idnits 2.17.1 draft-templin-seal-16.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- ** It looks like you're using RFC 3978 boilerplate. You should update this to the boilerplate described in the IETF Trust License Policy document (see https://trustee.ietf.org/license-info), which is required now. -- Found old boilerplate from RFC 3978, Section 5.1 on line 15. -- Found old boilerplate from RFC 3978, Section 5.5, updated by RFC 4748 on line 1027. -- Found old boilerplate from RFC 3979, Section 5, paragraph 1 on line 1038. -- Found old boilerplate from RFC 3979, Section 5, paragraph 2 on line 1045. -- Found old boilerplate from RFC 3979, Section 5, paragraph 3 on line 1051. 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 : ---------------------------------------------------------------------------- == There are 1 instance of lines with non-RFC2606-compliant FQDNs in the document. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust Copyright Line does not match the current year == The document seems to lack the recommended RFC 2119 boilerplate, even if it appears to use RFC 2119 keywords. (The document does seem to have the reference to RFC 2119 which the ID-Checklist requires). -- The document seems to lack a disclaimer for pre-RFC5378 work, but may have content which was first submitted before 10 November 2008. If you have contacted all the original authors and they are all willing to grant the BCP78 rights to the IETF Trust, then this is fine, and you can ignore this comment. If not, you may need to add the pre-RFC5378 disclaimer. (See the Legal Provisions document at https://trustee.ietf.org/license-info for more information.) -- The document date (June 1, 2008) is 5808 days in the past. Is this intentional? -- Found something which looks like a code comment -- if you have code sections in the document, please surround them with '' and '' lines. Checking references for intended status: Informational ---------------------------------------------------------------------------- ** Obsolete normative reference: RFC 2460 (Obsoleted by RFC 8200) -- Obsolete informational reference (is this intentional?): RFC 1063 (Obsoleted by RFC 1191) -- Obsolete informational reference (is this intentional?): RFC 1981 (Obsoleted by RFC 8201) Summary: 2 errors (**), 0 flaws (~~), 3 warnings (==), 10 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network Working Group F. Templin, Ed. 3 Internet-Draft Boeing Phantom Works 4 Intended status: Informational June 1, 2008 5 Expires: December 3, 2008 7 The Subnetwork Encapsulation and Adaptation Layer (SEAL) 8 draft-templin-seal-16.txt 10 Status of this Memo 12 By submitting this Internet-Draft, each author represents that any 13 applicable patent or other IPR claims of which he or she is aware 14 have been or will be disclosed, and any of which he or she becomes 15 aware will be disclosed, in accordance with Section 6 of BCP 79. 17 Internet-Drafts are working documents of the Internet Engineering 18 Task Force (IETF), its areas, and its working groups. Note that 19 other groups may also distribute working documents as Internet- 20 Drafts. 22 Internet-Drafts are draft documents valid for a maximum of six months 23 and may be updated, replaced, or obsoleted by other documents at any 24 time. It is inappropriate to use Internet-Drafts as reference 25 material or to cite them other than as "work in progress." 27 The list of current Internet-Drafts can be accessed at 28 http://www.ietf.org/ietf/1id-abstracts.txt. 30 The list of Internet-Draft Shadow Directories can be accessed at 31 http://www.ietf.org/shadow.html. 33 This Internet-Draft will expire on December 3, 2008. 35 Abstract 37 Subnetworks are connected network regions bounded by border nodes 38 that forward unicast and multicast packets over a virtual topology, 39 often manifested by encapsulation and/or tunneling. This virtual 40 topology may span multiple IP- and/or sub-IP layer forwarding hops, 41 and can introduce failure modes due to packet duplication and/or 42 links with diverse Maximum Transmission Units (MTUs). This document 43 specifies a Subnetwork Encapsulation and Adaptation Layer (SEAL) that 44 accommodates such virtual topologies over diverse underlying link 45 technologies. 47 Table of Contents 49 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3 50 2. Terminology and Requirements . . . . . . . . . . . . . . . . . 4 51 3. Applicability Statement . . . . . . . . . . . . . . . . . . . 5 52 4. SEAL Protocol Specification - Tunnel Mode . . . . . . . . . . 6 53 4.1. Model of Operation . . . . . . . . . . . . . . . . . . . . 6 54 4.2. ITE Specification . . . . . . . . . . . . . . . . . . . . 7 55 4.2.1. Tunnel Interface MTU . . . . . . . . . . . . . . . . . 7 56 4.2.2. Accounting for Headers . . . . . . . . . . . . . . . . 9 57 4.2.3. Segmentation and Encapsulation . . . . . . . . . . . . 9 58 4.2.4. Sending Probes . . . . . . . . . . . . . . . . . . . . 12 59 4.2.5. Packet Identification . . . . . . . . . . . . . . . . 12 60 4.2.6. Sending SEAL Protocol Packets . . . . . . . . . . . . 13 61 4.2.7. Processing Raw ICMPv4 Messages . . . . . . . . . . . . 13 62 4.2.8. Processing SEAL-Encapsulated ICMPv4 Messages . . . . . 13 63 4.3. ETE Specification . . . . . . . . . . . . . . . . . . . . 14 64 4.3.1. Reassembly Buffer Requirements . . . . . . . . . . . . 14 65 4.3.2. IPv4-Layer Reassembly . . . . . . . . . . . . . . . . 14 66 4.3.3. Generating SEAL-Encapsulated ICMPv4 Fragmentation 67 Needed Messages . . . . . . . . . . . . . . . . . . . 15 68 4.3.4. SEAL-Layer Reassembly . . . . . . . . . . . . . . . . 16 69 4.3.5. Decapsulation and Generating Other ICMPv4 Errors . . . 17 70 5. SEAL Protocol Specification - Transport Mode . . . . . . . . . 17 71 6. Link Requirements . . . . . . . . . . . . . . . . . . . . . . 18 72 7. End System Requirements . . . . . . . . . . . . . . . . . . . 18 73 8. Router Requirements . . . . . . . . . . . . . . . . . . . . . 18 74 9. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 18 75 10. Security Considerations . . . . . . . . . . . . . . . . . . . 18 76 11. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 19 77 12. References . . . . . . . . . . . . . . . . . . . . . . . . . . 19 78 12.1. Normative References . . . . . . . . . . . . . . . . . . . 19 79 12.2. Informative References . . . . . . . . . . . . . . . . . . 