idnits 2.17.00 (12 Aug 2021) /tmp/idnits14675/draft-ietf-nvo3-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.) ** There are 46 instances of too long lines in the document, the longest one being 2 characters in excess of 72. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year == Line 377 has weird spacing: '...of TLVs from ...' == Line 505 has weird spacing: '...lso, if the...' -- The document date (September 14, 2018) is 1345 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- == Missing Reference: 'RFC2119' is mentioned on line 149, but not defined == Missing Reference: 'I-D.ietf-nvo3-vxlan-gpe' is mentioned on line 720, but not defined == Missing Reference: 'I-D.herbert-gue-extensions' is mentioned on line 710, but not defined == Missing Reference: 'I-D.ietf-nvo3-gue' is mentioned on line 734, but not defined == Unused Reference: 'KEYWORDS' is defined on line 621, but no explicit reference was found in the text == Unused Reference: 'Geneve' is defined on line 626, but no explicit reference was found in the text == Unused Reference: 'GUE' is defined on line 628, but no explicit reference was found in the text == Unused Reference: 'NSH' is defined on line 629, but no explicit reference was found in the text == Unused Reference: 'VXLAN-GPE' is defined on line 630, but no explicit reference was found in the text == Outdated reference: draft-ietf-nvo3-geneve has been published as RFC 8926 == Outdated reference: draft-ietf-sfc-nsh has been published as RFC 8300 == Outdated reference: A later version (-05) exists of draft-smith-vxlan-group-policy-03 Summary: 2 errors (**), 0 flaws (~~), 15 warnings (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 INTERNET-DRAFT Sami Boutros(Ed.) 3 Intended Status: Informational VMware 5 Expires: March 18, 2019 September 14, 2018 7 NVO3 Encapsulation Considerations 8 draft-ietf-nvo3-encap-02 10 Abstract 12 As communicated by WG Chairs, the IETF NVO3 chairs and Routing Area 13 director have chartered a design team to take forward the 14 encapsulation discussion and see if there is potential to design a 15 common encapsulation that addresses the various technical concerns. 17 There are implications of different encapsulations in real 18 environments consisting of both software and hardware implementations 19 and spanning multiple data centers. For example, OAM functions such 20 as path MTU discovery become challenging with multiple encapsulations 21 along the data path. 23 The design team recommend Geneve with few modifications as the common 24 encapsulation, more details are described in section 7. 26 Status of this Memo 28 This Internet-Draft is submitted to IETF in full conformance with the 29 provisions of BCP 78 and BCP 79. 31 Internet-Drafts are working documents of the Internet Engineering 32 Task Force (IETF), its areas, and its working groups. Note that 33 other groups may also distribute working documents as 34 Internet-Drafts. 36 Internet-Drafts are draft documents valid for a maximum of six months 37 and may be updated, replaced, or obsoleted by other documents at any 38 time. It is inappropriate to use Internet-Drafts as reference 39 material or to cite them other than as "work in progress." 41 The list of current Internet-Drafts can be accessed at 42 http://www.ietf.org/1id-abstracts.html 43 The list of Internet-Draft Shadow Directories can be accessed at 44 http://www.ietf.org/shadow.html 46 Copyright and License Notice 48 Copyright (c) 2018 IETF Trust and the persons identified as the 49 document authors. All rights reserved. 51 This document is subject to BCP 78 and the IETF Trust's Legal 52 Provisions Relating to IETF Documents 53 (http://trustee.ietf.org/license-info) in effect on the date of 54 publication of this document. Please review these documents 55 carefully, as they describe your rights and restrictions with respect 56 to this document. Code Components extracted from this document must 57 include Simplified BSD License text as described in Section 4.e of 58 the Trust Legal Provisions and are provided without warranty as 59 described in the Simplified BSD License. 61 Table of Contents 63 1. Problem Statement . . . . . . . . . . . . . . . . . . . . . . . 4 64 2. Design Team Goals . . . . . . . . . . . . . . . . . . . . . . . 4 65 3. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . . 4 66 4. Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . 4 67 5. Issues with current Encapsulations . . . . . . . . . . . . . . 5 68 5.1 Geneve . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 69 5.2 GUE . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 70 5.3 VXLAN-GPE . . . . . . . . . . . . . . . . . . . . . . . . . 5 71 6. Common Encapsulation Considerations . . . . . . . . . . . . . . 6 72 6.1 Current Encapsulations . . . . . . . . . . . . . . . . . . . 6 73 6.2 Useful Extensions Use cases . . . . . . . . . . . . . . . . 6 74 6.2.1. Telemetry extensions. . . . . . . . . . . . . . . . . . 6 75 6.2.2. Security/Integrity extensions . . . . . . . . . . . . . 7 76 6.2.3. Group Base Policy . . . . . . . . . . . . . . . . . . . 