idnits 2.17.1 draft-ietf-nvo3-geneve-09.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- No issues found here. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (February 22, 2019) is 1889 days in the past. Is this intentional? Checking references for intended status: Proposed Standard ---------------------------------------------------------------------------- (See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) == Outdated reference: A later version (-13) exists of draft-ietf-intarea-tunnels-09 == Outdated reference: A later version (-12) exists of draft-ietf-nvo3-encap-02 -- Obsolete informational reference (is this intentional?): RFC 2460 (Obsoleted by RFC 8200) Summary: 0 errors (**), 0 flaws (~~), 3 warnings (==), 2 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network Working Group J. Gross, Ed. 3 Internet-Draft 4 Intended status: Standards Track I. Ganga, Ed. 5 Expires: August 26, 2019 Intel 6 T. Sridhar, Ed. 7 VMware 8 February 22, 2019 10 Geneve: Generic Network Virtualization Encapsulation 11 draft-ietf-nvo3-geneve-09 13 Abstract 15 Network virtualization involves the cooperation of devices with a 16 wide variety of capabilities such as software and hardware tunnel 17 endpoints, transit fabrics, and centralized control clusters. As a 18 result of their role in tying together different elements in the 19 system, the requirements on tunnels are influenced by all of these 20 components. Flexibility is therefore the most important aspect of a 21 tunnel protocol if it is to keep pace with the evolution of the 22 system. This draft describes Geneve, a protocol designed to 23 recognize and accommodate these changing capabilities and needs. 25 Status of This Memo 27 This Internet-Draft is submitted in full conformance with the 28 provisions of BCP 78 and BCP 79. 30 Internet-Drafts are working documents of the Internet Engineering 31 Task Force (IETF). Note that other groups may also distribute 32 working documents as Internet-Drafts. The list of current Internet- 33 Drafts is at https://datatracker.ietf.org/drafts/current/. 35 Internet-Drafts are draft documents valid for a maximum of six months 36 and may be updated, replaced, or obsoleted by other documents at any 37 time. It is inappropriate to use Internet-Drafts as reference 38 material or to cite them other than as "work in progress." 40 This Internet-Draft will expire on August 26, 2019. 42 Copyright Notice 44 Copyright (c) 2019 IETF Trust and the persons identified as the 45 document authors. All rights reserved. 47 This document is subject to BCP 78 and the IETF Trust's Legal 48 Provisions Relating to IETF Documents 49 (https://trustee.ietf.org/license-info) in effect on the date of 50 publication of this document. Please review these documents 51 carefully, as they describe your rights and restrictions with respect 52 to this document. Code Components extracted from this document must 53 include Simplified BSD License text as described in Section 4.e of 54 the Trust Legal Provisions and are provided without warranty as 55 described in the Simplified BSD License. 57 Table of Contents 59 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 60 1.1. Requirements Language . . . . . . . . . . . . . . . . . . 4 61 1.2. Terminology . . . . . . . . . . . . . . . . . . . . . . . 4 62 2. Design Requirements . . . . . . . . . . . . . . . . . . . . . 5 63 2.1. Control Plane Independence . . . . . . . . . . . . . . . 6 64 2.2. Data Plane Extensibility . . . . . . . . . . . . . . . . 7 65 2.2.1. Efficient Implementation . . . . . . . . . . . . . . 7 66 2.3. Use of Standard IP Fabrics . . . . . . . . . . . . . . . 8 67 3. Geneve Encapsulation Details . . . . . . . . . . . . . . . . 9 68 3.1. Geneve Packet Format Over IPv4 . . . . . . . . . . . . . 9 69 3.2. Geneve Packet Format Over IPv6 . . . . . . . . . . . . . 10 70 3.3. UDP Header . . . . . . . . . . . . . . . . . . . . . . . 12 71 3.4. Tunnel Header Fields . . . . . . . . . . . . . . . . . . 13 72 3.5. Tunnel Options . . . . . . . . . . . . . . . . . . . . . 14 73 3.5.1. Options Processing . . . . . . . . . . . . . . . . . 16 74 4. Implementation and Deployment Considerations . . . . . . . . 17 75 4.1. Applicability Statement . . . . . . . . . . . . . . . . . 17 76 4.2. Congestion Control Functionality . . . . . . . . . . . . 18 77 4.3. UDP Checksum . . . . . . . . . . . . . . . . . . . . . . 18 78 4.3.1. UDP Zero Checksum Handling with IPv6 . . . . . . . . 18 79 4.4. Encapsulation of Geneve in IP . . . . . . . . . . . . . . 20 80 4.4.1. IP Fragmentation . . . . . . . . . . . . . . . . . . 20 81 4.4.2. DSCP, ECN and TTL . . . . . . . . . . . . . . . . . . 21 82 4.4.3. Broadcast and Multicast . . . . . . . . . . . . . . . 22 83 4.4.4. Unidirectional Tunnels . . . . . . . . . . . . . . . 22 84 4.5. Constraints on Protocol Features . . . . . . . . . . . . 23 85 4.5.1. Constraints on Options . . . . . . . . . . . . . . . 23 86 4.6. NIC Offloads . . . . . . . . . . . . . . . . . . . . . . 23 87 4.7. Inner VLAN Handling . . . . . . . . . . . . . . . . . . . 24 88 5. Interoperability Issues . . . . . . . . . . . . . . . . . . . 24 89 6. Security Considerations . . . . . . . . . . . . . . . . . . . 25 90 6.1. Data Confidentiality . . . . . . . . . . . . . . . . . . 26 91 6.1.1. Inter-Data Center Traffic . . . . . . . . . . . . . . 26 92 6.2. Data Integrity . . . . . . . . . . . . . . . . . . . . . 27 93 6.3. Authentication of NVE peers . . . . . . . . . . . . . . . 27 94 6.4. Options Interpretation by Transit Devices . . . . . . . . 27 95 6.5. Multicast/Broadcast . . . . . . . . . . . . . . . . . . . 28 96 6.6. Control Plane Communications . . . . . . . . . . . . . . 28 98 7. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 28 99 8. Contributors . . . . . . . . . . . . . . . . . . . . . . . . 29 100 9. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 30 101 10. References . . . . . . . . . . . . . . . . . . . . . . . . . 31 102 10.1. Normative References . . . . . . . . . . . . . . . . . . 31 103 10.2. Informative References . . . . . . . . . . . . . . . . . 32 104 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 34 106 1. Introduction 108 Networking has long featured a variety of tunneling, tagging, and 109 other encapsulation mechanisms. However, the advent of network 110 virtualization has caused a surge of renewed interest and a 111 corresponding increase in the introduction of new protocols. The 112 large number of protocols in this space, ranging all the way from 113 VLANs [IEEE.802.1Q_2014] and MPLS [RFC3031] through the more recent 114 VXLAN [RFC7348], NVGRE [RFC7637], often leads to questions about the 115 need for new encapsulation formats and what it is about network 116 virtualization in particular that leads to their proliferation. 118 While many encapsulation protocols seek to simply partition the 119 underlay network or bridge between two domains, network 120 virtualization views the transit network as providing connectivity 121 between multiple components of a distributed system. In many ways 122 this system is similar to a chassis switch with the IP underlay 123 network playing the role of the backplane and tunnel endpoints on the 124 edge as line cards. When viewed in this light, the requirements 125 placed on the tunnel protocol are significantly different in terms of 126 the quantity of metadata necessary and the role of transit nodes. 128 Current work such as VL2 [VL2] and the NVO3 working group 129 [I-D.ietf-nvo3-dataplane-requirements] have described some of the 130 properties that the data plane must have to support network 131 virtualization. However, one additional defining requirement is the 132 need to carry system state along with the packet data. The use of 133 some metadata is certainly not a foreign concept - nearly all 134 protocols used for virtualization have at least 24 bits of identifier 135 space as a way to partition between tenants. This is often described 136 as overcoming the limits of 12-bit VLANs, and when seen in that 137 context, or any context where it is a true tenant identifier, 16 138 million possible entries is a large number. However, the reality is 139 that the metadata is not exclusively used to identify tenants and 140 encoding other information quickly starts to crowd the space. In 141 fact, when compared to the tags used to exchange metadata between 142 line cards on a chassis switch, 24-bit identifiers start to look 143 quite small. There are nearly endless uses for this metadata, 144 ranging from storing input ports for simple security policies to 145 service based context for interposing advanced middleboxes. 147 Existing tunnel protocols have each attempted to solve different 148 aspects of these new requirements, only to be quickly rendered out of 149 date by changing control plane implementations and advancements. 150 Furthermore, software and hardware components and controllers all 151 have different advantages and rates of evolution - a fact that should 152 be viewed as a benefit, not a liability or limitation. This draft 153 describes Geneve, a protocol which seeks to avoid these problems by 154 providing a framework for tunneling for network virtualization rather 155 than being prescriptive about the entire system. 