20 80 Appendix A. Historic Evolution of PMTUD (written 10/30/2002) . . 21 81 Author's Address . . . . . . . . . . . . . . . . . . . . . . . . . 22 82 Intellectual Property and Copyright Statements . . . . . . . . . . 24 84 1. Introduction 86 As Internet technology and communication has grown and matured, many 87 techniques have developed that use virtual topologies (frequently 88 tunnels of one form or another) over an actual IP network. Those 89 virtual topologies have elements which appear as one hop in the 90 virtual topology, but are actually multiple IP or sub-IP layer hops. 91 These multiple hops often have quite diverse properties which are 92 often not even visible to the end-points of the virtual hop. This 93 introduces many failure modes that are not dealt with well in current 94 approaches. 96 The use of IP encapsulation has long been considered as an 97 alternative for creating such virtual topologies. However, the 98 insertion of an outer IP header reduces the effective path MTU as- 99 seen by the IP layer. When IPv4 is used, this reduced MTU can be 100 accommodated through the use of IPv4 fragmentation, but unmitigated 101 in-the-network fragmentation has been shown to be harmful through 102 operational experience and studies conducted over the course of many 103 years [FRAG][FOLK][RFC4963]. Additionally, classical path MTU 104 discovery [RFC1191] has known operational issues that are exacerbated 105 by in-the-network tunnels [RFC2923][RFC4459]. 107 For the purpose of this document, subnetworks are defined as virtual 108 topologies that span connected network regions bounded by 109 encapsulating border nodes. Examples include the global Internet 110 interdomain routing core, Mobile Ad hoc Networks (MANETs) and 111 enterprise networks. Subnetwork border nodes support the Internet 112 protocols [RFC0791][RFC2460] and forward unicast and multicast IP 113 packets over the virtual topology across multiple IP- and/or sub-IP 114 layer forwarding hops which may introduce packet duplication and/or 115 traverse links with diverse Maximum Transmission Units (MTUs). 117 This document proposes a Subnetwork Encapsulation and Adaptation 118 Layer (SEAL) for tunnel-mode operation of IP over subnetworks that 119 connect the Ingress- and Egress Tunnel Endpoints (ITEs/ETEs) of 120 border nodes. Operation in transport mode is also supported when 121 subnetwork border node upper-layer protocols negotiate the use of 122 SEAL during connection establishment. SEAL accommodates links with 123 diverse MTUs and supports efficient duplicate packet detection by 124 introducing a minimal mid-layer encapsulation. The SEAL 125 encapsulation introduces an extended Identification field for packet 126 identification and a mid-layer segmentation and reassembly capability 127 that allows simplified cutting and pasting of packets without 128 invoking in-the-network IPv4 fragmentation. The SEAL encapsulation 129 layer and protocol is specified in the following sections. 131 2. Terminology and Requirements 133 The term "subnetwork" in this document refers to a virtual topology 134 that is configured over a connected network region bounded by border 135 nodes. 137 The terms "inner", "mid-layer" and "outer" respectively refer to the 138 innermost IP {layer, protocol, header, packet, etc.} before any 139 encapsulation, the mid-layer IP {protocol, header, packet, etc.) 140 after any mid-layer '*' encapsulation and the outermost IP {layer, 141 protocol, header, packet etc.} after SEAL/*/IPv4 encapsulation. 143 The notation IPvX/*/SEAL/*IPvY refers to an inner IPvX packet 144 encapsulated in any mid-layer '*' encapsulations followed by the SEAL 145 header followed by any outer '*' encapsulations followed by an outer 146 IPvY header. The notation "IP" means either IP protocol version 147 (IPv4 or IPv6). 149 The following abbreviations correspond to terms used within this 150 document and elsewhere in common Internetworking nomenclature: 152 Subnetwork - a connected network region bounded by border nodes 154 SEAL - Subnetwork Encapsulation and Adaptation Layer 156 ITE - Ingress Tunnel Endpoint 158 ETE - Egress Tunnel Endpoint 160 MTU - Maximum Transmission Unit 162 MHLEN - the length of any mid-layer '*' headers and trailers 164 OHLEN - the length of the outer encapsulating SEAL/*/IPv4 headers 166 S_MTU - the per-ETE SEAL Maximum Transmission Unit 168 S_MRU- the per-ETE SEAL Maximum Reassembly Unit 170 S_MSS - the SEAL Maximum Segment Size, derived from S_MTU 172 PTB - an ICMPv6 "Packet Too Big" or an ICMPv4 "Fragmentation 173 Needed" message 175 DF - the IPv4 header "Don't Fragment" flag 177 SEAL-ID - a 32-bit Identification value; randomly initialized and 178 monotonically incremented for each SEAL protocol packet 180 SEAL_PROTO - an IPv4 protocol number used for SEAL 182 SEAL_PORT - a TCP/UDP service port number used for SEAL 184 SEAL_OPTION - a TCP option number used for (transport-mode) SEAL 186 The keywords MUST, MUST NOT, REQUIRED, SHALL, SHALL NOT, SHOULD, 187 SHOULD NOT, RECOMMENDED, MAY, and OPTIONAL, when they appear in this 188 document, are to be interpreted as described in [RFC2119]. 190 3. Applicability Statement 192 SEAL was motivated by the specific use case of subnetwork abstraction 193 for Mobile Ad-hoc Networks (MANETs), however the domain of 194 applicability also extends to subnetwork abstractions of enterprise 195 networks, the interdomain routing core, etc. The domain of 196 application therefore also includes the map-and-encaps architecture 197 proposals in the IRTF Routing Research Group (RRG) (see: http:// 198 www3.tools.ietf.org/group/irtf/trac/wiki/RoutingResearchGroup). 200 SEAL introduces a minimal new sublayer for IPvX in IPvY encapsulation 201 (e.g., as IPv6/SEAL/IPv4), and appears as a subnetwork encapsulation 202 as seen by the inner IP layer. SEAL can also be used as a sublayer 203 for encapsulating inner IP packets within outer UDP/IPv4 header 204 (e.g., as IP/SEAL/UDP/IPv4) such as for the Teredo domain of 205 applicability [RFC4380]. When it appears immediately after the outer 206 IPv4 header, the SEAL header is processed exactly as for IPv6 207 extension headers. 