MPLS encapsulations 1490 [RFC3032]. The MPLS-in-UDP work [I-D.ietf-mpls-in-udp] and GRE over 1491 UDP [I-D.ietf-tsvwg-gre-in-udp-encap] have worked on some of the 1492 common issues around checksum and congestion control. MPLS also 1493 introduced a entropy label [RFC6790]. There is also a proposal for 1494 MPLS encryption [I-D.farrelll-mpls-opportunistic-encrypt]. 1496 The idea to use a UDP encapsulation with a UDP source port for 1497 entropy for the underlay routers' ECMP dates back to LISP [RFC6830]. 1499 The pseudo-wire work [RFC3985] is interesting in the notion of 1500 layering additional services/characteristics such as ordered delivery 1501 or timely deliver on top of an encapsulation. That layering approach 1502 might be useful for the new encapsulations as well. For instance, 1503 the control word [RFC4385]. There is also material on congestion 1504 control for pseudo-wires in [I-D.ietf-pwe3-congcons]. 1506 Both MPLS and L2TP [RFC3931] rely on some control or signaling to 1507 establish state (for the path/labels in the case of MPLS, and for the 1508 session in the case of L2TP). The NVO3, SFC, and BIER encapsulations 1509 will also have some separation between the data plane and control 1510 plane, but the type of separation appears to be different. 1512 IEEE 802.1 has defined encapsulations for L2 over L2, in the form of 1513 Provider backbone Bridging (PBB) [IEEE802.1Q-2014] and Equal Cost 1514 Multipath (ECMP) [IEEE802.1Q-2014]. The latter includes something 1515 very similar to the way the UDP source port is used as entropy: "The 1516 flow hash, carried in an F-TAG, serves to distinguish frames 1517 belonging to different flows and can be used in the forwarding 1518 process to distribute frames over equal cost paths" 1520 TRILL, which is also a L2 over L2 encapsulation, took a different 1521 approach to entropy but preserved the ability for OAM frames 1522 [RFC7174] to use the same entropy hence ECMP path as data frames. In 1523 [I-D.ietf-trill-oam-fm] there 96 bytes of headers for entropy in the 1524 OAM frames, followed by the actual OAM content. This ensures that 1525 any headers, which fit in those 96 bytes except the OAM bit in the 1526 TRILL header, can be used for ECMP hashing. 1528 As encapsulations evolve there might be a desire to fit multiple 1529 inner packets into one outer packet. The work in 1530 [I-D.saldana-tsvwg-simplemux] might be interesting for that purpose. 1532 21. Acknowledgements 1534 The authors acknowledge the comments from Alia Atlas, Fred Baker, 1535 David Black, Bob Briscoe, Stewart Bryant, Mike Cox, Andy Malis, Radia 1536 Perlman, Michael Smith, and Lucy Yong. 1538 22. Open Issues 1540 o Middleboxes: 1541 * Due to OAM there are constraints on middleboxes in general. If 1542 middleboxes inspect the packet past the outer IP+UDP and 1543 encapsulation header and look for inner IP and TCP/UDP headers, 1544 that might violate the assumption that OAM packets will be 1545 handled the same as regular data packets. That issue is 1546 broader than just QoS - applies to firewall filters etc. 1547 * Firewalls looking at inner payload? How does that work for OAM 1548 frames? Even if it only drops ... TRILL approach might be an 1549 option? Would that encourage more middleboxes making the 1550 network more fragile? 1551 * Editorially perhaps we should pull the above two into a 1552 separate section about middlebox considerations? 1553 o Next-protocol indication - should it be common across different 1554 encapsulation headers? We will have different ways to indicate 1555 the presence of the first encapsulation header in a packet (could 1556 be a UDP destination port, an Ethernet type, etc depending on the 1557 outer delivery header). But for the next protocol past an 1558 encapsulation header one could envision creating or adoption a 1559 common scheme. Such a would also need to be able to identify 1560 following headers like Ethernet, IPv4/IPv6, ESP, etc. 1561 o Common OAM error reporting protocol? 