157 1.1. Requirements Language 159 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 160 "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and 161 "OPTIONAL" in this document are to be interpreted as described in BCP 162 14 [RFC2119] [RFC8174] when, and only when, they appear in all 163 capitals, as shown here. 165 1.2. Terminology 167 The NVO3 framework [RFC7365] defines many of the concepts commonly 168 used in network virtualization. In addition, the following terms are 169 specifically meaningful in this document: 171 Checksum offload. An optimization implemented by many NICs which 172 enables computation and verification of upper layer protocol 173 checksums in hardware on transmit and receive, respectively. This 174 typically includes IP and TCP/UDP checksums which would otherwise be 175 computed by the protocol stack in software. 177 Clos network. A technique for composing network fabrics larger than 178 a single switch while maintaining non-blocking bandwidth across 179 connection points. ECMP is used to divide traffic across the 180 multiple links and switches that constitute the fabric. Sometimes 181 termed "leaf and spine" or "fat tree" topologies. 183 ECMP. Equal Cost Multipath. A routing mechanism for selecting from 184 among multiple best next hop paths by hashing packet headers in order 185 to better utilize network bandwidth while avoiding reordering of 186 packets within a flow. 188 Geneve. Generic Network Virtualization Encapsulation. The tunnel 189 protocol described in this draft. 191 LRO. Large Receive Offload. The receive-side equivalent function of 192 LSO, in which multiple protocol segments (primarily TCP) are 193 coalesced into larger data units. 195 NIC. Network Interface Card. A NIC could be part of a tunnel 196 endpoint or transit device and can either process Geneve packets or 197 aid in the processing of Geneve packets. 199 Transit device. A forwarding element along the path of the tunnel 200 making up part of the Underlay Network. A transit device MAY be 201 capable of understanding the Geneve packet format but does not 202 originate or terminate Geneve packets. 204 LSO. Large Segmentation Offload. A function provided by many 205 commercial NICs that allows data units larger than the MTU to be 206 passed to the NIC to improve performance, the NIC being responsible 207 for creating smaller segments of size less than or equal to the MTU 208 with correct protocol headers. When referring specifically to TCP/ 209 IP, this feature is often known as TSO (TCP Segmentation Offload). 211 Tunnel endpoint. A component performing encapsulation and 212 decapsulation of packets, such as Ethernet frames or IP datagrams, in 213 Geneve headers. As the ultimate consumer of any tunnel metadata, 214 tunnel endpoints have the highest level of requirements for parsing 215 and interpreting tunnel headers. Tunnel endpoints may consist of 216 either software or hardware implementations or a combination of the 217 two. Tunnel endpoints are frequently a component of an NVE (Network 218 Virtualization Edge) but may also be found in middleboxes or other 219 elements making up an NVO3 Network. 221 VM. Virtual Machine. 223 2. Design Requirements 225 Geneve is designed to support network virtualization use cases, where 226 tunnels are typically established to act as a backplane between the 227 virtual switches residing in hypervisors, physical switches, or 228 middleboxes or other appliances. An arbitrary IP network can be used 229 as an underlay although Clos networks composed using ECMP links are a 230 common choice to provide consistent bisectional bandwidth across all 231 connection points. Figure 1 shows an example of a hypervisor, top of 232 rack switch for connectivity to physical servers, and a WAN uplink 233 connected using Geneve tunnels over a simplified Clos network. These 234 tunnels are used to encapsulate and forward frames from the attached 235 components such as VMs or physical links. 237 +---------------------+ +-------+ +------+ 238 | +--+ +-------+---+ | |Transit|--|Top of|==Physical 239 | |VM|--| | | | +------+ /|Router | | Rack |==Servers 240 | +--+ |Virtual|NIC|---|Top of|/ +-------+\/+------+ 241 | +--+ |Switch | | | | Rack |\ +-------+/\+------+ 242 | |VM|--| | | | +------+ \|Transit| |Uplink| WAN 243 | +--+ +-------+---+ | |Router |--| |=========> 244 +---------------------+ +-------+ +------+ 245 Hypervisor 247 ()===================================() 248 Switch-Switch Geneve Tunnels 250 Figure 1: Sample Geneve Deployment 252 To support the needs of network virtualization, the tunnel protocol 253 should be able to take advantage of the differing (and evolving) 254 capabilities of each type of device in both the underlay and overlay 255 networks. This results in the following requirements being placed on 256 the data plane tunneling protocol: 258 o The data plane is generic and extensible enough to support current 259 and future control planes. 261 o Tunnel components are efficiently implementable in both hardware 262 and software without restricting capabilities to the lowest common 263 denominator. 265 o High performance over existing IP fabrics. 267 These requirements are described further in the following 268 subsections. 270 2.1. Control Plane Independence 272 Although some protocols for network virtualization have included a 273 control plane as part of the tunnel format specification (most 274 notably, the original VXLAN spec prescribed a multicast learning- 275 based control plane), these specifications have largely been treated 276 as describing only the data format. The VXLAN packet format has 277 actually seen a wide variety of control planes built on top of it. 279 There is a clear advantage in settling on a data format: most of the 280 protocols are only superficially different and there is little 281 advantage in duplicating effort. However, the same cannot be said of 282 control planes, which are diverse in very fundamental ways. The case 283 for standardization is also less clear given the wide variety in 284 requirements, goals, and deployment scenarios. 286 As a result of this reality, Geneve aims to be a pure tunnel format 287 specification that is capable of fulfilling the needs of many control 288 planes by explicitly not selecting any one of them. This 289 simultaneously promotes a shared data format and increases the 290 chances that it will not be obsoleted by future control plane 291 enhancements. 293 2.2. Data Plane Extensibility 295 Achieving the level of flexibility needed to support current and 296 future control planes effectively requires an options infrastructure 297 to allow new metadata types to be defined, deployed, and either 298 finalized or retired. Options also allow for differentiation of 299 products by encouraging independent development in each vendor's core 300 specialty, leading to an overall faster pace of advancement. By far 301 the most common mechanism for implementing options is Type-Length- 302 Value (TLV) format. 304 It should be noted that while options can be used to support non- 305 wirespeed control packets, they are equally important on data packets 306 as well to segregate and direct forwarding (for instance, the 307 examples given before of input port based security policies and 308 service interposition both require tags to be placed on data 309 packets). Therefore, while it would be desirable to limit the 310 extensibility to only control packets for the purposes of simplifying 311 the datapath, that would not satisfy the design requirements. 313 2.2.1. Efficient Implementation 315 There is often a conflict between software flexibility and hardware 316 performance that is difficult to resolve. For a given set of 317 functionality, it is obviously desirable to maximize performance. 318 However, that does not mean new features that cannot be run at that 319 speed today should be disallowed. Therefore, for a protocol to be 320 efficiently implementable means that a set of common capabilities can 321 be reasonably handled across platforms along with a graceful 322 mechanism to handle more advanced features in the appropriate 323 situations. 325 The use of a variable length header and options in a protocol often 326 raises questions about whether it is truly efficiently implementable 327 in hardware. To answer this question in the context of Geneve, it is 328 important to first divide "hardware" into two categories: tunnel 329 endpoints and transit devices. 331 Tunnel endpoints must be able to parse the variable header, including 332 any options, and take action. Since these devices are actively 333 participating in the protocol, they are the most affected by Geneve. 335 However, as tunnel endpoints are the ultimate consumers of the data, 336 transmitters can tailor their output to the capabilities of the 337 recipient. As new functionality becomes sufficiently well defined to 338 add to tunnel endpoints, supporting options can be designed using 339 ordering restrictions and other techniques to ease parsing. 341 Options, if present in the packet, MUST be generated and terminated 342 by tunnel endpoints. Transit devices MAY be able to interpret the 343 options, however, as non-terminating devices, transit devices do not 344 originate or terminate the Geneve packet, hence MUST NOT modify 345 Geneve headers and MUST NOT insert or delete options, which is the 346 responsibility of tunnel endpoints. The participation of transit 347 devices in interpreting options is OPTIONAL. 349 Further, either tunnel endpoints or transit devices MAY use offload 350 capabilities of NICs such as checksum offload to improve the 351 performance of Geneve packet processing. The presence of a Geneve 352 variable length header SHOULD NOT prevent the tunnel endpoints and 353 transit devices from using such offload capabilities. 