7 77 6.3 Hardware Considerations . . . . . . . . . . . . . . . . . . 7 78 6.4 Extension Size . . . . . . . . . . . . . . . . . . . . . . . 8 79 6.5 Extension Ordering . . . . . . . . . . . . . . . . . . . . . 9 80 6.6 TLV vs Bit Fields . . . . . . . . . . . . . . . . . . . . . 9 81 6.7 Control Plane Considerations . . . . . . . . . . . . . . . . 10 82 6.8 Split NVE . . . . . . . . . . . . . . . . . . . . . . . . . 11 83 6.9 Larger VNI Considerations . . . . . . . . . . . . . . . . . 11 84 7. Design team recommendations . . . . . . . . . . . . . . . . . . 11 85 8. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 14 86 9. Security Considerations . . . . . . . . . . . . . . . . . . . . 14 87 10. References . . . . . . . . . . . . . . . . . . . . . . . . . 14 88 10.1 Normative References . . . . . . . . . . . . . . . . . . . 14 89 10.2 Informative References . . . . . . . . . . . . . . . . . . 15 90 11. Appendix A . . . . . . . . . . . . . . . . . . . . . . . . . . 15 91 11.1. Overview . . . . . . . . . . . . . . . . . . . . . . . . 15 92 11.2. Extensibility . . . . . . . . . . . . . . . . . . . . . . 15 93 11.2.1. Native Extensibility Support . . . . . . . . . . . . 15 94 11.2.2. Extension Parsing . . . . . . . . . . . . . . . . . . 15 95 11.2.3. Critical Extensions . . . . . . . . . . . . . . . . . 16 96 11.2.4. Maximal Header Length . . . . . . . . . . . . . . . . 16 97 11.3. Encapsulation Header . . . . . . . . . . . . . . . . . . 16 98 11.3.1. Virtual Network Identifier (VNI) . . . . . . . . . . 16 99 11.3.2. Next Protocol . . . . . . . . . . . . . . . . . . . . 17 100 11.3.3. Other Header Fields . . . . . . . . . . . . . . . . . 17 101 11.4. Comparison Summary . . . . . . . . . . . . . . . . . . . 17 102 Authors' Addresses (In alphabetical order) . . . . . . . . . . . . 18 104 1. Problem Statement 106 As communicated by WG Chairs, the NVO3 WG charter states that it may 107 produce requirements for network virtualization data planes based on 108 encapsulation of virtual network traffic over an IP-based underlay 109 data plane. Such requirements should consider OAM and security. Based 110 on these requirements the WG will select, extend, and/or develop one 111 or more data plane encapsulation format(s). 113 This has led to drafts describing three encapsulations being adopted 114 by the working group: 116 - draft-ietf-nvo3-geneve-03 118 - draft-ietf-nvo3-gue-04 120 - draft-ietf-nvo3-vxlan-gpe-02 122 Discussion on the list and in face-to-face meetings has identified a 123 number of technical problems with each of these encapsulations. 124 Furthermore, there was clear consensus at the IETF meeting in Berlin 125 that it is undesirable for the working group to progress more than 126 one data plane encapsulation. Although consensus could not be reached 127 on the list, the overall consensus was for a single encapsulation 128 (RFC2418, Section 3.3). Nonetheless there has been resistance to 129 converging on a single encapsulation format. 131 2. Design Team Goals 133 As communicated by WG Chairs, the design team should take one of the 134 proposed encapsulations and enhance it to address the technical 135 concerns. The simple evolution of deployed networks as well as 136 applicability to all locations in the NVO3 architecture are goals. 137 The DT should specifically avoid a design that is burdensome on 138 hardware implementations, but should allow future extensibility. The 139 chosen design should also operate well with ICMP and in ECMP 140 environments. If further extensibility is required, then it should be 141 done in such a manner that it does not require the consent of an 142 entity outside of the IETF. 144 3. Terminology 146 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 147 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 148 document are to be interpreted as described in RFC 2119 [RFC2119]. 150 4. Abbreviations 151 NVO3 Network Virtualization Overlays over Layer 3 153 OAM Operations, Administration, and Maintenance 155 TLV Type, Length, and Value 157 VNI Virtual Network Identifier 159 NVE Network Virtualization Edge 161 NVA Network Virtualization Authority 163 NIC Network interface card 165 Transit device Underlay network devices between NVE(s). 167 5. Issues with current Encapsulations 169 As summarized by WG Chairs. 171 5.1 Geneve 173 - Can't be implemented cost-effectively in all use cases because 174 variable length header and order of the TLVs makes is costly (in 175 terms of number of gates) to implement in hardware 177 - Header doesn't fit into largest commonly available parse buffer 178 (256 bytes in NIC). Cannot justify doubling buffer size unless it is 179 mandatory for hardware to process additional option fields. 181 5.2 GUE 183 - There were a significant number of objections related to the 184 complexity of implementation in hardware, similar to those noted for 185 Geneve above. 