1562 o There is discussion about timestamps, sequence numbers, etc in 1563 three different parts of the document. OAM, Congestion 1564 Considerations, and Service Model, where the latter argues that a 1565 pseudo-wire service should really be layered on top of the 1566 encapsulation using its own header. Those recommendations seem to 1567 be at odds with each other. Do we envision sequence numbers, 1568 timestamps, etc as potential extensions for OAM and CC? If so, 1569 those extensions could be used to provide a service which doesn't 1570 reorder packets. 1572 23. Change Log 1574 The changes from draft-rtg-dt-encap-01 based on feedback at the 1575 Dallas IETF meeting: 1576 o Setting the context that not all common issues might apply to all 1577 encapsulations, but that they should all be understood before 1578 being dismissed. 1579 o Clarified that IPv6 flow label is useful for entropy in 1580 combination with a UDP source port. 1581 o Editorially added a "summary" set of bullets to most sections. 1582 o Editorial clarifications in the next protocol section to more 1583 clearly state the three areas. 1584 o Folded the two next protocol sections into one. 1585 o Mention the MPLS first nibble issue in the next protocol section. 1586 o Mention that viewing the next protocol as an index to a table with 1587 processing instructions can provide additional flexibility in the 1588 protocol evolution. 1589 o For the OAM "don't forward to end stations" added that defining a 1590 bit seems better than using a special next-protocol value. 1591 o Added mention of DTLS in addition to IPsec for security. 1592 o Added some mention of IPv6 hob-by-hop options of other headers 1593 than potentially can be copied from inner to outer header. 1594 o Added text on architectural considerations when it might make 1595 sense to define an additional header/protocol as opposed to using 1596 the extensibility mechanism in the existing encapsulation 1597 protocol. 1598 o Clarified the "unconstrained TLVs" in the hardware friendly 1599 section. 1600 o Clarified the text around checksum verification and full vs. 1601 header checksums. 1602 o Added wording that the considerations might apply for encaps 1603 outside of the routing area. 1604 o Added references to draft-ietf-pwe3-congcons, 1605 draft-ietf-tsvwg-rfc5405bis, RFC2473, and RFC7325 1606 o Removed reference to RFC3948. 1607 o Updated the acknowledgements section. 1608 o Added this change log section. 1610 24. References 1612 24.1. Normative References 1614 [I-D.ietf-tsvwg-rfc5405bis] 1615 Eggert, L., Fairhurst, G., and G. Shepherd, "UDP Usage 1616 Guidelines", draft-ietf-tsvwg-rfc5405bis-02 (work in 1617 progress), April 2015. 1619 [RFC2460] Deering, S. and R. Hinden, "Internet Protocol, Version 6 1620 (IPv6) Specification", RFC 2460, December 1998. 1622 [RFC2473] Conta, A. and S. Deering, "Generic Packet Tunneling in 1623 IPv6 Specification", RFC 2473, December 1998. 1625 [RFC2890] Dommety, G., "Key and Sequence Number Extensions to GRE", 1626 RFC 2890, September 2000. 1628 [RFC2983] Black, D., "Differentiated Services and Tunnels", 1629 RFC 2983, October 2000. 1631 [RFC3032] Rosen, E., Tappan, D., Fedorkow, G., Rekhter, Y., 1632 Farinacci, D., Li, T., and A. Conta, "MPLS Label Stack 1633 Encoding", RFC 3032, January 2001. 1635 [RFC3931] Lau, J., Townsley, M., and I. Goyret, "Layer Two Tunneling 1636 Protocol - Version 3 (L2TPv3)", RFC 3931, March 2005. 1638 [RFC3985] Bryant, S. and P. Pate, "Pseudo Wire Emulation Edge-to- 1639 Edge (PWE3) Architecture", RFC 3985, March 2005. 1641 [RFC4385] Bryant, S., Swallow, G., Martini, L., and D. McPherson, 1642 "Pseudowire Emulation Edge-to-Edge (PWE3) Control Word for 1643 Use over an MPLS PSN", RFC 4385, February 2006. 1645 [RFC5405] Eggert, L. and G. Fairhurst, "Unicast UDP Usage Guidelines 1646 for Application Designers", BCP 145, RFC 5405, 1647 November 2008. 