355 2.3. Use of Standard IP Fabrics 357 IP has clearly cemented its place as the dominant transport mechanism 358 and many techniques have evolved over time to make it robust, 359 efficient, and inexpensive. As a result, it is natural to use IP 360 fabrics as a transit network for Geneve. Fortunately, the use of IP 361 encapsulation and addressing is enough to achieve the primary goal of 362 delivering packets to the correct point in the network through 363 standard switching and routing. 365 In addition, nearly all underlay fabrics are designed to exploit 366 parallelism in traffic to spread load across multiple links without 367 introducing reordering in individual flows. These equal cost 368 multipathing (ECMP) techniques typically involve parsing and hashing 369 the addresses and port numbers from the packet to select an outgoing 370 link. However, the use of tunnels often results in poor ECMP 371 performance without additional knowledge of the protocol as the 372 encapsulated traffic is hidden from the fabric by design and only 373 tunnel endpoint addresses are available for hashing. 375 Since it is desirable for Geneve to perform well on these existing 376 fabrics, it is necessary for entropy from encapsulated packets to be 377 exposed in the tunnel header. The most common technique for this is 378 to use the UDP source port, which is discussed further in 379 Section 3.3. 381 3. Geneve Encapsulation Details 383 The Geneve packet format consists of a compact tunnel header 384 encapsulated in UDP over either IPv4 or IPv6. A small fixed tunnel 385 header provides control information plus a base level of 386 functionality and interoperability with a focus on simplicity. This 387 header is then followed by a set of variable options to allow for 388 future innovation. Finally, the payload consists of a protocol data 389 unit of the indicated type, such as an Ethernet frame. Section 3.1 390 and Section 3.2 illustrate the Geneve packet format transported (for 391 example) over Ethernet along with an Ethernet payload. 393 3.1. Geneve Packet Format Over IPv4 395 0 1 2 3 396 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 397 Outer Ethernet Header: 398 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 399 | Outer Destination MAC Address | 400 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 401 | Outer Destination MAC Address | Outer Source MAC Address | 402 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 403 | Outer Source MAC Address | 404 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 405 |Optional Ethertype=C-Tag 802.1Q| Outer VLAN Tag Information | 406 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 407 | Ethertype=0x0800 | 408 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 410 Outer IPv4 Header: 411 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 412 |Version| IHL |Type of Service| Total Length | 413 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 414 | Identification |Flags| Fragment Offset | 415 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 416 | Time to Live |Protocol=17 UDP| Header Checksum | 417 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 418 | Outer Source IPv4 Address | 419 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 420 | Outer Destination IPv4 Address | 421 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 423 Outer UDP Header: 424 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 425 | Source Port = xxxx | Dest Port = 6081 | 426 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 427 | UDP Length | UDP Checksum | 428 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 430 Geneve Header: 431 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 432 |Ver| Opt Len |O|C| Rsvd. | Protocol Type | 433 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 434 | Virtual Network Identifier (VNI) | Reserved | 435 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 436 | Variable Length Options | 437 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 439 Inner Ethernet Header (example payload): 440 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 441 | Inner Destination MAC Address | 442 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 443 | Inner Destination MAC Address | Inner Source MAC Address | 444 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 445 | Inner Source MAC Address | 446 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 447 |Optional Ethertype=C-Tag 802.1Q| Inner VLAN Tag Information | 448 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 450 Payload: 451 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 452 | Ethertype of Original Payload | | 453 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | 454 | Original Ethernet Payload | 455 | | 456 | (Note that the original Ethernet Frame's FCS is not included) | 457 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 459 Frame Check Sequence: 460 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 461 | New FCS (Frame Check Sequence) for Outer Ethernet Frame | 462 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 464 3.2. Geneve Packet Format Over IPv6 466 0 1 2 3 467 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 468 Outer Ethernet Header: 469 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 470 | Outer Destination MAC Address | 471 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 472 | Outer Destination MAC Address | Outer Source MAC Address | 473 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 474 | Outer Source MAC Address | 475 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 476 |Optional Ethertype=C-Tag 802.1Q| Outer VLAN Tag Information | 477 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 478 | Ethertype=0x86DD | 479 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 481 Outer IPv6 Header: 482 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 483 |Version| Traffic Class | Flow Label | 484 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 485 | Payload Length | NxtHdr=17 UDP | Hop Limit | 486 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 487 | | 488 + + 489 | | 490 + Outer Source IPv6 Address + 491 | | 492 + + 493 | | 494 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 495 | | 496 + + 497 | | 498 + Outer Destination IPv6 Address + 499 | | 500 + + 501 | | 502 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 504 Outer UDP Header: 505 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 506 | Source Port = xxxx | Dest Port = 6081 | 507 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 508 | UDP Length | UDP Checksum | 509 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 511 Geneve Header: 512 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 513 |Ver| Opt Len |O|C| Rsvd. | Protocol Type | 514 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 515 | Virtual Network Identifier (VNI) | Reserved | 516 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 517 | Variable Length Options | 518 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 520 Inner Ethernet Header (example payload): 521 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 522 | Inner Destination MAC Address | 523 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 524 | Inner Destination MAC Address | Inner Source MAC Address | 525 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 526 | Inner Source MAC Address | 527 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 528 |Optional Ethertype=C-Tag 802.1Q| Inner VLAN Tag Information | 529 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 531 Payload: 532 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 533 | Ethertype of Original Payload | | 534 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | 535 | Original Ethernet Payload | 536 | | 537 | (Note that the original Ethernet Frame's FCS is not included) | 538 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 540 Frame Check Sequence: 541 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 542 | New FCS (Frame Check Sequence) for Outer Ethernet Frame | 543 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 545 3.3. UDP Header 547 The use of an encapsulating UDP [RFC0768] header follows the 548 connectionless semantics of Ethernet and IP in addition to providing 549 entropy to routers performing ECMP. The header fields are therefore 550 interpreted as follows: 552 Source port: A source port selected by the originating tunnel 553 endpoint. This source port SHOULD be the same for all packets 554 belonging to a single encapsulated flow to prevent reordering due 555 to the use of different paths. To encourage an even distribution 556 of flows across multiple links, the source port SHOULD be 557 calculated using a hash of the encapsulated packet headers using, 558 for example, a traditional 5-tuple. Since the port represents a 559 flow identifier rather than a true UDP connection, the entire 560 16-bit range MAY be used to maximize entropy. 562 Dest port: IANA has assigned port 6081 as the fixed well-known 563 destination port for Geneve. Although the well-known value should 564 be used by default, it is RECOMMENDED that implementations make 565 this configurable. The chosen port is used for identification of 566 Geneve packets and MUST NOT be reversed for different ends of a 567 connection as is done with TCP. 569 UDP length: The length of the UDP packet including the UDP header. 571 UDP checksum: In order to protect the Geneve header, options and 572 payload from potential data corruption, UDP checksum SHOULD be 573 generated as specified in [RFC0768] and [RFC1112] when Geneve is 574 encapsulated in IPv4. To protect the IP header, Geneve header, 575 options and payload from potential data corruption, the UDP 576 checksum MUST be generated by default as specified in [RFC0768] 577 and [RFC2460] when Geneve is encapsulated in IPv6. Upon receiving 578 such packets with non-zero UDP checksum, the receiving tunnel 579 endpoints MUST validate the checksum. If the checksum is not 580 correct, the packet MUST be dropped, otherwise the packet MUST be 581 accepted for decapsulation. 