187 5.3 VXLAN-GPE 189 - GPE is not day-1 backwards compatible with VXLAN. Although the 190 frame format is similar, it uses a different UDP port, so would 191 require changes to existing implementations even if the rest of the 192 GPE frame is the same. 194 - GPE is insufficiently extensible. Numerous extensions and options 195 have been designed for GUE and Geneve. Note that these have not yet 196 been validated by the WG. 198 - Security e.g. of the VNI has not been addressed by GPE. Although a 199 shim header could be used for security and other extensions, this has 200 not been defined yet and its implications on offloading in NICs are 201 not understood. 203 6. Common Encapsulation Considerations 205 6.1 Current Encapsulations 207 Appendix A includes a detailed comparison between the three proposed 208 encapsulations. The comparison indicates several common properties, 209 but also three major differences among the encapsulations: 211 - Extensibility: Geneve and GUE were defined with built-in 212 extensibility, while VXLAN-GPE is not inherently extensible. Note 213 that any of the three encapsulations can be extended using the 214 Network Service Header (NSH). 216 - Extension method: Geneve is extensible using Type/Length/Value 217 (TLV) fields, while GUE uses a small set of possible extensions, and 218 a set of flags that indicate which extension is present. 220 - Length field: Geneve and GUE include a Length field, indicating the 221 length of the encapsulation header, while VXLAN-GPE does not include 222 such a field. 224 6.2 Useful Extensions Use cases 226 Non vendor specific TLV MUST follow the standardization process. The 227 following use cases for extensions shows that there is a strong 228 requirement to support variable length extensions with possible 229 different subtypes. 231 6.2.1. Telemetry extensions. 233 In several scenarios it is beneficial to make information about the 234 path a packet took through the network or through a network device as 235 well as associated telemetry information available to the operator. 237 This includes not only tasks like debugging, troubleshooting, as well 238 as network planning and network optimization but also policy or 239 service level agreement compliance checks. 241 Packet scheduling algorithms, especially for balancing traffic across 242 equal cost paths or links, often leverage information contained 243 within the packet, such as protocol number, IP-address or MAC- 244 address. Probe packets would thus either need to be sent from the 245 exact same endpoints with the exact same parameters, or probe packets 246 would need to be artificially constructed as "fake" packets and 247 inserted along the path. Both approaches are often not feasible from 248 an operational perspective, be it that access to the end-system is 249 not feasible, or that the diversity of parameters and associated 250 probe packets to be created is simply too large. An in-band telemetry 251 mechanism in extensions is an alternative in those cases. 253 6.2.2. Security/Integrity extensions 255 Since the currently proposed NVO3 encapsulations do not protect their 256 headers a single bit corruption in the VNI field could deliver a 257 packet to the wrong tenant. Extensions are needed to use any 258 sophisticated security. 260 The possibility of VNI spoofing with an NVO3 protocol is exacerbated 261 by the use of UDP. Systems typically have no restrictions on 262 applications being able to send to any UDP port so an unprivileged 263 application can trivially spoof for instance, VXLAN packets, 264 including using arbitrary VNIs. 266 One can envision HMAC-like support in some NVO3 extension to 267 authenticate the header and the outer IP addresses, thereby 268 preventing attackers from injecting packets with spoofed VNIs. 270 An other aspect of security is payload security. Essentially this is 271 to make packets that look like IP|UDP|NVO3 Encap|DTLS/IPSEC-ESP 272 Extension|payload. This is nice since we still have the UDP header 273 for ECMP, the NVO3 header is in plain text so it can by read by 274 network elements, and different security or other payload transforms 275 can be supported on a single UDP port (we don't need a separate UDP 276 for DTLS/IPSEC). 278 6.2.3. Group Base Policy 280 Another use case would be to carry the Group Based Policy (GBP) 281 source group information within a NVO3 header extension in a similar 282 manner as has been implemented for VXLAN [VXLAN-GBP]. This allows 283 various forms of policy such as access control and QoS to be applied 284 between abstract groups rather than coupled to specific endpoint 285 addresses. 287 6.3 Hardware Considerations 289 Hardware restrictions should be taken into consideration along with 290 future hardware enhancements that may provide more flexible metadata 291 processing. However, the set of options that need to and will be 292 implemented in hardware will be a subset of what is implemented in 293 software, since software NVEs are likely to grow features, and hence 294 option support, at a more rapid rate. 296 We note that it is hard to predict which options will be implemented 297 in which piece of hardware and when. That depends on whether the 298 hardware will be in the form of a NIC providing increasing offload 299 capabilities to software NVEs, or a switch chip being used as an NVE 300 gateway towards non-NVO3 parts of the network, or even an transit 301 devices that participates in the NVO3 dataplane e.g. for OAM 302 purposes. 