209 SEAL can also be used in "transport-mode", e.g., when the inner layer 210 includes upper layer protocol data rather than an encapsulated IP 211 packet. For instance, TCP peers can negotiate the use of SEAL for 212 the carriage of protocol data encapsulated as TCP/SEAL/IPv4. In this 213 sense, the "subnetwork" becomes the entire end-to-end path between 214 the TCP peers and may potentially span the entire Internet. 216 The current document version is specific to the use of IPv4 as the 217 outer encapsulation layer, however the same principles apply when 218 IPv6 is used as the outer layer. 220 4. SEAL Protocol Specification - Tunnel Mode 222 4.1. Model of Operation 224 SEAL supports the encapsulation of inner IP packets in mid-layer and 225 outer encapsulating headers/trailers. For example, an inner IP 226 packet would appear as IP/*/SEAL/*/IPv4 after mid-layer and outer 227 encapsulations, where '*' denotes zero or more additional 228 encapsulation sublayers. Ingres Tunnel Endpoints (ITEs) add mid- 229 layer '*' and outer SEAL/*/IPv4 encapsulations to the inner packets 230 they inject into a subnetwork, where the outermost IPv4 header 231 contains the source and destination addresses of the subnetwork 232 entry/exit points (i.e., the ITE/ETE), respectively. SEAL defines a 233 new IP protocol type and a new encapsulation sublayer for both 234 unicast and multicast. The ITE encapsulates an inner IP packet in 235 mid-layer and outer encapsulations as shown in Figure 1: 237 +-------------------------+ 238 | | 239 ~ Outer */IPv4 headers ~ 240 | | 241 I +-------------------------+ 242 n | SEAL Header | 243 n +-------------------------+ +-------------------------+ 244 e ~ Any mid-layer * headers ~ ~ Any mid-layer * headers ~ 245 r +-------------------------+ +-------------------------+ 246 | | | | 247 I --> ~ Inner IP ~ --> ~ Inner IP ~ 248 P --> ~ Packet ~ --> ~ Packet ~ 249 | | | | 250 P +-------------------------+ +-------------------------+ 251 a ~ Any mid-layer trailers ~ ~ Any mid-layer trailers ~ 252 c +-------------------------+ +-------------------------+ 253 k ~ Any outer trailers ~ 254 e +-------------------------+ 255 t 256 (After mid-layer encaps.) (After SEAL/*/IPv4 encaps.) 258 Figure 1: SEAL Encapsulation 260 where the SEAL header is inserted as follows: 262 o For simple IP/IPv4 encapsulations (e.g., 263 [RFC2003][RFC2004][RFC4213]), the SEAL header is inserted between 264 the inner IP and outer IPv4 headers as: IP/SEAL/IPv4. 266 o For tunnel-mode IPsec encapsulations over IPv4, [RFC4301], the 267 SEAL header is inserted between the {AH,ESP} header and outer IPv4 268 headers as: IP/*/{AH,ESP}/SEAL/IPv4. 270 o For IP encapsulations over transports such as UDP, the SEAL header 271 is inserted immediately after the outer transport layer header, 272 e.g., as IP/*/SEAL/UDP/IPv4. 274 SEAL-encapsulated packets include a 32-bit SEAL-ID formed from the 275 concatenation of the 16-bit ID Extension field in the SEAL header as 276 the most-significant bits, and with the 16-bit ID value in the outer 277 IPv4 header as the least-significant bits. (For tunnels that 278 traverse IPv4 Network Address Translators, the SEAL-ID is instead 279 maintained only within the 16-bit ID Extension field in the SEAL 280 header.) Routers within the subnetwork use the SEAL-ID for duplicate 281 packet detection, and ITEs/ETEs use the SEAL-ID for SEAL segmentation 282 and reassembly. 284 SEAL enables a multi-level segmentation and reassembly capability. 285 First, the ITE can use IPv4 fragmentation to fragment inner IPv4 286 packets with DF=0 before SEAL encapsulation to avoid lower-level 287 segmentation and reassembly. Secondly, the SEAL layer itself 288 provides a simple mid-layer cutting-and-pasting of mid-layer packets 289 to avoid IPv4 fragmentation on the outer packet. Finally, ordinary 290 IPv4 fragmentation is permitted on the outer packet after SEAL 291 encapsulation and used to detect and dampen any in-the-network 292 fragmentation as quickly as possible. 294 The following sections specifiy the SEAL-related operations of the 295 ITE and ETE, respectively: 297 4.2. ITE Specification 299 4.2.1. Tunnel Interface MTU 301 The ITE configures a tunnel virtual interface over one or more 302 underlying links that connect the border node to the subnetwork. The 303 tunnel interface must present a fixed MTU to the inner IP layer 304 (i.e., Layer 3) as the size for admission of inner IP packets into 305 the tunnel. Since the tunnel interface may support a potentially 306 large set of ETEs, however, care must be taken in setting a greatest- 307 common-denominator MTU for all ETEs while still upholding end system 308 expectations. 310 Due to the ubiquitous deployment of standard Ethernet and similar 311 networking gear, the nominal Internet cell size has become 1500 312 bytes; this is the de facto size that end systems have come to expect 313 will either be delivered by the network without loss due to an MTU 314 restriction on the path or a suitable PTB message returned. However, 315 the network may not always deliver the necessary PTBs, leading to 316 MTU-related black holes [RFC2923]. The ITE therefore requires a 317 means for conveying 1500 byte (or smaller) packets to the ETE without 318 loss due to MTU restrictions and without dependence on PTB messages 319 from within the subnetwork. 321 In common deployments, there may be many forwarding hops between the 322 original source and the ITE. Within those hops, there may be 323 additional encapsulations (IPSec, L2TP, etc.) such that a 1500 byte 324 packet sent by the original source might grow to a larger size by the 325 time it reaches the ITE for encapsulation as an inner IP packet. 326 Similarly, additional encapsulations on the path from the ITE to the 327 ETE could cause the encapsulated packet to become larger still and 328 trigger in-the-network fragmentation. In order to preserve the end 329 system expectations, the ITE therefore requires a means for conveying 330 these larger packets to the ETE even though there may be links within 331 the subnetwork that configure a smaller MTU. 333 The ITE should therefore set a tunnel virtual interface MTU of 1500 334 bytes plus extra room to accommodate any additional encapsulations 335 that may occur on the path from the original source (i.e., even if 336 the underlying links do not support an MTU of this size). The ITE 337 can set larger MTU values still (up to the maximum MTU size of the 338 underlying links), but should select a value that is not so large as 339 to cause excessive PTBs coming from within the tunnel interface (see: 340 Sections 4.2.2 and 4.2.6). The ITE can also set smaller MTU values, 341 however care must be taken not to set so small a value that original 342 sources would experience an MTU underflow. In particular, IPv6 343 sources must see a minimum path MTU of 1280 bytes, and IPv4 sources 344 should see a minimum path MTU of 576 bytes. 