1649 [RFC6040] Briscoe, B., "Tunnelling of Explicit Congestion 1650 Notification", RFC 6040, November 2010. 1652 [RFC6790] Kompella, K., Drake, J., Amante, S., Henderickx, W., and 1653 L. Yong, "The Use of Entropy Labels in MPLS Forwarding", 1654 RFC 6790, November 2012. 1656 [RFC6830] Farinacci, D., Fuller, V., Meyer, D., and D. Lewis, "The 1657 Locator/ID Separation Protocol (LISP)", RFC 6830, 1658 January 2013. 1660 [RFC6935] Eubanks, M., Chimento, P., and M. Westerlund, "IPv6 and 1661 UDP Checksums for Tunneled Packets", RFC 6935, April 2013. 1663 [RFC6936] Fairhurst, G. and M. Westerlund, "Applicability Statement 1664 for the Use of IPv6 UDP Datagrams with Zero Checksums", 1665 RFC 6936, April 2013. 1667 [RFC7174] Salam, S., Senevirathne, T., Aldrin, S., and D. Eastlake, 1668 "Transparent Interconnection of Lots of Links (TRILL) 1669 Operations, Administration, and Maintenance (OAM) 1670 Framework", RFC 7174, May 2014. 1672 [RFC7325] Villamizar, C., Kompella, K., Amante, S., Malis, A., and 1673 C. Pignataro, "MPLS Forwarding Compliance and Performance 1674 Requirements", RFC 7325, August 2014. 1676 [RFC7348] Mahalingam, M., Dutt, D., Duda, K., Agarwal, P., Kreeger, 1677 L., Sridhar, T., Bursell, M., and C. Wright, "Virtual 1678 eXtensible Local Area Network (VXLAN): A Framework for 1679 Overlaying Virtualized Layer 2 Networks over Layer 3 1680 Networks", RFC 7348, August 2014. 1682 [RFC7364] Narten, T., Gray, E., Black, D., Fang, L., Kreeger, L., 1683 and M. Napierala, "Problem Statement: Overlays for Network 1684 Virtualization", RFC 7364, October 2014. 1686 [RFC7365] Lasserre, M., Balus, F., Morin, T., Bitar, N., and Y. 1687 Rekhter, "Framework for Data Center (DC) Network 1688 Virtualization", RFC 7365, October 2014. 1690 24.2. Informative References 1692 [I-D.briscoe-conex-data-centre] 1693 Briscoe, B. and M. Sridharan, "Network Performance 1694 Isolation in Data Centres using Congestion Policing", 1695 draft-briscoe-conex-data-centre-02 (work in progress), 1696 February 2014. 1698 [I-D.farrelll-mpls-opportunistic-encrypt] 1699 Farrel, A. and S. Farrell, "Opportunistic Security in MPLS 1700 Networks", draft-farrelll-mpls-opportunistic-encrypt-04 1701 (work in progress), January 2015. 1703 [I-D.gross-geneve] 1704 Gross, J., Sridhar, T., Garg, P., Wright, C., Ganga, I., 1705 Agarwal, P., Duda, K., Dutt, D., and J. Hudson, "Geneve: 1706 Generic Network Virtualization Encapsulation", 1707 draft-gross-geneve-02 (work in progress), October 2014. 1709 [I-D.herbert-gue] 1710 Herbert, T., Yong, L., and O. Zia, "Generic UDP 1711 Encapsulation", draft-herbert-gue-03 (work in progress), 1712 March 2015. 1714 [I-D.herbert-remotecsumoffload] 1715 Herbert, T., "Remote checksum offload for encapsulation", 1716 draft-herbert-remotecsumoffload-01 (work in progress), 1717 November 2014. 1719 [I-D.ietf-mpls-in-udp] 1720 Xu, X., Sheth, N., Yong, L., Callon, R., and D. Black, 1721 "Encapsulating MPLS in UDP", draft-ietf-mpls-in-udp-11 1722 (work in progress), January 2015. 1724 [I-D.ietf-nvo3-arch] 1725 Black, D., Hudson, J., Kreeger, L., Lasserre, M., and T. 1726 Narten, "An Architecture for Overlay Networks (NVO3)", 1727 draft-ietf-nvo3-arch-03 (work in progress), March 2015. 1729 [I-D.ietf-pwe3-congcons] 1730 Stein, Y., Black, D., and B. Briscoe, "Pseudowire 1731 Congestion Considerations", draft-ietf-pwe3-congcons-02 1732 (work in progress), July 2014. 1734 [I-D.ietf-sfc-architecture] 1735 Halpern, J. and C. Pignataro, "Service Function Chaining 1736 (SFC) Architecture", draft-ietf-sfc-architecture-08 (work 1737 in progress), May 2015. 1739 [I-D.ietf-sfc-problem-statement] 1740 Quinn, P. and T. Nadeau, "Service Function Chaining 1741 Problem Statement", draft-ietf-sfc-problem-statement-13 1742 (work in progress), February 2015. 