304 A result of this is that it doesn't look useful to prescribe some 305 order of the option so that the ones that are likely to be 306 implemented in hardware come first; we can't decide such an order 307 when we define the options, however a control plane can enforce such 308 order for some hardware implementations. 310 We do know that hardware needs to initially be able to efficiently 311 skip over the NVO3 header to find the inner payload. That is needed 312 for both NICs doing e.g. TCP offload and transit devices and NVEs 313 applying policy/ACLs to the inner payload. 315 6.4 Extension Size 317 Extension header length has a significant impact to hardware and 318 software implementations. A total header length that is too small 319 will unnecessarily constrained software flexibility. A total header 320 length that is too large will place a nontrivial cost on hardware 321 implementations. Thus, the design team recommends that there be a 322 minimum and maximum total extension header length selected. The 323 maximum total header length is determined by the bits allocated for 324 the total extension header length field. The risk with this approach 325 is that it may be difficult to extend the total header size in the 326 future. The minimum total header length is determined by a 327 requirement in the specifications that all implementations must meet. 328 The risk with this approach is that all implementations will only 329 implement the minimum total header length which would then become the 330 de facto maximum total header length. The recommended minimum total 331 header length is 64 bytes. 333 Single Extension size should always be 4 bytes aligned. 335 The maximum length of a single option should be large enough to meet 336 the different extension use case requirements e.g. in-band telemetry 337 and future use. 339 6.5 Extension Ordering 341 In order to support hardware nodes at the tunnel endpoint or at the 342 transit that can process one or few extensions TLVs in TCAM. A 343 control plane in such a deployment can signal a capability to ensure 344 a specific TLV will always appear in a specific order for example the 345 first one in the packet. 347 The order of the TLVs should be HW friendly for both the sender and 348 the receiver and possibly the transit node too. 350 Transit nodes doesn't participate in control plane communication 351 between the end points and are not required to process the options 352 however, if they do, they need to process only a small subset of 353 options that will be consumed by tunnel endpoints. 355 6.6 TLV vs Bit Fields 357 If there is a well-known initial set of options that are likely to be 358 implemented in software and in hardware, it can be efficient to use 359 the bit-field approach as in GUE. However, as described in section 360 6.3, if options are added over time and different subsets of options 361 are likely to be implemented in different pieces of hardware, then it 362 would be hard for the IETF to specify which options should get the 363 early bit fields. TLVs are a lot more flexible, which avoids the need 364 to determine the relative importance different options. However, 365 general TLV of arbitrary order, size, and repetition of the same 366 order is difficult to implement in hardware. A middle ground is to 367 use TLV with restrictions on the size and alignment, observing that 368 individual TLVs can have a fixed length, and support in the control 369 plane such that an NVE will only receive options that to needs and 370 implements. The control plane approach can potentially be used to 371 control the order of the TLVs sent to a particular NVE. Note that 372 transit devices are not likely to participate in the control plane 373 hence to the extent that they need to participate in option 374 processing they need more effort, But transit devices would have 375 issues with future GUE bits being defined for future options as well. 377 A benefit of TLVs from a HW perspective is that they are self 378 describing i.e., all the information is in the TLV. In a Bit fields 379 approach the hardware needs to look up the bit to determine the 380 length of the data associated with the bit through some separate 381 table, which would add hardware complexity. 383 There are use cases where multiple modules of software are running on 384 NVE. This can be modules such as a diagnostic module by one vendor 385 that does packet sampling and another module from a different vendor 386 that does a firewall. Using a TLV format, it is easier to have 387 different software modules process different TLVs, which could be 388 standard extensions or vendor specific extensions defined by the 389 different vendors, without conflicting with each other. This can help 390 with hardware modularity as well. There are some implementations with 391 options that allows different software like mac learning and security 392 handle different options. 