346 The inner IP layer consults the tunnel interface MTU when admitting a 347 packet into the interface. For inner IPv4 packets larger than the 348 tunnel interface MTU and with the IPv4 Don't Fragment (DF) bit set to 349 0, the inner IP layer uses IPv4 fragmentation to break the packet 350 into IPv4 fragments no larger than the tunnel interface MTU then 351 admits each fragment into the tunnel as an independent packet. For 352 all other inner packets (IPv4 or IPv6), the ITE admits the packet if 353 it is no larger than the tunnel interface MTU; otherwise, it drops 354 the packet and sends an PTB message with an MTU value of the tunnel 355 interface MTU to the source. 357 4.2.2. Accounting for Headers 359 As for any upper-layer protocol, ITEs use the MTU of the underlying 360 IPv4 interface, the length of any mid-layer '*' headers and trailers, 361 and the length of the outer SEAL/*/IPv4 headers to determine the 362 maximum-sized upper layer payload. For example, when the underlying 363 IPv4 interface advertises an MTU of 1500 bytes and the ITE inserts a 364 minimum-length (i.e., 20 byte) IPv4 header, the ITE sees an upper 365 layer payload of 1480 bytes. When the ITE inserts IPv4 header 366 options, the size is further reduced by as many as 40 additional 367 bytes (the maximum length for IPv4 options) such that as few as 1440 368 bytes may be available for the upper layer payload. When the ITE 369 inserts outer '*' encapsulations before the SEAL header and mid-layer 370 '* encapsulations after the SEAL header, the available MTU for the 371 upper layer payload is reduced further still. 373 The ITE must additionally account for the length of the SEAL header 374 itself as an extra encapsulation that further reduces the size 375 available for the upper layer payload. The length of the SEAL header 376 is not incorporated in the IPv4 header length, therefore the network 377 does not observe the SEAL header as an IPv4 option. In this way, the 378 SEAL header is inserted after the IPv4 options but before the upper 379 layer payload in exactly the same manner as for IPv6 extension 380 headers. 382 4.2.3. Segmentation and Encapsulation 384 The ITE maintains a SEAL Maximum Transmission Unit (S_MTU) value for 385 each ETE as soft state within the tunnel interface (e.g., in the IPv4 386 destination cache). The ITE initializes S_MTU to the MTU of the 387 underlying IPv4 interface, and decreases or increases S_MTU based on 388 any ICMPv4 Fragmentation Needed messages received (see: Section 389 4.2.6). The ITE additionally maintains a SEAL Maximum Reassembly 390 Unit (S_MRU) value for each ETE. The ITE initializes S_MRU to a 391 value no larger than 2KB (2048 bytes), and uses this value to 392 determine when to set the "Dont Reassemble" bit (see below). 394 The ITE performs segmentation and encapsulation on inner packets that 395 have been admitted into the tunnel interface. For each inner packet, 396 the ITE maintains the length of any mid-layer '*' encapsulation 397 headers and trailers (e.g., for '*' = AH, ESP, NULL, etc.) in a per- 398 packet variable 'MHLEN' and maintains the length of the outer SEAL/*/ 399 IPv4 encapsulation headers in a per-packet variable 'OHLEN'. The ITE 400 also sets a per-packet variable 'S_MSS" to (S_MTU-OHLEN). Next, for 401 inner IPv4 packets with the DF bit set to 0, if the length of the 402 inner packet plus MHLEN is larger than S_MRU the ITE uses IPv4 403 fragmentation to break the packet into IPv4 fragments no larger than 404 S_MRU - MHLEN. For unfragmentable inner packets (e.g., IPv6 packets, 405 IPv4 packets with DF=1, etc.), if the length of the inner packet plus 406 MHLEN is larger than MAX(S_MSS, S_MRU) the ITE drops the packet and 407 sends an ICMP PTB message with an MTU value of (MAX(S_MSS, S_MRU) - 408 MHLEN) back to the original source. 410 The ITE then encapsulates each inner packet/fragment in the MHLEN 411 bytes of mid-layer '*' headers and trailers. For each such resulting 412 mid-layer packet, if the packet is no larger than S_MRU but is larger 413 than S_MSS, the ITE breaks it into N segments (N <= 16) that are no 414 larger than S_MSS bytes each. Each segment except the final one MUST 415 be of equal length, while the final segment MUST be no larger than 416 the initial segment. The first byte of each segment MUST begin 417 immediately after the final byte of the previous segment, i.e., the 418 segments MUST NOT overlap. 420 Note that this SEAL segmentation is used only for mid-layer packets 421 that are no larger than S_MRU; packets that are larger than S_MRU 422 (and also no larger than S_MSS) are instead encapsulated as a single 423 segment. Note also that this SEAL segmentation ignores the fact that 424 the mid-layer packet may be unfragmentable. This segmentation 425 process is a mid-layer (not an IP layer) operation employed by the 426 ITE to adapt the mid-layer packet to the subnetwork path 427 characteristics, and the ETE will restore the packet to its original 428 form during decapsulation. Therefore, the fact that the packet may 429 have been segmented within the subnetwork is not observable after 430 decapsulation. 432 The ITE next encapsulates each segment in a SEAL header formatted as 433 follows: 435 0 1 2 3 436 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 1 437 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 438 | ID Extension |P|R|D|M|Segment| Next Header | 439 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 441 Figure 2: SEAL Header Format 443 where the header fields are defined as follows: 445 ID Extension (16) 446 a 16-bit extension of the ID field in the outer IPv4 header; 447 encodes the most-significant 16 bits of a 32 bit SEAL-ID value. 449 P (1) 450 the "Probe" bit. Set to 1 if the ITE wishes to receive an 451 explicit acknowledgement from the ETE. 453 R (1) 454 the "Report Fragmentation" bit. Set to 1 if the ITE wishes to 455 receive a report from the ETE if any IPv4 fragmentation occurs. 