583 Under certain conditions, the UDP checksum MAY be set to zero on 584 transmit for packets encapsulated in both IPv4 and IPv6 [RFC6935]. 585 See Section 4.3 for additional requirements that apply for using 586 zero UDP checksum with IPv4 and IPv6. Disabling the use of UDP 587 checksums is an operational consideration that should take into 588 account the risks and effects of packet corruption. 590 3.4. Tunnel Header Fields 592 Ver (2 bits): The current version number is 0. Packets received by 593 a tunnel endpoint with an unknown version MUST be dropped. 594 Transit devices interpreting Geneve packets with an unknown 595 version number MUST treat them as UDP packets with an unknown 596 payload. 598 Opt Len (6 bits): The length of the options fields, expressed in 599 four byte multiples, not including the eight byte fixed tunnel 600 header. This results in a minimum total Geneve header size of 8 601 bytes and a maximum of 260 bytes. The start of the payload 602 headers can be found using this offset from the end of the base 603 Geneve header. 605 O (1 bit): Control packet. This packet contains a control message. 606 Control messages are sent between tunnel endpoints. Tunnel 607 Endpoints MUST NOT forward the payload and transit devices MUST 608 NOT attempt to interpret it. Since these are infrequent control 609 messages, it is RECOMMENDED that tunnel endpoints direct these 610 packets to a high priority control queue (for example, to direct 611 the packet to a general purpose CPU from a forwarding ASIC or to 612 separate out control traffic on a NIC). Transit devices MUST NOT 613 alter forwarding behavior on the basis of this bit, such as ECMP 614 link selection. 616 C (1 bit): Critical options present. One or more options has the 617 critical bit set (see Section 3.5). If this bit is set then 618 tunnel endpoints MUST parse the options list to interpret any 619 critical options. On tunnel endpoints where option parsing is not 620 supported the packet MUST be dropped on the basis of the 'C' bit 621 in the base header. If the bit is not set tunnel endpoints MAY 622 strip all options using 'Opt Len' and forward the decapsulated 623 packet. Transit devices MUST NOT drop packets on the basis of 624 this bit. 626 The critical bit allows hardware implementations the flexibility 627 to handle options processing in the hardware fastpath or in the 628 exception (slow) path without the need to process all the options. 629 For example, a critical option such as secure hash to provide 630 Geneve header integrity check must be processed by tunnel 631 endpoints and typically processed in the hardware fastpath. 633 Rsvd. (6 bits): Reserved field, which MUST be zero on transmission 634 and MUST be ignored on receipt. 636 Protocol Type (16 bits): The type of the protocol data unit 637 appearing after the Geneve header. This follows the EtherType 638 [ETYPES] convention with Ethernet itself being represented by the 639 value 0x6558. 641 Virtual Network Identifier (VNI) (24 bits): An identifier for a 642 unique element of a virtual network. In many situations this may 643 represent an L2 segment, however, the control plane defines the 644 forwarding semantics of decapsulated packets. The VNI MAY be used 645 as part of ECMP forwarding decisions or MAY be used as a mechanism 646 to distinguish between overlapping address spaces contained in the 647 encapsulated packet when load balancing across CPUs. 649 Reserved (8 bits): Reserved field which MUST be zero on transmission 650 and ignored on receipt. 652 Transit devices MUST maintain consistent forwarding behavior 653 irrespective of the value of 'Opt Len', including ECMP link 654 selection. These devices SHOULD be able to forward packets 655 containing options without resorting to a slow path. 657 3.5. Tunnel Options 659 0 1 2 3 660 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 661 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 662 | Option Class | Type |R|R|R| Length | 663 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 664 | Variable Option Data | 665 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 667 Geneve Option 669 The base Geneve header is followed by zero or more options in Type- 670 Length-Value format. Each option consists of a four byte option 671 header and a variable amount of option data interpreted according to 672 the type. 674 Option Class (16 bits): Namespace for the 'Type' field. IANA will 675 be requested to create a "Geneve Option Class" registry to 676 allocate identifiers for organizations, technologies, and vendors 677 that have an interest in creating types for options. Each 678 organization may allocate types independently to allow 679 experimentation and rapid innovation. It is expected that over 680 time certain options will become well known and a given 681 implementation may use option types from a variety of sources. In 682 addition, IANA will be requested to reserve specific ranges for 683 standardized and experimental options. 685 Type (8 bits): Type indicating the format of the data contained in 686 this option. Options are primarily designed to encourage future 687 extensibility and innovation and so standardized forms of these 688 options will be defined in a separate document. 690 The high order bit of the option type indicates that this is a 691 critical option. If the receiving tunnel endpoint does not 692 recognize this option and this bit is set then the packet MUST be 693 dropped. If the 'C' bit (critical bit) is set in any option then 694 the 'C' bit in the Geneve base header MUST also be set. Transit 695 devices MUST NOT drop packets on the basis of this bit. The 696 following figure shows the location of the 'C' bit in the 'Type' 697 field: 699 0 1 2 3 4 5 6 7 8 700 +-+-+-+-+-+-+-+-+ 701 |C| Type | 702 +-+-+-+-+-+-+-+-+ 704 The requirement to drop a packet with an unknown option with the 705 'C' bit set applies to the entire tunnel endpoint system and not a 706 particular component of the implementation. For example, in a 707 system comprised of a forwarding ASIC and a general purpose CPU, 708 this does not mean that the packet must be dropped in the ASIC. 709 An implementation may send the packet to the CPU using a rate- 710 limited control channel for slow-path exception handling. 712 R (3 bits): Option control flags reserved for future use. MUST be 713 zero on transmission and ignored on receipt. 715 Length (5 bits): Length of the option, expressed in four byte 716 multiples excluding the option header. The total length of each 717 option may be between 4 and 128 bytes. A value of 0 in the Length 718 field implies an option with only the option header without the 719 variable option data. Packets in which the total length of all 720 options is not equal to the 'Opt Len' in the base header are 721 invalid and MUST be silently dropped if received by a tunnel 722 endpoint that processes the options. 724 Variable Option Data: Option data interpreted according to 'Type'. 726 3.5.1. Options Processing 728 Geneve options are intended to be originated and processed by tunnel 729 endpoints. However, options MAY be interpreted by transit devices 730 along the tunnel path. Transit devices not interpreting Geneve 731 headers (that may or may not include Options) SHOULD handle Geneve 732 packets as any other UDP packet and maintain consistent forwarding 733 behavior. 735 In tunnel endpoints, the generation and interpretation of options is 736 determined by the control plane, which is out of the scope of this 737 document. However, to ensure interoperability between heterogeneous 738 devices some requirements are imposed on options and the devices that 739 process them: 741 o Receiving tunnel endpoints MUST drop packets containing unknown 742 options with the 'C' bit set in the option type. Conversely, 743 transit devices MUST NOT drop packets as a result of encountering 744 unknown options, including those with the 'C' bit set. 746 o Some options may be defined in such a way that the position in the 747 option list is significant. Options MUST NOT be changed by 748 transit devices. 750 o An option SHOULD NOT be dependent upon any other option in the 751 packet, i.e., options can be processed independent of one another. 752 An option MUST NOT affect the parsing or interpretation of any 753 other option. 755 When designing a Geneve option, it is important to consider how the 756 option will evolve in the future. Once an option is defined it is 757 reasonable to expect that implementations may come to depend on a 758 specific behavior. As a result, the scope of any future changes must 759 be carefully described upfront. 761 Unexpectedly significant interoperability issues may result from 762 changing the length of an option that was defined to be a certain 763 size. A particular option is specified to have either a fixed 764 length, which is constant, or a variable length, which may change 765 over time or for different use cases. This property is part of the 766 definition of the option and conveyed by the 'Type'. For fixed 767 length options, some implementations may choose to ignore the length 768 field in the option header and instead parse based on the well known 769 length associated with the type. In this case, redefining the length 770 will impact not only parsing of the option in question but also any 771 options that follow. Therefore, options that are defined to be fixed 772 length in size MUST NOT be redefined to a different length. Instead, 773 a new 'Type' should be allocated. 