394 6.7 Control Plane Considerations 396 Given that we want to allow large flexibility and extensibility for 397 e.g. software NVEs, yet be able to support key extensions in less 398 flexible e.g. hardware NVEs, it is useful to consider the control 399 plane. By control plane in this context we mean both protocols such 400 as EVPN and others, and also deployment specific configuration. 402 If each NVE can express in the control plane that they only care 403 about particular extensions (could be a single extension, or a few), 404 and the source NVEs only include requested extensions in the NVO3 405 packets, then the target NVE can both use a simpler parser (e.g., a 406 TCAM might be usable to look for a single NVO3 extension) and the 407 depth of the inner payload in the NVO3 packet will be minimized. 408 Furthermore, if the target NVE cares about a few extensions and can 409 express in the control plane the desired order of those extensions in 410 the NVO3 packets, then it can provide useful functionality with 411 minimal hardware requirements. 413 Note that transit devices that are not aware of the NVO3 extensions 414 somewhat benefit from such an approach, since the inner payload is 415 less deep in the packet if no extraneous extensions are included in 416 the packet. However, in general a transit device is not likely to 417 participate in the NVO3 control plane. (However, configuration 418 mechanisms can take into account limitations of the transit devices 419 used in particular deployments.) 421 Note that in this approach different NVEs could desire different 422 (sets of) extensions, which means that the source NVE needs to be 423 able to place different sets of extensions in different NVO3 packets, 424 and perhaps in different order. It also assumes that underlay 425 multicast or replication servers are not used together with NVO3 426 extensions. 428 There is a need to consider mandatory extensions versus optional 429 extensions. Mandatory extensions require the receiver to drop the 430 packet if the extension is unknown. A control plane mechanism can 431 prevent the need for dropping unknown extensions, since they would 432 not be included to targets that do not support them. 434 The control planes defined today need to add the ability to describe 435 the different encapsulations. Thus perhaps EVPN, and any other 436 control plane protocol that the IETF defines, should have a way to 437 enumerate the supported NVO3 extensions and their order. 439 The WG should consider developing a separate draft on guidance for 440 option processing and control plane participation. This should 441 provide examples/guidance on range of usage models and deployments 442 scenarios for specific options and ordering that are relevant for 443 that specific deployment. This includes end points and middle boxes 444 using the options. So, having the control plane negotiate the 445 constraints is most appropriate and flexible way to address these 446 requirements. 448 6.8 Split NVE 450 If the working group sees a need for having the hosts send and 451 receive options in a split NVE case, this is possible using any of 452 the existing extensible encapsulations (Geneve, GUE, GPE+NSH) by 453 defining a way to carry those over other transports. NSH can already 454 be used over different transports. 456 If we need to do this with other encapsulations it can be done by 457 defining an Ether type for other encapsulations so that it can be 458 carried over Ethernet and 802.1Q. 460 If we need to carry other encapsulations over MPLS, it would require 461 an EVPN control plane to signal that other encapsulation header + 462 options will be present in front of the L2 packet. The VNI can be 463 ignored in the header, and the MPLS label will be the one used to 464 identify the EVPN L2 instance. 466 6.9 Larger VNI Considerations 468 We discussed whether we should make VNI 32-bits or larger. The 469 benefit of 24-bit VNI would be to avoid unnecessary changes with 470 existing proposals and implementations that are almost all, if not 471 all, are using 24-bit VNI. If we need a larger VNI, an extension can 472 be used to support that. 474 7. Design team recommendations 476 We concluded that Geneve is most suitable as a starting point for 477 proposed standard for network virtualization, for the following 478 reasons: 480 1. We studied whether VNI should be in base header or in extensions 481 and whether it should be 24-bit or 32-bit. The design team agreed 482 that VNI is critical information for network virtualization and MUST 483 be present in all packets. Design team also agreed that 24-bit VNI 484 matches the existing widely used encapsulation format i.e. VxLAN and 485 NVGRE and hence more suitable to use going forward. 487 2. Geneve has the total options length that allow skipping over the 488 options for NIC offload operations, and will allow transit devices to 489 view flow information in the inner payload. 