457 D (1) 458 the "Dont Reassemble" bit. Set to 1 if the reassembled SEAL 459 protocol packet is to be discarded by the ETE if any IPv4 460 reassemly is required. 462 M (1) 463 the "More Segments" bit. Set to 1 if this SEAL protocol packet 464 contains a non-final segment of a multi-segment mid-layer packet. 466 Segment (4) 467 a 4-bit Segment number. Encodes a segment number between 0 - 15. 469 Next Header (8) an 8-bit field that encodes an IP protocol number 470 the same as for the IPv4 protocol and IPv6 next header fields. 472 For single-segment mid-layer packets, the ITE encapsulates the 473 segment in a SEAL header with (M=0; Segment=0). For N-segment mid- 474 layer packets (N <= 16), the ITE encapsulates each segment in a SEAL 475 header with (M=1; Segment=0) for the first segment, (M=1; Segment=1) 476 for the second segment, etc., with the final segment setting (M=0; 477 Segment=N-1). For each encapsulated segment, the ITE sets D=0 in the 478 SEAL header if the ETE is to accept the packet even if it arrives as 479 multiple IPv4 fragments; for example, the ITE may set D=0 in the SEAL 480 header of each segment for all mid-layer packets no larger than 481 S_MRU. The ITE instead sets D=1 in the SEAL header if the ETE is to 482 discard the packet if it arrives as multiple IPv4 fragments; in 483 particular, the ITE should set D=1 in the SEAL header of each segment 484 for all mid-layer packets larger than S_MRU. 486 The ITE next sets the P and R bits in the SEAL header of each segment 487 depending on whether the packet is to be used as an explicit/implicit 488 probe as specified in Section 4.2.4, then writes the IP protocol 489 number corresponding to the mid-layer packet in the SEAL 'Next 490 Header' field. Next, the ITE encapsulates the segment in the 491 requisite */IPv4 outer headers according to the specific 492 encapsulation format (e.g., [RFC2003], [RFC4213], [RFC4380], etc.), 493 except that it writes 'SEAL_PROTO' in the protocol field of the outer 494 IPv4 header (when simple IPv4 encapsualtion is used) or writes 495 'SEAL_PORT' in the outer destination service port field (e.g., when 496 UDP/IPv4 encapsulation is used). The ITE finally sets packet 497 identification values as specified in Section 4.2.5 and sends the 498 packets as specified in Section 4.2.6. 500 4.2.4. Sending Probes 502 When S_MSS is larger than 128, the ITE sends ordinary data packets as 503 implicit probes to detect in-the-network IPv4 fragmentation and to 504 determine new values for S_MTU. The ITE sets R=1 in the SEAL header 505 and DF=0 in the outer IPv4 header of each segment of a SEAL-segmented 506 packet to be used as an implicit probe, and will receive ICMPv4 507 Fragmentation Needed messages from the ETE if any IPv4 fragmentation 508 occurs. When S_MSS is no larger than 128, the ITE instead sets R=0 509 in the SEAL header to avoid generating fragmentation reports for 510 unavoidable in-the-network fragmentation. 512 The ITE may additionally send explicit probes periodically to manage 513 a window of SEAL-IDs of outstanding probes as a means to validate any 514 ICMPv4 messages it receives. The ITE sets P=1 in the SEAL header of 515 each segment of a SEAL-segmented packet to be used as an explicit 516 probe, where the probe can be either an ordinary data packet or a 517 NULL packet created by setting the 'Next Header' field in the SEAL 518 header to a value of "No Next Header" (see: [RFC2460], Section 4.7. 520 The ITE should periodically probe to detect increases in S_MTU. The 521 ITE can 1) reset S_MTU to the MTU of the underlying IPv4 interface, 522 and/or 2) send probes that are larger than the current S_MTU using 523 either a NULL packet or an ordinary data packet that is padded at the 524 end by setting the outer IPv4 length field to a larger value than the 525 packet's true length. 527 4.2.5. Packet Identification 529 For the purpose of packet identification, the ITE maintains a 32-bit 530 SEAL-ID value as per-ETE soft state, e.g. in the IPv4 destination 531 cache. The ITE randomly-initializes SEAL-ID when the soft state is 532 created and monotonically increments it (modulo 2^32) for each 533 successive SEAL protocol packet it sends to the ETE. For each 534 packet, the ITE writes the least-significant 16 bits of the SEAL-ID 535 value in the ID field in the outer IPv4 header, and writes the most- 536 significant 16 bits in the ID Extension field in the SEAL header. 538 For packets that may traverse IPv4 Network Address Translators 539 (NATs), the ITE instead maintains SEAL-ID as a 16-bit value that it 540 randomly-initializes when the soft state is created and monotonically 541 increments (modulo 2^16) for each successive SEAL protocol packet. 542 For each packet, the ITE writes SEAL-ID in the ID extension field of 543 the SEAL header and writes a random 16-bit value in the ID field in 544 the outer IPv4 header. This requires that both the ITE and ETE 545 participate in this alternate scheme. 547 4.2.6. Sending SEAL Protocol Packets 549 Following SEAL segmentation and encapsulation, the ITE sets DF=0 in 550 the outer IPv4 header of every outer packet it sends. For 551 "expendable" packets (e.g., for NULL packets used as probes - see: 552 Section 4.2.6), the ITE may optionally set DF=1. 554 The ITE then sends each outer packet that encapsulates a segment of 555 the same mid-layer packet into the tunnel in canonical order, i.e., 556 Segment 0 first, followed by Segment 1, etc. and finally Segment N-1. 558 4.2.7. Processing Raw ICMPv4 Messages 560 The ITE may receive "raw" ICMPv4 error messages from routers within 561 the subnetwork that comprise an outer IPv4 header followed by an 562 ICMPv4 header followed by a portion of the SEAL packet that generated 563 the error (also known as the "packet-in-error"). For such messages, 564 the ITE can use the 32-bit SEAL ID encoded in the packet-in-error as 565 a nonce to confirm that the ICMP message came from an on-path router 566 within the subnetwork. The ITE MAY process raw ICMPv4 messages as 567 soft errors indicating that the path to the ETE may be failing, but 568 it discards any raw ICMPv4 Fragmentation Needed messages for which 569 the IPv4 header of the packet-in-error has the DF=0. 