1744 [I-D.ietf-trill-oam-fm] 1745 Senevirathne, T., Finn, N., Salam, S., Kumar, D., 1746 Eastlake, D., Aldrin, S., and L. Yizhou, "TRILL Fault 1747 Management", draft-ietf-trill-oam-fm-11 (work in 1748 progress), October 2014. 1750 [I-D.ietf-tsvwg-circuit-breaker] 1751 Fairhurst, G., "Network Transport Circuit Breakers", 1752 draft-ietf-tsvwg-circuit-breaker-01 (work in progress), 1753 March 2015. 1755 [I-D.ietf-tsvwg-gre-in-udp-encap] 1756 Crabbe, E., Yong, L., Xu, X., and T. Herbert, "GRE-in-UDP 1757 Encapsulation", draft-ietf-tsvwg-gre-in-udp-encap-06 (work 1758 in progress), March 2015. 1760 [I-D.ietf-tsvwg-port-use] 1761 Touch, J., "Recommendations on Using Assigned Transport 1762 Port Numbers", draft-ietf-tsvwg-port-use-11 (work in 1763 progress), April 2015. 1765 [I-D.quinn-sfc-nsh] 1766 Quinn, P., Guichard, J., Surendra, S., Smith, M., 1767 Henderickx, W., Nadeau, T., Agarwal, P., Manur, R., 1768 Chauhan, A., Halpern, J., Majee, S., Elzur, U., Melman, 1769 D., Garg, P., McConnell, B., Wright, C., and K. Kevin, 1770 "Network Service Header", draft-quinn-sfc-nsh-07 (work in 1771 progress), February 2015. 1773 [I-D.saldana-tsvwg-simplemux] 1774 Saldana, J., "Simplemux. A generic multiplexing protocol", 1775 draft-saldana-tsvwg-simplemux-02 (work in progress), 1776 January 2015. 1778 [I-D.shepherd-bier-problem-statement] 1779 Shepherd, G., Dolganow, A., and a. 1780 arkadiy.gulko@thomsonreuters.com, "Bit Indexed Explicit 1781 Replication (BIER) Problem Statement", 1782 draft-shepherd-bier-problem-statement-02 (work in 1783 progress), February 2015. 1785 [I-D.sridharan-virtualization-nvgre] 1786 Garg, P. and Y. Wang, "NVGRE: Network Virtualization using 1787 Generic Routing Encapsulation", 1788 draft-sridharan-virtualization-nvgre-08 (work in 1789 progress), April 2015. 1791 [I-D.wei-tsvwg-tunnel-congestion-feedback] 1792 Wei, X., Zhu, L., and L. Deng, "Tunnel Congestion 1793 Feedback", draft-wei-tsvwg-tunnel-congestion-feedback-03 1794 (work in progress), October 2014. 1796 [I-D.wijnands-bier-architecture] 1797 Wijnands, I., Rosen, E., Dolganow, A., Przygienda, T., and 1798 S. Aldrin, "Multicast using Bit Index Explicit 1799 Replication", draft-wijnands-bier-architecture-05 (work in 1800 progress), March 2015. 1802 [I-D.wijnands-mpls-bier-encapsulation] 1803 Wijnands, I., Rosen, E., Dolganow, A., Tantsura, J., and 1804 S. Aldrin, "Encapsulation for Bit Index Explicit 1805 Replication in MPLS Networks", 1806 draft-wijnands-mpls-bier-encapsulation-02 (work in 1807 progress), December 2014. 1809 [I-D.xu-bier-encapsulation] 1810 Xu, X., Somasundaram, S., Jacquenet, C., and R. Raszuk, 1811 "BIER Encapsulation", draft-xu-bier-encapsulation-02 (work 1812 in progress), February 2015. 1814 [IEEE802.1Q-2014] 1815 IEEE, "IEEE Standard for Local and metropolitan area 1816 networks--Bridges and Bridged Networks", IEEE Std 802.1Q- 1817 2014, 2014, 1818 . 1820 (Access Controlled link within page) 1822 Authors' Addresses 1824 Erik Nordmark 1825 Arista Networks 1826 5453 Great America Parkway 1827 Santa Clara, CA 95054 1828 USA 1830 Email: nordmark@arista.com 1832 Albert Tian 1833 Ericsson Inc. 1834 300 Holger Way 1835 San Jose, California 95134 1836 USA 1838 Email: albert.tian@ericsson.com 1840 Jesse Gross 1841 VMware 1842 3401 Hillview Ave. 1843 Palo Alto, CA 94304 1844 USA 1846 Email: jgross@vmware.com 1847 Jon Hudson 1848 Brocade Communications Systems, Inc. 1849 130 Holger Way 1850 San Jose, CA 95134 1851 USA 1853 Email: jon.hudson@gmail.com 1855 Lawrence Kreeger 1856 Cisco Systems, Inc. 1857 170 W. Tasman Avenue 1858 San Jose, CA 95134 1859 USA 1861 Email: kreeger@cisco.com 1863 Pankaj Garg 1864 Microsoft 1865 1 Microsoft Way 1866 Redmond, WA 98052 1867 USA 1869 Email: pankajg@microsoft.com 1871 Patricia Thaler 1872 Broadcom Corporation 1873 3151 Zanker Road 1874 San Jose, CA 95134 1875 USA 1877 Email: pthaler@broadcom.com 1879 Tom Herbert 1880 Google 1881 1600 Amphitheatre Parkway 1882 Mountain View, CA 1883 USA 1885 Email: therbert@google.com