491 3. We considered the option of using NSH with VxLAN-GPE but given 492 that NSH is targeted at service chaining and contains service 493 chaining information, it is less suitable for the network 494 virtualization use case. The other downside for VxLAN-GPE was lack of 495 header length in VxLAN-GPE and hence makes skipping over the headers 496 to process inner payload more difficult. Total Option Length is 497 present in Geneve. It is not possible to skip any options in the 498 middle with VxLAN-GPE. In principle a split between a base header and 499 a header with options is interesting (whether that options header is 500 NSH or some new header without ties to a service path). We explored 501 whether it would make sense to either use NSH for this, or define a 502 new NVO3 options header. However, we observed that this makes it 503 slightly harder to find the inner payload since the length field is 504 not in the NVO3 header itself. Thus one more field would have to be 505 extracted to compute the start of the inner payload. Also, if the 506 experience with IPv6 extension headers is a guidance, there would be 507 a risk that key pieces of hardware might not implement the options 508 header, resulting in future calls to deprecate its use. Making the 509 options part of the base NVO3 header has less of those issues. Even 510 though the implementation of any particular option can not be 511 predicted ahead of time, the option mechanism and ability to skip the 512 options is likely to be broadly implemented. 514 4. We compared the TLV vs Bit-fields style extension and it was 515 deemed that parsing both TLV and bit-fields is expensive and while 516 bit-fields may be simpler to parse, it is also more restrictive and 517 requires guessing which extensions will be widely implemented so they 518 can get early bit assignments, given that half the bits are already 519 assigned in GUE, a widely deployed extension may appear in a flag 520 extension, and this will require extra processing, to dig the flag 521 from the flag extension and then look for the extension itself. As 522 well Bit-fields are not flexible enough to address the requirements 523 from OAM, Telemetry and security extensions, for variable length 524 option and different subtypes of the same option. While TLV are more 525 flexible, a control plane can restrict the number of option TLVs as 526 well the order and size of the TLVs to make it simpler for a 527 dataplane implementation to handle. 529 5. We briefly discussed multi-vendor NVE case, and the need to allow 530 vendors to put their own extensions in the NVE header. This is 531 possible with TLVs. 533 6. We also agreed that the C bit in Geneve is helpful to allow 534 receiver NVE to easily decide whether to process options or not. For 535 example a UUID based packet trace and how an optional extension such 536 as that can be ignored by receiver NVE and thus make it easy for NVE 537 to skip over the options. Thus the C-bit remains as defined in 538 Geneve. 540 7. There are already some extensions that are being discussed (see 541 section 6.2) of varying sizes, by using Geneve option it is possible 542 to get in band parameters like: switch id, ingress port, egress port, 543 internal delay, and queue in telemetry defined extension TLV from 544 switches. It is also possible to add Security extension TLVs like 545 HMAC and DTLS/IPSEC to authenticate the Geneve packet header and 546 secure the Geneve packet payload by software or hardware tunnel 547 endpoints. As well, a Group Based Policy extension TLV can be 548 carried. 550 8. There are implemented Geneve options today in production. There 551 are as well new HW supporting Geneve TLV parsing. In addition In-band 552 Telemetry (INT) specification being developed by P4.org illustrates 553 the option of INT meta data carried over Geneve. OVN/OVS have also 554 defined some option TLV(s) for Geneve. 556 9. The DT has addressed the usage models while considering the 557 requirements and implementations in general that includes software 558 and hardware. 560 There seems to be interest to standardize some well known secure 561 option TLVs to secure the header and payload to guarantee 562 encapsulation header integrity and tenant data privacy. The design 563 team recommends that the working group consider standardizing such 564 option(s). 566 We recommend the following enhancements to Geneve to make it more 567 suitable to hardware and yet provide the flexibility for software: 569 We would propose a text such as, while TLV are more flexible, a 570 control plane can restrict the number of option TLVs as well the 571 order and size of the TLVs to make it simpler for a data plane 572 implementation in software or hardware to handle. For example, there 573 may be some critical information such as secure hash that must be 574 processed in certain order at lowest latency. 576 A control plane can negotiate a subset of option TLVs and certain TLV 577 ordering, as well can limit the total number of option TLVs present 578 in the packet, for example, to allow hardware capable of processing 579 fewer options. Hence, the control planes need to have the ability to 580 describe the supported TLVs subset and their order. 