571 4.2.8. Processing SEAL-Encapsulated ICMPv4 Messages 573 In addition to any raw ICMPv4 messages, the ITE may receive SEAL- 574 encapsulated ICMPv4 messages from subnetwork border nodes that 575 comprise outer ICMPv4/*/SEAL/*/IPv4 headers followed by a portion of 576 the SEAL-encapsulated packet-in-error. The ITE can use the 32-bit 577 SEAL ID encoded in the packet-in-error as well as information in the 578 outer IPv4 and SEAL headers as nonces to confirm that the ICMP 579 message came from a legitimate ETE. The ITE then verifies that the 580 SEAL-ID encoded in the packet-in-error is within the current window 581 of transmitted SEAL-IDs for this ETE. If the SEAL-ID is outside of 582 the window, the ITE discards the message; otherwise, it advances the 583 window and processes the message. 585 The ITE processes SEAL-encapsulated ICMPv4 messages other than ICMPv4 586 Fragmentation Needed exactly as specified in [RFC0792]. For all 587 SEAL-encapsulated ICMPv4 Fragmentation Needed messages, if the MTU 588 value is larger than S_MTU the ITE sets S_MTU to this new value. For 589 SEAL-encapsulated packets-in-error that the first-fragment of an IPv4 590 fragmented packet, the ITE instead sets S_MTU to this new value if 591 the value is no smaller than 576 and sets S_MTU to MAX(S_MTU/2, 128) 592 if the value is smaller than 576. 594 Note that in the above 576 is assumed as the nominal minimum MTU for 595 common IPv4 links, and accounts for normal-case IPv4 first fragments. 596 When an ETE reports an MTU smaller than 576, the ITE performs a 597 "limited halving" of S_MTU that accounts for IPv4 links with 598 unusually small MTUs or cases in which the ETE otherwise receives an 599 undersized IPv4 first-fragment. This limited halving may require 600 multiple iterations of sending probes and receiving ICMPv4 601 Fragmentation Needed messages, but will soon converge to a stable 602 S_MTU value. 604 After deterimining a new value for S_MTU, if the IPv4 header of the 605 packet-in-error has M=1 and its SEAL header has D=1 the ITE MAY 606 transcribe the message into an ICMP PTB message to send back to the 607 original source. To do so, the ITE discards the SEAL/*/IPv4 headers 608 plus any mid-layer '*' headers/trailers then encapsulates the 609 remaining inner IP packet portion in a PTB message with the MTU field 610 set to MAX((S_MSS, S_MRU) - MHLEN). Note that this may not be 611 possible when the inner IP packet portion was encrypted (e.g. via 612 IPsec/ESP) and is otherwise not entirely necessary, since the ITE 613 will discard subsequent large packets and send back an ICMP PTB 614 *before* encapsulating them and sending to the ETE. Transcribing 615 ICMPv4 Fragmentation Needed messages into ICMP PTBs is therefore 616 offered only as an optional optimization. 618 4.3. ETE Specification 620 4.3.1. Reassembly Buffer Requirements 622 ETEs MUST be capable of using IPv4-layer reassembly to reassemble 623 SEAL protocol outer IPv4 packets of at least 2KB plus the size of the 624 maximum-length outer SEAL/*/IPv4 plus mid-layer '*' headers and 625 trailers. For example, for simple IP/SEAL/IPv4 encapsulation, the 626 ETE must be capable of reassembling an outer IPv4 packet of 2KB + 4 + 627 60 bytes 629 The ETE MUST also be capable of using SEAL-layer reassembly to 630 reassemble mid-layer packets of 2KB. 632 4.3.2. IPv4-Layer Reassembly 634 The ETE performs IPv4 reassembly as-normal, and should maintain a 635 conservative high- and low-water mark for the number of outstanding 636 reassemblies pending for each ITE. When the size of the reassembly 637 buffer exceeds this high-water mark, the ETE actively discards 638 incomplete reassemblies (e.g., using an Active Queue Management (AQM) 639 strategy) until the size falls below the low-water mark. The ETE 640 should also use a reduced IPv4 maximum segment lifetime value (e.g., 641 15 seconds), i.e., the time after which it will discard an incomplete 642 IPv4 reassembly for a SEAL protocol packet. 644 After reassembly, the ETE either accepts or discards the reassembled 645 packet based on the current status of the IPv4 reassembly cache 646 (congested vs uncongested). The SEAL-ID included in the IPv4 first- 647 fragment provides an additional level of reassembly assurance, since 648 it can record a distinct arrival timestamp useful for associating the 649 first-fragment with its corresponding non-initial fragments. The 650 choice of accepting/discarding a reassembly may also depend on the 651 strength of the upper-layer integrity check if known (e.g., IPSec/ESP 652 provides a strong upper-layer integrity check) and/or the corruption 653 tolerance of the data (e.g., multicast streaming audio/video may be 654 more corruption-tolerant than file transfer, etc.). In the limiting 655 case, the ETE may choose to discard all IPv4 reassemblies and process 656 only the IPv4 first-fragment for SEAL-encapsulated error generation 657 purposes (see the following sections). 659 4.3.3. Generating SEAL-Encapsulated ICMPv4 Fragmentation Needed 660 Messages 662 During IPv4-layer reassembly, the ETE determines whether the packet 663 belongs to the SEAL protocol by checking for SEAL_PROTO in the outer 664 IPv4 header (i.e., for simple IPv4 encapsulation) or for SEAL_PORT in 665 the outer */IPv4 header (e.g., for '*'=UDP). When the ETE processes 666 the IPv4 first-fragment (i.e, one with DF=1 and Offset =0 in the IPv4 667 header) of a SEAL protocol IPv4 packet with (R=1; Segment=0) in the 668 SEAL header, it sends a SEAL-encapsulated ICMPv4 Fragmentation Needed 669 message back to the ITE with the MTU value set to the length of the 670 IPv4 first-fragment. 672 When the ETE processes a SEAL protocol IPv4 packet with (P=1; 673 Segment=0) for which no IPv4 reassembly was required, it sends a 674 SEAL-encapsulated ICMPv4 Fragmentation Needed message back to the ITE 675 with the MTU value set to the length of the whole IPv4 packet. Note 676 that explicit probes with (P=1; Segment=0) are only acknowledged by 677 the ETE if they arrive as an unfragmented IPv4 packet. The ITE 678 should therefore set both R=1 and P=1 if it requires an 679 acknowledgement whether or not any IPv4 fragmentation occurred. 