775 Options may be processed by NIC hardware utilizing offloads (e.g. 776 LSO and LRO) as described in Section 4.6. Careful consideration 777 should be given to how the offload capabilities outlined in 778 Section 4.6 impact an option's design. 780 4. Implementation and Deployment Considerations 782 4.1. Applicability Statement 784 Geneve is a network virtualization overlay encapsulation protocol 785 designed to establish tunnels between NVEs over an existing IP 786 network. It is intended for use in public or private data center 787 environments, for deploying multi-tenant overlay networks over an 788 existing IP underlay network. 790 Geneve is an UDP based encapsulation protocol transported over 791 existing IPv4 and IPv6 networks. Hence, as an UDP based protocol, 792 Geneve needs to meet the UDP usage guidelines as specified in 793 [RFC8085]. The applicability of these guidelines are dependent on 794 the underlay IP network and the nature of Geneve payload protocol 795 (example TCP/IP, IP/Ethernet). 797 [RFC8085] outlines two applicability scenarios for UDP applications, 798 1) general Internet and 2) controlled environment. The controlled 799 environment means a single administrative domain or adjacent set of 800 cooperating domains. A network in a controlled environment can be 801 managed to operate under certain conditions whereas in general 802 Internet this cannot be done. Hence requirements for a tunnel 803 protocol operating under a controlled environment can be less 804 restrictive than the requirements of general internet. 806 Geneve is intended to be deployed in a data center network 807 environment operated by a single operator or adjacent set of 808 cooperating network operators that fits with the definition of 809 controlled environments in [RFC8085]. 811 For the purpose of this document, a traffic-managed controlled 812 environment (TMCE) is defined as an IP network that is traffic- 813 engineered and/or otherwise managed (e.g., via use of traffic rate 814 limiters) to avoid congestion. The concept of TMCE is outlined in 815 [RFC8086]. Significant portions of text in Section 4.1 through 816 Section 4.3 are based on [RFC8086] as applicable to Geneve. 818 It is the responsibility of the operator to ensure that the 819 guidelines/requirements in this section are followed as applicable to 820 their Geneve deployment(s). 822 4.2. Congestion Control Functionality 824 Geneve does not natively provide congestion control functionality and 825 relies on the payload protocol traffic for congestion control. As 826 such Geneve MUST be used with congestion controlled traffic or within 827 a network that is traffic managed to avoid congestion (TMCE). An 828 operator of a traffic managed network (TMCE) may avoid congestion by 829 careful provisioning of their networks, rate-limiting of user data 830 traffic and traffic engineering according to path capacity. 832 4.3. UDP Checksum 834 In order to provide integrity of Geneve headers, options and payload, 835 for example to avoid mis-delivery of payload to different tenant 836 systems in case of data corruption, outer UDP checksum SHOULD be used 837 with Geneve when transported over IPv4. An operator MAY choose to 838 disable UDP checksum and use zero checksum if Geneve packet integrity 839 is provided by other data integrity mechanisms such as IPsec or 840 additional checksums or if one of the conditions in Section 4.3.1 a, 841 b, c are met. 843 By default, UDP checksum MUST be used when Geneve is transported over 844 IPv6. A tunnel endpoint MAY be configured for use with zero UDP 845 checksum if additional requirements in Section 4.3.1 are met. 847 4.3.1. UDP Zero Checksum Handling with IPv6 849 When Geneve is used over IPv6, UDP checksum is used to protect IPv6 850 headers, UDP headers and Geneve headers, options and payload from 851 potential data corruption. As such by default Geneve MUST use UDP 852 checksum when transported over IPv6. An operator MAY choose to 853 configure to operate with zero UDP checksum if operating in a traffic 854 managed controlled environment as stated in Section 4.1 if one of the 855 following conditions are met. 857 a. It is known that the packet corruption is exceptionally unlikely 858 (perhaps based on knowledge of equipment types in their underlay 859 network) and the operator is willing to take a risk of undetected 860 packet corruption 862 b. It is judged through observational measurements (perhaps through 863 historic or current traffic flows that use non zero checksum) 864 that the level of packet corruption is tolerably low and where 865 the operator is willing to take the risk of undetected 866 corruption. 868 c. Geneve payload is carrying applications that are tolerant of 869 misdelivered or corrupted packets (perhaps through higher layer 870 checksum validation and/or reliability through retransmission) 872 In addition Geneve tunnel implementations using Zero UDP checksum 873 MUST meet the following requirements: 875 1. Use of UDP checksum over IPv6 MUST be the default configuration 876 for all Geneve tunnels. 878 2. If Geneve is used with zero UDP checksum over IPv6 then such 879 tunnel endpoint implementation MUST meet all the requirements 880 specified in section 4 of [RFC6936] and requirements 1 as 881 specified in section 5 of [RFC6936]. 883 3. The Geneve tunnel endpoint that decapsulates the tunnel SHOULD 884 check the source and destination IPv6 addresses are valid for the 885 Geneve tunnel that is configured to receive Zero UDP checksum and 886 discard other packets for which such check fails. 888 4. The Geneve tunnel endpoint that encapsulates the tunnel MAY use 889 different IPv6 source addresses for each Geneve tunnel that uses 890 Zero UDP checksum mode in order to strengthen the decapsulator's 891 check of the IPv6 source address (i.e the same IPv6 source 892 address is not to be used with more than one IPv6 destination 893 address, irrespective of whether that destination address is a 894 unicast or multicast address). When this is not possible, it is 895 RECOMMENDED to use each source address for as few Geneve tunnels 896 that use zero UDP checksum as is feasible. 898 5. Measures SHOULD be taken to prevent Geneve traffic over IPv6 with 899 zero UDP checksum from escaping into the general Internet. 900 Examples of such measures include employing packet filters at the 901 Gateways or edge of Geneve network and/or keeping logical or 902 physical separation of Geneve network from networks carrying 903 General Internet. 905 The above requirements do not change either the requirements 906 specified in [RFC2460] as modified by [RFC6935] or the requirements 907 specified in [RFC6936]. 909 The requirement to check the source IPv6 address in addition to the 910 destination IPv6 address, plus the recommendation against reuse of 911 source IPv6 addresses among Geneve tunnels collectively provide some 912 mitigation for the absence of UDP checksum coverage of the IPv6 913 header. A traffic-managed controlled environment that satisfies at 914 least one of three conditions listed at the beginning of this section 915 provides additional assurance. 917 Editorial Note (The following paragraph to be removed by the RFC 918 Editor before publication) 920 It was discussed during TSVART early review if the level of 921 requirement for using different IPv6 source addresses for different 922 tunnel destinations would need to be "MAY" or "SHOULD". The 923 discussion concluded that it was appropriate to keep this as "MAY", 924 since it was considered not realistic for control planes having to 925 maintain a high level of state on a per tunnel destination basis. In 926 addition, the text above provides sufficient guidance to operators 927 and implementors on possible mitigations. 929 4.4. Encapsulation of Geneve in IP 931 As an IP-based tunnel protocol, Geneve shares many properties and 932 techniques with existing protocols. The application of some of these 933 are described in further detail, although in general most concepts 934 applicable to the IP layer or to IP tunnels generally also function 935 in the context of Geneve. 937 4.4.1. IP Fragmentation 939 It is strongly RECOMMENDED that Path MTU Discovery ([RFC1191], 940 [RFC8201]) be used by setting the DF bit in the IP header when Geneve 941 packets are transmitted over IPv4 (this is the default with IPv6). 942 The use of Path MTU Discovery on the transit network provides the 943 encapsulating tunnel endpoint with soft-state about the link that it 944 may use to prevent or minimize fragmentation depending on its role in 945 the virtualized network. The NVE control plane MAY use configuration 946 mechanism or path discovery information to maintain the MTU size of 947 the tunnel link(s) associated with the tunnel endpoint, so if a 948 tenant system sends large packets that when encapsulated exceed the 949 MTU size of the tunnel link, the tunnel endpoint can discard such 950 packets and send exception messages to the tenant system(s). If the 951 tunnel endpoint is associated with a routing or forwarding function 952 and/or has the capability to send ICMP messages, the encapsulating 953 tunnel endpoint MAY send ICMP fragmentation needed [RFC0792] or 954 Packet Too Big [RFC4443] messages to the tenant system(s). For 955 example, recommendations/guidance for handling fragmentation in 956 similar overlay encapsulation services like PWE3 are provided in 957 section 5.3 of [RFC3985]. 