582 The Geneve draft could specify that the subset and order of option 583 TLVs should be configurable for each remote NVE in the absence of a 584 protocol control plane. 586 We recommend Geneve to follow fragmentation recommendations in 587 overlay services like PWE3, and L2/L3 VPN recommendation to guarantee 588 larger MTU for the tunnel overhead 589 https://tools.ietf.org/html/rfc3985#section-5.3 591 We request Geneve to provide a recommendation for critical bit 592 processing - text could look like how critical bits can be used with 593 control plane specifying the critical options. 595 Given that there is a telemetry option use case for a length of 256 596 bytes, we recommend Geneve to increase the Single TLV option length 597 to 256. 599 We request Geneve to address Requirements for OAM considerations for 600 alternate marking and for performance measurements that need 2 bits 601 in the header. And clarify the need of the current OAM bit in the 602 Geneve Header. 604 We recommend the WG to work on security options for Geneve. 606 8. Acknowledgements 608 The authors would like to thank Tom Herbert for providing the 609 motivation for the Security/Integrity extension, and for his valuable 610 comments, and would like to thank T. Sridhar for his valuable 611 comments and feedback. 613 9. Security Considerations 615 This document does not introduce any additional security constraints. 617 10. References 619 10.1 Normative References 621 [KEYWORDS] Bradner, S., "Key words for use in RFCs to Indicate 622 Requirement Levels", BCP 14, RFC 2119, March 1997. 624 10.2 Informative References 626 [Geneve] Generic Network Virtualization Encapsulation [I-D.ietf-nvo3- 627 geneve] 628 [GUE] Generic UDP Encapsulation [I-D.ietf-nvo3-gue] 629 [NSH] Network Service Header [I-D.ietf-sfc-nsh] 630 [VXLAN-GPE] Virtual eXtensible Local Area Network - Generic Protocol 631 Extension [I-D.ietf-nvo3-vxlan-gpe] 633 [VXLAN-GBP] VXLAN Group Policy Option - [I-D.draft-smith-vxlan-group- 634 policy-03] 636 11. Appendix A 638 11.1. Overview 640 This section presents a comparison of the three NVO3 encapsulation 641 proposals, Geneve, GUE, and VXLAN-GPE. The three encapsulations use 642 an outer UDP/IP transport. Geneve and VXLAN-GPE use an 8-octet 643 header, while GUE uses a 4-octet header. In addition to the base 644 header, optional extensions may be included in the encapsulation, as 645 discussed in Section 3.2 below. 647 11.2. Extensibility 649 11.2.1. Native Extensibility Support 651 The Geneve and GUE encapsulations both enable optional headers to be 652 incorporated at the end of the base encapsulation header. 654 VXLAN-GPE does not provide native support for header extensions. 655 However, as discussed in [I-D.ietf-nvo3-vxlan-gpe], extensibility can 656 be attained to some extent if the Network Service Header (NSH) [I- 657 D.ietf-sfc-nsh] is used immediately following the VXLAN-GPE header. 658 NSH supports either a fixed-size extension (MD Type 1), or a 659 variable-size TLV-based extension (MD Type 2). It should be noted 660 that NSH-over-VXLAN-GPE implies an additional overhead of the 8- 661 octets NSH header, in addition to the VXLAN-GPE header. 663 11.2.2. Extension Parsing 665 The Geneve Variable Length Options are defined as 666 Type/Length/Value(TLV) extensions. Similarly, VXLAN-GPE, when using 667 NSH, can include NSH TLV-based extensions. In contrast, GUE defines 668 a small set of possible extension fields (proposed in [I-D.herbert- 669 gue-extensions]), and a set of flags in the GUE header that indicate 670 for each extension type whether it is present or not. 672 TLV-based extensions, as defined in Geneve, provide the flexibility 673 for a large number of possible extension types. Similar behavior can 674 be supported in NSH-over-VXLAN-GPE when using MD Type 2. The flag- 675 based approach taken in GUE strives to simplify implementations by 676 defining a small number of possible extensions, used in a fixed 677 order. 679 The Geneve and GUE headers both include a length field, defining the 680 total length of the encapsulation, including the optional extensions. 682 The length field simplifies the parsing of transit devices that skip 683 the encapsulation header without parsing its extensions. 685 11.2.3. Critical Extensions 687 The Geneve encapsulation header includes the 'C' field, which 688 indicates whether the current Geneve header includes critical 689 options, which must be parsed by the tunnel endpoint. If the endpoint 690 is not able to process the critical option, the packet is discarded. 692 11.2.4. Maximal Header Length 694 The maximal header length in Geneve, including options, is 260 695 octets. GUE defines the maximal header to be 128 octets. VXLAN-GPE 696 uses a fixed-length header of 8 octets, unless NSH-over-VXLAN-GPE is 697 used, yielding an encapsulation header of up to 264 octets. 699 11.3. Encapsulation Header 701 11.3.1. Virtual Network Identifier (VNI) 703 The Geneve and VXLAN-GPE headers both include a 24-bit VNI field. 