681 The ETE prepares the ICMPv4 Fragmentation Needed message by 682 encapsulating as much of the first fragment (or the whole IPv4 683 packet) as possible in outer */SEAL/*/IPv4 headers without the length 684 of the message exceeding 576 bytes as shown in Figure 3: 686 +-------------------------+ - 687 | | \ 688 ~ Outer */SEAL/*/IPv4 hdrs~ | 689 | | | 690 +-------------------------+ | 691 | ICMPv4 Header | | 692 |(Dest Unreach; Frag Need)| | 693 +-------------------------+ | 694 | | > Up to 576 bytes 695 ~ IP/*/SEAL/*/IPv4 ~ | 696 ~ hdrs of IPv4 packet ~ | 697 | | | 698 +-------------------------+ | 699 | | | 700 ~ Data of IPv4 packet ~ | 701 | | / 702 +-------------------------+ - 704 Figure 3: SEAL-encapsulated ICMPv4 Fragmentation Needed Message 706 The ETE next sets D=0, P=0, R=0, M=0 and Segment=0 in the outer SEAL 707 header, sets the SEAL-ID the same as for any SEAL packet, then sets 708 the SEAL Next Header field and the fields of the outer */IPv4 headers 709 the same as for ordinay SEAL encapsulation. The ETE then sets outer 710 IPv4 destination address to the source address of the first-fragment 711 and sets the outer IPv4 source address to the destination address of 712 the first-fragment. If the destination address in the first-fragment 713 was multicast, the ETE instead sets the outer IPv4 source address to 714 an address assigned to the underlying IPv4 interface. The ETE 715 finally sends the SEAL-encapsulated ICMPv4 message to the ITE the 716 same as specified in Section 4.2.5, except that the ETE may send the 717 messages subject to rate limiting since it is not entirely critical 718 that all fragmentation be reported to the ITE. 720 4.3.4. SEAL-Layer Reassembly 722 Following IPv4 reassembly of a SEAL protocol packet, the ETE adds the 723 SEAL packet to a SEAL-Layer pending-reassembly queue (if necessary). 724 If the packet arrived as multiple IPv4 fragments and with D=1 in the 725 SEAL header, the ETE marks the packet and/or pending reassembly queue 726 as "discard following reassembly". The ETE also marks the packet as 727 "discard following reassembly" if the (Next Header, P, R, D) fields 728 of the packet's SEAL header differ from their respective values in 729 other SEAL segments already in the queue, i.e., the (Next Header, P, 730 R, D)-tuple serves as a reassembly nonce. 732 The ETE performs SEAL-layer reassembly for multi-segment mid-layer 733 packets through simple in-order concatenation of the encapsulated 734 segments from N consecutive SEAL protocol packets from the same mid- 735 layer packet. SEAL-layer reassembly requires the ETE to maintain a 736 cache of recently received SEAL packet segments for a hold time that 737 would allow for reasonable inter-segment delays. The ETE uses a SEAL 738 maximum segment lifetime of 15 seconds for this purpose, i.e., the 739 time after which it will discard an incomplete reassembly. However, 740 the ETE should also actively discard any pending reassemblies that 741 clearly have no opportunity for completion, e.g., when a considerable 742 number of new SEAL packets have been received before a packet that 743 completes a pending reassembly has arrived. 745 The ETE reassembles the mid-layer packet segments in SEAL protocol 746 packets that contain Segment numbers 0 through N-1, with M=1/0 in 747 non-final/final segments, respectively, and with consecutive SEAL-ID 748 values. That is, for an N-segment mid-layer packet, reassembly 749 entails the concatenation of the SEAL-encapsulated segments with 750 (Segment 0, SEAL-ID i), followed by (Segment 1, SEAL-ID ((i + 1) mod 751 2^32)), etc. up to (Segment N-1, SEAL-ID ((i + N-1) mod 2^32)). (For 752 tunnels that may traverse IPv4 NATs, the ETE instead uses only a 16- 753 bit SEAL-ID value, and uses mod 2^16 arithmetic to associate the 754 segments of the same packet.) 756 4.3.5. Decapsulation and Generating Other ICMPv4 Errors 758 Following SEAL-layer reassembly, if the packet had the value "No Next 759 Header" in the SEAL header's Next Header field, or if the packet was 760 marked "discard following reassembly" and IPv4 fragmentat was 761 experienced, the ETE silently discards the reassembled mid-layer 762 packet. Otherwise, the ETE decapsulates the inner packet and 763 processes it as normal. If the ETE determines that the decapsulated 764 inner packet cannot be processed further, it drops the packet and 765 prepares an appropriate SEAL-encapsulated ICMPv4 error message and 766 sends it back to the ITE exactly as for ICMPv4 Fragmentation Needed 767 messages (See: Section 4.3.3). 769 5. SEAL Protocol Specification - Transport Mode 771 Section 4 specifies the operation of SEAL in "tunnel mode", i.e., 772 when there is both an inner and outer IP layer and with a SEAL 773 encapsulation layer between. However, the SEAL protocol can also be 774 used in a "transport mode" of operation in which the inner layer 775 corresponds to an upper layer protocol (e.g., UDP, TCP, etc.) instead 776 of an additional IP layer. 778 For example, two TCP endpoints connected to the same subnetwork 779 region can negotiate the use of transport-mode SEAL for a connection 780 by inserting a 'SEAL_OPTION' TCP option during the connection 781 establishment phase. If both TCPs agree on the use of SEAL, their 782 protocol messages will be carriaged as TCP/SEAL/IPv4 and the 783 connection will be serviced by the SEAL protocol using TCP (nstead of 784 an encapsulating tunnel endpoint) as the upper layer protocol. The 785 SEAL protocol for transport mode otherwise observes the same 786 specifications as for Section 4. 788 6. Link Requirements 790 Subnetwork designers are strongly encouraged to follow the 791 recommendations in [RFC3819] when configuring link MTUs, where all 792 IPv4 links SHOULD configure a minimum MTU of 576 bytes. Links that 793 cannot configure an MTU of at least 576 bytes (e.g., due to 794 performance characteristics) SHOULD implement transparent link-layer 795 segmentation and reassembly such that an MTU of at least 576 can 796 still be presented to the IP layer. 798 7. End System Requirements 800 SEAL provides robust mechanisms for returning PTB messages to the 801 original source, however end systems that send unfragmentable IP 802 packets larger than 1500 bytes are strongly encouraged to use 803 Packetization Layer Path MTU Discovery per [RFC4821]. 