959 Note that some implementations may not be capable of supporting 960 fragmentation or other less common features of the IP header, such as 961 options and extension headers. For example, some of the issues 962 associated with MTU size and fragmentation in IP tunneling and use of 963 ICMP messages is outlined in section 4.2 of 964 [I-D.ietf-intarea-tunnels]. 966 Editorial Note (The following paragraph to be removed by the RFC 967 Editor before publication) 969 It was discussed during TSVART early review if the level of 970 requirement for maintaining tunnel MTU at the ingress has to be "MAY" 971 or "SHOULD". The discussion concluded that it was appropriate to 972 leave this as "MAY", considering the high level of state to be 973 maintained. 975 4.4.2. DSCP, ECN and TTL 977 When encapsulating IP (including over Ethernet) packets in Geneve, 978 there are several considerations for propagating DSCP and ECN bits 979 from the inner header to the tunnel on transmission and the reverse 980 on reception. 982 [RFC2983] provides guidance for mapping DSCP between inner and outer 983 IP headers. Network virtualization is typically more closely aligned 984 with the Pipe model described, where the DSCP value on the tunnel 985 header is set based on a policy (which may be a fixed value, one 986 based on the inner traffic class, or some other mechanism for 987 grouping traffic). Aspects of the Uniform model (which treats the 988 inner and outer DSCP value as a single field by copying on ingress 989 and egress) may also apply, such as the ability to remark the inner 990 header on tunnel egress based on transit marking. However, the 991 Uniform model is not conceptually consistent with network 992 virtualization, which seeks to provide strong isolation between 993 encapsulated traffic and the physical network. 995 [RFC6040] describes the mechanism for exposing ECN capabilities on IP 996 tunnels and propagating congestion markers to the inner packets. 997 This behavior MUST be followed for IP packets encapsulated in Geneve. 999 Though Uniform or Pipe models could be used for TTL (or Hop Limit in 1000 case of IPv6) handling when tunneling IP packets, Pipe model is more 1001 aligned with network virtualization. [RFC2003] provides guidance on 1002 handling TTL between inner IP header and outer IP tunnels; this model 1003 is more aligned with the Pipe model and is recommended for use with 1004 Geneve for network virtualization applications. 1006 4.4.3. Broadcast and Multicast 1008 Geneve tunnels may either be point-to-point unicast between two 1009 tunnel endpoints or may utilize broadcast or multicast addressing. 1010 It is not required that inner and outer addressing match in this 1011 respect. For example, in physical networks that do not support 1012 multicast, encapsulated multicast traffic may be replicated into 1013 multiple unicast tunnels or forwarded by policy to a unicast location 1014 (possibly to be replicated there). 1016 With physical networks that do support multicast it may be desirable 1017 to use this capability to take advantage of hardware replication for 1018 encapsulated packets. In this case, multicast addresses may be 1019 allocated in the physical network corresponding to tenants, 1020 encapsulated multicast groups, or some other factor. The allocation 1021 of these groups is a component of the control plane and therefore 1022 outside of the scope of this document. When physical multicast is in 1023 use, the 'C' bit in the Geneve header may be used with groups of 1024 devices with heterogeneous capabilities as each device can interpret 1025 only the options that are significant to it if they are not critical. 1027 In addition, [RFC8293] provides examples of various mechanisms that 1028 can be used for multicast handling in network virtualization overlay 1029 networks. 1031 4.4.4. Unidirectional Tunnels 1033 Generally speaking, a Geneve tunnel is a unidirectional concept. IP 1034 is not a connection oriented protocol and it is possible for two 1035 tunnel endpoints to communicate with each other using different paths 1036 or to have one side not transmit anything at all. As Geneve is an 1037 IP-based protocol, the tunnel layer inherits these same 1038 characteristics. 1040 It is possible for a tunnel to encapsulate a protocol, such as TCP, 1041 which is connection oriented and maintains session state at that 1042 layer. In addition, implementations MAY model Geneve tunnels as 1043 connected, bidirectional links, such as to provide the abstraction of 1044 a virtual port. In both of these cases, bidirectionality of the 1045 tunnel is handled at a higher layer and does not affect the operation 1046 of Geneve itself. 1048 4.5. Constraints on Protocol Features 1050 Geneve is intended to be flexible to a wide range of current and 1051 future applications. As a result, certain constraints may be placed 1052 on the use of metadata or other aspects of the protocol in order to 1053 optimize for a particular use case. For example, some applications 1054 may limit the types of options which are supported or enforce a 1055 maximum number or length of options. Other applications may only 1056 handle certain encapsulated payload types, such as Ethernet or IP. 1057 This could be either globally throughout the system or, for example, 1058 restricted to certain classes of devices or network paths. 1060 These constraints may be communicated to tunnel endpoints either 1061 explicitly through a control plane or implicitly by the nature of the 1062 application. As Geneve is defined as a data plane protocol that is 1063 control plane agnostic, the exact mechanism is not defined in this 1064 document. 1066 4.5.1. Constraints on Options 1068 While Geneve options are more flexible, a control plane may restrict 1069 the number of option TLVs as well as the order and size of the TLVs, 1070 between tunnel endpoints, to make it simpler for a data plane 1071 implementation in software or hardware to handle 1072 [I-D.ietf-nvo3-encap]. For example, there may be some critical 1073 information such as a secure hash that must be processed in a certain 1074 order to provide lowest latency. 1076 A control plane may negotiate a subset of option TLVs and certain TLV 1077 ordering, as well may limit the total number of option TLVs present 1078 in the packet, for example, to accommodate hardware capable of 1079 processing fewer options [I-D.ietf-nvo3-encap]. Hence, a control 1080 plane needs to have the ability to describe the supported TLVs subset 1081 and their order to the tunnel endpoints. In the absence of a control 1082 plane, alternative configuration mechanisms may be used for this 1083 purpose. The exact mechanism is not defined in this document. 1085 4.6. NIC Offloads 1087 Modern NICs currently provide a variety of offloads to enable the 1088 efficient processing of packets. The implementation of many of these 1089 offloads requires only that the encapsulated packet be easily parsed 1090 (for example, checksum offload). However, optimizations such as LSO 1091 and LRO involve some processing of the options themselves since they 1092 must be replicated/merged across multiple packets. In these 1093 situations, it is desirable to not require changes to the offload 1094 logic to handle the introduction of new options. To enable this, 1095 some constraints are placed on the definitions of options to allow 1096 for simple processing rules: 1098 o When performing LSO, a NIC MUST replicate the entire Geneve header 1099 and all options, including those unknown to the device, onto each 1100 resulting segment. However, a given option definition may 1101 override this rule and specify different behavior in supporting 1102 devices. Conversely, when performing LRO, a NIC MAY assume that a 1103 binary comparison of the options (including unknown options) is 1104 sufficient to ensure equality and MAY merge packets with equal 1105 Geneve headers. 1107 o Options MUST NOT be reordered during the course of offload 1108 processing, including when merging packets for the purpose of LRO. 1110 o NICs performing offloads MUST NOT drop packets with unknown 1111 options, including those marked as critical, unless explicitly 1112 configured. 1114 There is no requirement that a given implementation of Geneve employ 1115 the offloads listed as examples above. However, as these offloads 1116 are currently widely deployed in commercially available NICs, the 1117 rules described here are intended to enable efficient handling of 1118 current and future options across a variety of devices. 1120 4.7. Inner VLAN Handling 1122 Geneve is capable of encapsulating a wide range of protocols and 1123 therefore a given implementation is likely to support only a small 1124 subset of the possibilities. However, as Ethernet is expected to be 1125 widely deployed, it is useful to describe the behavior of VLANs 1126 inside encapsulated Ethernet frames. 1128 As with any protocol, support for inner VLAN headers is OPTIONAL. In 1129 many cases, the use of encapsulated VLANs may be disallowed due to 1130 security or implementation considerations. However, in other cases 1131 trunking of VLAN frames across a Geneve tunnel can prove useful. As 1132 a result, the processing of inner VLAN tags upon ingress or egress 1133 from a tunnel endpoint is based upon the configuration of the tunnel 1134 endpoint and/or control plane and not explicitly defined as part of 1135 the data format. 1137 5. Interoperability Issues 1139 Viewed exclusively from the data plane, Geneve does not introduce any 1140 interoperability issues as it appears to most devices as UDP packets. 