704 GUE, on the other hand, enables the use of a 32-bit field called 705 VNID; this field is not included in the GUE header, but was defined 706 as an optional extension in [I-D.herbert-gue-extensions]. 708 The VXLAN-GPE header includes the 'I' bit, indicating that the VNI 709 field is valid in the current header. A similar indicator is defined 710 as a flag in the GUE header [I-D.herbert-gue-extensions]. 712 11.3.2. Next Protocol 714 The three encapsulation headers include a field that specifies the 715 type of the next protocol header, which resides after the NVO3 716 encapsulation header. The Geneve header includes a 16-bit field that 717 uses the IEEE Ethertype convention. GUE uses an 8-bit field, which 718 uses the IANA Internet protocol numbering. The VXLAN-GPE header 719 incorporates an 8-bit Next Protocol field, using a VXLAN-GPE-specific 720 registry, defined in [I-D.ietf-nvo3-vxlan-gpe]. 722 The VXLAN-GPE header also includes the 'P' bit, which explicitly 723 indicates whether the Next Protocol field is present in the current 724 header. 726 11.3.3. Other Header Fields 728 The OAM bit, which is defined in Geneve and in VXLAN-GPE, indicates 729 whether the current packet is an OAM packet. The GUE header includes 730 a similar field, but uses different terminology; the GUE 'C-bit' 731 specifies whether the current packet is a control packet. Note that 732 the GUE control bit can potentially be used in a large set of 733 protocols that are not OAM protocols. However, the control packet 734 examples discussed in [I-D.ietf-nvo3-gue] are OAM-related. 736 Each of the three NVO3 encapsulation headers includes a 2-bit Version 737 field, which is currently defined to be zero. 739 The Geneve and VXLAN-GPE headers include reserved fields; 14 bits in 740 the Geneve header, and 27 bits in the VXLAN-GPE header are reserved. 742 11.4. Comparison Summary 744 The following table summarizes the comparison between the three NVO3 745 encapsulations. 747 +----------------+----------------+----------------+----------------+ 748 | | Geneve | GUE | VXLAN-GPE | 749 +----------------+----------------+----------------+----------------+ 750 | Outer transport| UDP/IP | UDP/IP | UDP/IP | 751 +----------------+----------------+----------------+----------------+ 752 | Base header | 8 octets | 4 octets | 8 octets | 753 | length | | | (16 octets | 754 | | | | using NSH) | 755 +----------------+----------------+----------------+----------------+ 756 | Extensibility |Variable length |Extension fields| No native ext- | 757 | | options | | ensibility. | 758 | | | | Extensible | 759 | | | | using NSH. | 760 +----------------+----------------+----------------+----------------+ 761 | Extension | TLV-based | Flag-based | TLV-based | 762 | parsing method | | |(using NSH with | 763 | | | | MD Type 2) | 764 +----------------+----------------+----------------+----------------+ 765 | Extension | Variable | Fixed | Variable | 766 | order | | | (using NSH) | 767 +----------------+----------------+----------------+----------------+ 768 | Length field | + | + | - | 769 +----------------+----------------+----------------+----------------+ 770 | Max Header | 260 octets | 128 octets | 8 octets | 771 | Length | | |(264 using NSH) | 772 +----------------+----------------+----------------+----------------+ 773 | Critical exte- | + | - | - | 774 | nsion bit | | | | 775 +----------------+----------------+----------------+----------------+ 776 | VNI field size | 24 bits | 32 bits | 24 bits | 777 | | | (extension) | | 778 +----------------+----------------+----------------+----------------+ 779 | Next protocol | 16 bits | 8 bits | 8 bits | 780 | field | Ethertype | Internet prot- | New registry | 781 | | registry | ocol registry | | 782 +----------------+----------------+----------------+----------------+ 783 | Next protocol | - | - | + | 784 | indicator | | | | 785 +----------------+----------------+----------------+----------------+ 786 | OAM / control | OAM bit | Control bit | OAM bit | 787 | field | | | | 788 +----------------+----------------+----------------+----------------+ 789 | Version field | 2 bits | 2 bits | 2 bits | 790 +----------------+----------------+----------------+----------------+ 791 | Reserved bits | 14 bits | - | 27 bits | 792 +----------------+----------------+----------------+----------------+ 794 Figure 1: NVO3 Encapsulation Comparison 796 Authors' Addresses (In alphabetical order) 798 Sami Boutros 799 VMware 800 Email: sboutros@vmware.com 802 Ilango Ganga 803 Intel 804 Email: ilango.s.ganga@intel.com 805 Pankaj Garg 806 Microsoft 807 Email: pankajg@microsoft.com 809 Rajeev Manur 810 Broadcom 811 Email: rajeev.manur@broadcom.com 813 Tal Mizrahi 814 Marvell 815 Email: talmi@marvell.com 817 David Mozes 818 Email: mosesster@gmail.com 820 Erik Nordmark 821 Email: nordmark@sonic.net 823 Michael Smith 824 Cisco 825 Email: michsmit@cisco.com 827 Sam Aldrin 828 Google 829 Email: aldrin.ietf@gmail.com 831 Ignas Bagdonas 832 Equinix 833 Email: ibagdona.ietf@gmail.com