1141 However, as there are already a number of tunnel protocols deployed 1142 in network virtualization environments, there is a practical question 1143 of transition and coexistence. 1145 Since Geneve is a superset of the functionality of the most common 1146 protocols used for network virtualization (VXLAN,NVGRE) it should be 1147 straightforward to port an existing control plane to run on top of it 1148 with minimal effort. With both the old and new packet formats 1149 supporting the same set of capabilities, there is no need for a hard 1150 transition - tunnel endpoints directly communicating with each other 1151 use any common protocol, which may be different even within a single 1152 overall system. As transit devices are primarily forwarding packets 1153 on the basis of the IP header, all protocols appear similar and these 1154 devices do not introduce additional interoperability concerns. 1156 To assist with this transition, it is strongly suggested that 1157 implementations support simultaneous operation of both Geneve and 1158 existing tunnel protocols as it is expected to be common for a single 1159 node to communicate with a mixture of other nodes. Eventually, older 1160 protocols may be phased out as they are no longer in use. 1162 6. Security Considerations 1164 As encapsulated within an UDP/IP packet, Geneve does not have any 1165 inherent security mechanisms. As a result, an attacker with access 1166 to the underlay network transporting the IP packets has the ability 1167 to snoop or inject packets. Compromised tunnel endpoints may also 1168 spoof identifiers in the tunnel header to gain access to networks 1169 owned by other tenants. 1171 Within a particular security domain, such as a data center operated 1172 by a single service provider, the most common and highest performing 1173 security mechanism is isolation of trusted components. Tunnel 1174 traffic can be carried over a separate VLAN and filtered at any 1175 untrusted boundaries. In addition, tunnel endpoints should only be 1176 operated in environments controlled by the service provider, such as 1177 the hypervisor itself rather than within a customer VM. 1179 When crossing an untrusted link, such as the public Internet, IPsec 1180 [RFC4301] may be used to provide authentication and/or encryption of 1181 the IP packets formed as part of Geneve encapsulation. 1183 Geneve does not otherwise affect the security of the encapsulated 1184 packets. As per the guidelines of BCP 72 [RFC3552], the following 1185 sections describe potential security risks that may be applicable to 1186 Geneve deployments and approaches to mitigate such risks. It is also 1187 noted that not all such risks are applicable to all Geneve deployment 1188 scenarios, i.e., only a subset may be applicable to certain 1189 deployments. So an operator has to make an assessment based on their 1190 network environment and determine the risks that are applicable to 1191 their specific environment and use appropriate mitigation approaches 1192 as applicable. 1194 6.1. Data Confidentiality 1196 Geneve is a network virtualization overlay encapsulation protocol 1197 designed to establish tunnels between NVEs over an existing IP 1198 network. It can be used to deploy multi-tenant overlay networks over 1199 an existing IP underlay network in a public or private data center. 1200 The overlay service is typically provided by a service provider, for 1201 example a cloud services provider or a private data center operator, 1202 this may or not may be the same provider as an underlay service 1203 provider. Due to the nature of multi-tenancy in such environments, a 1204 tenant system may expect data confidentiality to ensure its packet 1205 data is not tampered with (active attack) in transit or a target of 1206 unauthorized monitoring (passive attack). A tenant may expect the 1207 overlay service provider to provide data confidentiality as part of 1208 the service or a tenant may bring its own data confidentiality 1209 mechanisms like IPsec or TLS to protect the data end to end between 1210 its tenant systems. 1212 If an operator determines data confidentiality is necessary in their 1213 environment based on their risk analysis, for example as in multi- 1214 tenant environments, then an encryption mechanism SHOULD be used to 1215 encrypt the tenant data end to end between the NVEs. The NVEs may 1216 use existing well established encryption mechanisms such as IPsec, 1217 DTLS, etc. 1219 6.1.1. Inter-Data Center Traffic 1221 A tenant system in a customer premises (private data center) may want 1222 to connect to tenant systems on their tenant overlay network in a 1223 public cloud data center or a tenant may want to have its tenant 1224 systems located in multiple geographically separated data centers for 1225 high availability. Geneve data traffic between tenant systems across 1226 such separated networks should be protected from threats when 1227 traversing public networks. Any Geneve overlay data leaving the data 1228 center network beyond the operator's security domain SHOULD be 1229 secured by encryption mechanisms such as IPsec or other VPN 1230 mechanisms to protect the communications between the NVEs when they 1231 are geographically separated over untrusted network links. 1232 Specification of data protection mechanisms employed between data 1233 centers is beyond the scope of this document. 1235 6.2. Data Integrity 1237 Geneve encapsulation is used between NVEs to establish overlay 1238 tunnels over an existing IP underlay network. In a multi-tenant data 1239 center, a rogue or compromised tenant system may try to launch a 1240 passive attack such as monitoring the traffic of other tenants, or an 1241 active attack such as trying to inject unauthorized Geneve 1242 encapsulated traffic such as spoofing, replay, etc., into the 1243 network. To prevent such attacks, an NVE MUST NOT propagate Geneve 1244 packets beyond the NVE to tenant systems and SHOULD employ packet 1245 filtering mechanisms so as not to forward unauthorized traffic 1246 between TSs in different tenant networks. 1248 A compromised network node or a transit device within a data center 1249 may launch an active attack trying to tamper with the Geneve packet 1250 data between NVEs. Malicious tampering of Geneve header fields may 1251 cause the packet from one tenant to be forwarded to a different 1252 tenant network. If an operator determines the possibility of such 1253 threat in their environment, the operator may choose to employ data 1254 integrity mechanisms between NVEs. In order to prevent such risks, a 1255 data integrity mechanism SHOULD be used in such environments to 1256 protect the integrity of Geneve packets including packet headers, 1257 options and payload on communications between NVE pairs. A 1258 cryptographic data protection mechanism such as IPsec may be used to 1259 provide data integrity protection. A data center operator may choose 1260 to deploy any other data integrity mechanisms as applicable and 1261 supported in their underlay networks. 1263 6.3. Authentication of NVE peers 1265 A rogue network device or a compromised NVE in a data center 1266 environment might be able to spoof Geneve packets as if it came from 1267 a legitimate NVE. In order to mitigate such a risk, an operator 1268 SHOULD use an authentication mechanism, such as IPsec to ensure that 1269 the Geneve packet originated from the intended NVE peer, in 1270 environments where the operator determines spoofing or rogue devices 1271 is a potential threat. Other simpler source checks such as ingress 1272 filtering for VLAN/MAC/IP address, reverse path forwarding checks, 1273 etc., may be used in certain trusted environments to ensure Geneve 1274 packets originated from the intended NVE peer. 1276 6.4. Options Interpretation by Transit Devices 1278 Options, if present in the packet, are generated and terminated by 1279 tunnel endpoints. As indicated in Section 2.2.1, transit devices may 1280 interpret the options. However, if the packet is protected by tunnel 1281 endpoint to tunnel endpoint encryption, for example through IPsec, 1282 transit devices will not have visibility into the Geneve header or 1283 options in the packet. In cases where options are interpreted by 1284 transit devices, the operator MUST ensure that transit devices are 1285 trusted and not compromised. Implementation of a mechanism to ensure 1286 this trust is beyond the scope of this document. 1288 6.5. Multicast/Broadcast 1290 In typical data center networks where IP multicasting is not 1291 supported in the underlay network, multicasting may be supported 1292 using multiple unicast tunnels. The same security requirements as 1293 described in the above sections can be used to protect Geneve 1294 communications between NVE peers. If IP multicasting is supported in 1295 the underlay network and the operator chooses to use it for multicast 1296 traffic among tunnel endpoints, then the operator in such 1297 environments may use data protection mechanisms such as IPsec with 1298 Multicast extensions [RFC5374] to protect multicast traffic among 1299 Geneve NVE groups. 1301 6.6. Control Plane Communications 1303 A Network Virtualization Authority (NVA) as outlined in [RFC8014] may 1304 be used as a control plane for configuring and managing the Geneve 1305 NVEs. The data center operator is expected to use security 1306 mechanisms to protect the communications between the NVA to NVEs and 1307 use authentication mechanisms to detect any rogue or compromised NVEs 1308 within their administrative domain. Data protection mechanisms for 1309 control plane communication or authentication mechanisms between the 1310 NVA and the NVEs is beyond the scope of this document. 1312 7. IANA Considerations 1314 IANA has allocated UDP port 6081 as the well-known destination port 1315 for Geneve. Upon publication, the registry should be updated to cite 1316 this document. The original request was: 1318 Service Name: geneve 1319 Transport Protocol(s): UDP 1320 Assignee: Jesse Gross 1321 Contact: Jesse Gross 1322 Description: Generic Network Virtualization Encapsulation (Geneve) 1323 Reference: This document 1324 Port Number: 6081 1326 In addition, IANA is requested to create a "Geneve Option Class" 1327 registry to allocate Option Classes. This shall be a registry of 1328 16-bit hexadecimal values along with descriptive strings. The 1329 identifiers 0x0-0xFF are to be reserved for standardized options for 1330 allocation by IETF Review [RFC8126] and 0xFFF0-0xFFFF for 1331 Experimental Use. Otherwise, identifiers are to be assigned to any 1332 organization with an interest in creating Geneve options on a First 1333 Come First Served basis. The registry is to be populated with the 1334 following initial values: 1336 +----------------+--------------------------------------+ 1337 | Option Class | Description | 1338 +----------------+--------------------------------------+ 1339 | 0x0000..0x00FF | Unassigned - IETF Review | 1340 | 0x0100 | Linux | 1341 | 0x0101 | Open vSwitch (OVS) | 1342 | 0x0102 | Open Virtual Networking (OVN) | 1343 | 0x0103 | In-band Network Telemetry (INT) | 1344 | 0x0104 | VMware, Inc. | 1345 | 0x0105 | Amazon.com, Inc. | 1346 | 0x0106 | Cisco Systems, Inc. | 1347 | 0x0107 | Oracle Corporation | 1348 | 0x0108..0xFFEF | Unassigned - First Come First Served | 1349 | 0xFFF0..FFFF | Experimental | 1350 +----------------+--------------------------------------+ 1352 8. Contributors 1354 The following individuals were authors of an earlier version of this 1355 document and made significant contributions: 1357 Pankaj Garg 1358 Microsoft Corporation 1359 1 Microsoft Way 1360 Redmond, WA 98052 1361 USA 1363 Email: pankajg@microsoft.com 1365 Chris Wright 1366 Red Hat Inc. 1367 1801 Varsity Drive 1368 Raleigh, NC 27606 1369 USA 1371 Email: chrisw@redhat.com 1373 Puneet Agarwal 1374 Innovium, Inc. 1375 6001 America Center Drive 1376 San Jose, CA 95002 1377 USA 1378 Email: puneet@innovium.com 1380 Kenneth Duda 1381 Arista Networks 1382 5453 Great America Parkway 1383 Santa Clara, CA 95054 1384 USA 1386 Email: kduda@arista.com 1388 Dinesh G. Dutt 1389 Cumulus Networks 1390 140C S. Whisman Road 1391 Mountain View, CA 94041 1392 USA 1394 Email: ddutt@cumulusnetworks.com 1396 Jon Hudson 1397 Independent 1399 Email: jon.hudson@gmail.com 1401 Ariel Hendel 1402 Facebook, Inc. 1403 1 Hacker Way 1404 Menlo Park, CA 94025 1405 USA 1407 Email: ahendel@fb.com 1409 9. Acknowledgements 1411 The authors wish to thank Martin Casado, Bruce Davie and Dave Thaler 1412 for their input, feedback, and helpful suggestions. 1414 The authors would like to thank Magnus Nystrom for his reviews and 1415 feedback during the SECDIR early review. 1417 Thanks to Daniel Migault, Anoop Ghanwani, Greg Mirksy, and Tal 1418 Mizrahi for their reviews, comments and feedback during the Working 1419 Group Last Call process. 1421 The authors would like to thank David Black for his detailed reviews 1422 and valuable inputs during the TSVART early review. 1424 Thanks to Sami Boutros for his inputs and helpful feedback. 1426 The authors would like to thank Matthew Bocci, Sam Aldrin, Benson 1427 Schliesser, Martin Vigoureux, and Alia Atlas for their guidance 1428 throughout the process. 1430 10. References 1432 10.1. Normative References 1434 [RFC0768] Postel, J., "User Datagram Protocol", STD 6, RFC 768, 1435 DOI 10.17487/RFC0768, August 1980, 1436 . 1438 [RFC0792] Postel, J., "Internet Control Message Protocol", STD 5, 1439 RFC 792, DOI 10.17487/RFC0792, September 1981, 1440 . 1442 [RFC1112] Deering, S., "Host extensions for IP multicasting", STD 5, 1443 RFC 1112, DOI 10.17487/RFC1112, August 1989, 1444 . 1446 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 1447 Requirement Levels", BCP 14, RFC 2119, 1448 DOI 10.17487/RFC2119, March 1997, 1449 . 1451 [RFC4443] Conta, A., Deering, S., and M. Gupta, Ed., "Internet 1452 Control Message Protocol (ICMPv6) for the Internet 1453 Protocol Version 6 (IPv6) Specification", STD 89, 1454 RFC 4443, DOI 10.17487/RFC4443, March 2006, 1455 . 1457 [RFC6935] Eubanks, M., Chimento, P., and M. Westerlund, "IPv6 and 1458 UDP Checksums for Tunneled Packets", RFC 6935, 1459 DOI 10.17487/RFC6935, April 2013, 1460 . 1462 [RFC6936] Fairhurst, G. and M. Westerlund, "Applicability Statement 1463 for the Use of IPv6 UDP Datagrams with Zero Checksums", 1464 RFC 6936, DOI 10.17487/RFC6936, April 2013, 1465 . 1467 [RFC8126] Cotton, M., Leiba, B., and T. Narten, "Guidelines for 1468 Writing an IANA Considerations Section in RFCs", BCP 26, 1469 RFC 8126, DOI 10.17487/RFC8126, June 2017, 1470 . 1472 [RFC8174] Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC 1473 2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174, 1474 May 2017, . 1476 10.2. Informative References 1478 [ETYPES] The IEEE Registration Authority, "IEEE 802 Numbers", 2013, 1479 . 1482 [I-D.ietf-intarea-tunnels] 1483 Touch, J. and M. Townsley, "IP Tunnels in the Internet 1484 Architecture", draft-ietf-intarea-tunnels-09 (work in 1485 progress), July 2018. 1487 [I-D.ietf-nvo3-dataplane-requirements] 1488 Bitar, N., Lasserre, M., Balus, F., Morin, T., Jin, L., 1489 and B. Khasnabish, "NVO3 Data Plane Requirements", draft- 1490 ietf-nvo3-dataplane-requirements-03 (work in progress), 1491 April 2014. 1493 [I-D.ietf-nvo3-encap] 1494 Boutros, S., "NVO3 Encapsulation Considerations", draft- 1495 ietf-nvo3-encap-02 (work in progress), September 2018. 1497 [IEEE.802.1Q_2014] 1498 IEEE, "IEEE Standard for Local and metropolitan area 1499 networks--Bridges and Bridged Networks", IEEE 802.1Q-2014, 1500 DOI 10.1109/ieeestd.2014.6991462, December 2014, 1501 . 1504 [RFC1191] Mogul, J. and S. Deering, "Path MTU discovery", RFC 1191, 1505 DOI 10.17487/RFC1191, November 1990, 1506 . 1508 [RFC2003] Perkins, C., "IP Encapsulation within IP", RFC 2003, 1509 DOI 10.17487/RFC2003, October 1996, 1510 . 1512 [RFC2460] Deering, S. and R. Hinden, "Internet Protocol, Version 6 1513 (IPv6) Specification", RFC 2460, DOI 10.17487/RFC2460, 1514 December 1998, . 1516 [RFC2983] Black, D., "Differentiated Services and Tunnels", 1517 RFC 2983, DOI 10.17487/RFC2983, October 2000, 1518 . 1520 [RFC3031] Rosen, E., Viswanathan, A., and R. Callon, "Multiprotocol 1521 Label Switching Architecture", RFC 3031, 1522 DOI 10.17487/RFC3031, January 2001, 1523 . 1525 [RFC3552] Rescorla, E. and B. Korver, "Guidelines for Writing RFC 1526 Text on Security Considerations", BCP 72, RFC 3552, 1527 DOI 10.17487/RFC3552, July 2003, 1528 . 1530 [RFC3985] Bryant, S., Ed. and P. Pate, Ed., "Pseudo Wire Emulation 1531 Edge-to-Edge (PWE3) Architecture", RFC 3985, 1532 DOI 10.17487/RFC3985, March 2005, 1533 . 1535 [RFC4301] Kent, S. and K. Seo, "Security Architecture for the 1536 Internet Protocol", RFC 4301, DOI 10.17487/RFC4301, 1537 December 2005, . 1539 [RFC5374] Weis, B., Gross, G., and D. Ignjatic, "Multicast 1540 Extensions to the Security Architecture for the Internet 1541 Protocol", RFC 5374, DOI 10.17487/RFC5374, November 2008, 1542 . 1544 [RFC6040] Briscoe, B., "Tunnelling of Explicit Congestion 1545 Notification", RFC 6040, DOI 10.17487/RFC6040, November 1546 2010, . 1548 [RFC7348] Mahalingam, M., Dutt, D., Duda, K., Agarwal, P., Kreeger, 1549 L., Sridhar, T., Bursell, M., and C. Wright, "Virtual 1550 eXtensible Local Area Network (VXLAN): A Framework for 1551 Overlaying Virtualized Layer 2 Networks over Layer 3 1552 Networks", RFC 7348, DOI 10.17487/RFC7348, August 2014, 1553 . 1555 [RFC7365] Lasserre, M., Balus, F., Morin, T., Bitar, N., and Y. 1556 Rekhter, "Framework for Data Center (DC) Network 1557 Virtualization", RFC 7365, DOI 10.17487/RFC7365, October 1558 2014, . 1560 [RFC7637] Garg, P., Ed. and Y. Wang, Ed., "NVGRE: Network 1561 Virtualization Using Generic Routing Encapsulation", 1562 RFC 7637, DOI 10.17487/RFC7637, September 2015, 1563 . 1565 [RFC8014] Black, D., Hudson, J., Kreeger, L., Lasserre, M., and T. 1566 Narten, "An Architecture for Data-Center Network 1567 Virtualization over Layer 3 (NVO3)", RFC 8014, 1568 DOI 10.17487/RFC8014, December 2016, 1569 . 1571 [RFC8085] Eggert, L., Fairhurst, G., and G. Shepherd, "UDP Usage 1572 Guidelines", BCP 145, RFC 8085, DOI 10.17487/RFC8085, 1573 March 2017, . 1575 [RFC8086] Yong, L., Ed., Crabbe, E., Xu, X., and T. Herbert, "GRE- 1576 in-UDP Encapsulation", RFC 8086, DOI 10.17487/RFC8086, 1577 March 2017, . 1579 [RFC8201] McCann, J., Deering, S., Mogul, J., and R. Hinden, Ed., 1580 "Path MTU Discovery for IP version 6", STD 87, RFC 8201, 1581 DOI 10.17487/RFC8201, July 2017, 1582 . 1584 [RFC8293] Ghanwani, A., Dunbar, L., McBride, M., Bannai, V., and R. 1585 Krishnan, "A Framework for Multicast in Network 1586 Virtualization over Layer 3", RFC 8293, 1587 DOI 10.17487/RFC8293, January 2018, 1588 . 1590 [VL2] Greenberg, A., et al., "VL2: A Scalable and Flexible Data 1591 Center Network", ACM SIGCOMM Computer Communication 1592 Review, DOI 10.1145/1594977.1592576, 2009, 1593 . 1596 Authors' Addresses 1598 Jesse Gross (editor) 1600 Email: jesse@kernel.org 1602 Ilango Ganga (editor) 1603 Intel Corporation 1604 2200 Mission College Blvd. 1605 Santa Clara, CA 95054 1606 USA 1608 Email: ilango.s.ganga@intel.com 1609 T. Sridhar (editor) 1610 VMware, Inc. 1611 3401 Hillview Ave. 1612 Palo Alto, CA 94304 1613 USA 1615 Email: tsridhar@vmware.com