805 8. Router Requirements 807 IPv4 routers within the subnetwork are strongly encouraged to 808 implement IPv4 fragmentation such that the first fragment is the 809 largest and approximately the size of the underlying link MTU. 811 9. IANA Considerations 813 SEAL_PROTO, SEAL_PORT and SEAL_OPTION are taken from their respective 814 range of experimental values documented in [RFC3692][RFC4727]. These 815 values are for experimentation purposes only, and not to be used for 816 any kind of deployments (i.e., they are not to be shipped in any 817 products). This document therefore has no actions for IANA. 819 10. Security Considerations 821 Unlike IPv4 fragmentation, overlapping fragment attacks are not 822 possible due to the requirement that SEAL segments be non- 823 overlapping. 825 An amplification/reflection attack is possible when an attacker sends 826 IPv4 first-fragments with spoofed source addresses to an ETE, 827 resulting in a stream of ICMPv4 Fragmentation Needed messages 828 returned to a victim ITE. The encapsulated segment of the spoofed 829 IPv4 first-fragment provides mitigation for the ITE to detect and 830 discard spurious ICMPv4 Fragmentation Needed messages. 832 The SEAL header is sent in-the-clear (outside of any IPsec/ESP 833 encapsulations) the same as for the IPv4 header. As for IPv6 834 extension headers, the SEAL header is protected only by L2 integrity 835 checks and is not covered under any L3 integrity checks. 837 11. Acknowledgments 839 Path MTU determination through the report of fragmentation 840 experienced by the final destination was first proposed by Charles 841 Lynn of BBN on the TCP-IP mailing list in May 1987. An historical 842 analysis of the evolution of path MTU discovery appears in 843 http://www.tools.ietf.org/html/draft-templin-v6v4-ndisc-01 and is 844 reproduced in Appendix A of this document. 846 The following individuals are acknowledged for helpful comments and 847 suggestions: Jari Arkko, Fred Baker, Teco Boot, Iljitsch van Beijnum, 848 Brian Carpenter, Steve Casner, Ian Chakeres, Remi Denis-Courmont, 849 Aurnaud Ebalard, Gorry Fairhurst, Joel Halpern, John Heffner, Bob 850 Hinden, Christian Huitema, Joe Macker, Matt Mathis, Dan Romascanu, 851 Dave Thaler, Joe Touch, Magnus Westerlund, Robin Whittle, James 852 Woodyatt and members of the Boeing PhantomWorks DC&NT group. 854 12. References 856 12.1. Normative References 858 [RFC0791] Postel, J., "Internet Protocol", STD 5, RFC 791, 859 September 1981. 861 [RFC0792] Postel, J., "Internet Control Message Protocol", STD 5, 862 RFC 792, September 1981. 864 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 865 Requirement Levels", BCP 14, RFC 2119, March 1997. 867 [RFC2460] Deering, S. and R. Hinden, "Internet Protocol, Version 6 868 (IPv6) Specification", RFC 2460, December 1998. 870 12.2. Informative References 872 [FOLK] C, C., D, D., and k. k, "Beyond Folklore: Observations on 873 Fragmented Traffic", December 2002. 875 [FRAG] Kent, C. and J. Mogul, "Fragmentation Considered Harmful", 876 October 1987. 878 [MTUDWG] "IETF MTU Discovery Working Group mailing list, 879 gatekeeper.dec.com/pub/DEC/WRL/mogul/mtudwg-log, November 880 1989 - February 1995.". 882 [RFC1063] Mogul, J., Kent, C., Partridge, C., and K. McCloghrie, "IP 883 MTU discovery options", RFC 1063, July 1988. 885 [RFC1191] Mogul, J. and S. Deering, "Path MTU discovery", RFC 1191, 886 November 1990. 888 [RFC1981] McCann, J., Deering, S., and J. Mogul, "Path MTU Discovery 889 for IP version 6", RFC 1981, August 1996. 891 [RFC2003] Perkins, C., "IP Encapsulation within IP", RFC 2003, 892 October 1996. 894 [RFC2004] Perkins, C., "Minimal Encapsulation within IP", RFC 2004, 895 October 1996. 897 [RFC2923] Lahey, K., "TCP Problems with Path MTU Discovery", 898 RFC 2923, September 2000. 900 [RFC3692] Narten, T., "Assigning Experimental and Testing Numbers 901 Considered Useful", BCP 82, RFC 3692, January 2004. 903 [RFC3819] Karn, P., Bormann, C., Fairhurst, G., Grossman, D., 904 Ludwig, R., Mahdavi, J., Montenegro, G., Touch, J., and L. 905 Wood, "Advice for Internet Subnetwork Designers", BCP 89, 906 RFC 3819, July 2004. 908 [RFC4213] Nordmark, E. and R. Gilligan, "Basic Transition Mechanisms 909 for IPv6 Hosts and Routers", RFC 4213, October 2005. 911 [RFC4301] Kent, S. and K. Seo, "Security Architecture for the 912 Internet Protocol", RFC 4301, December 2005. 914 [RFC4380] Huitema, C., "Teredo: Tunneling IPv6 over UDP through 915 Network Address Translations (NATs)", RFC 4380, 916 February 2006. 918 [RFC4459] Savola, P., "MTU and Fragmentation Issues with In-the- 919 Network Tunneling", RFC 4459, April 2006. 921 [RFC4727] Fenner, B., "Experimental Values In IPv4, IPv6, ICMPv4, 922 ICMPv6, UDP, and TCP Headers", RFC 4727, November 2006. 924 [RFC4821] Mathis, M. and J. Heffner, "Packetization Layer Path MTU 925 Discovery", RFC 4821, March 2007. 927 [RFC4963] Heffner, J., Mathis, M., and B. Chandler, "IPv4 Reassembly 928 Errors at High Data Rates", RFC 4963, July 2007. 930 [TCP-IP] "TCP-IP mailing list archives, 931 http://www-mice.cs.ucl.ac.uk/multimedia/mist/tcpip, May 932 1987 - May 1990.". 934 Appendix A. Historic Evolution of PMTUD (written 10/30/2002) 936 The topic of Path MTU discovery (PMTUD) saw a flurry of discussion 937 and numerous proposals in the late 1980's through early 1990. The 938 initial problem was posed by Art Berggreen on May 22, 1987 in a 939 message to the TCP-IP discussion group [TCP-IP]. The discussion that 940 followed provided significant reference material for [FRAG]. An IETF 941 Path MTU Discovery Working Group [MTUDWG] was formed in late 1989 942 with charter to produce an RFC. Several variations on a very few 943 basic proposals were entertained, including: 945 1. Routers record the PMTUD estimate in ICMP-like path probe 946 messages (proposed in [FRAG] and later [RFC1063]) 948 2. The destination reports any fragmentation that occurs for packets 949 received with the "RF" (Report Fragmentation) bit set (Steve 950 Deering's 1989 adaptation of Charles Lynn's Nov. 1987 proposal) 952 3. A hybrid combination of 1) and Charles Lynn's Nov. 1987 proposal 953 (straw RFC draft by McCloughrie, Fox and Mogul on Jan 12, 1990) 955 4. Combination of the Lynn proposal with TCP (Fred Bohle, Jan 30, 956 1990) 958 5. Fragmentation avoidance by setting "IP_DF" flag on all packets 959 and retransmitting if ICMPv4 "fragmentation needed" messages 960 occur (Geof Cooper's 1987 proposal; later adapted into [RFC1191] 961 by Mogul and Deering). 963 Option 1) seemed attractive to the group at the time, since it was 964 believed that routers would migrate more quickly than hosts. Option 965 2) was a strong contender, but repeated attempts to secure an "RF" 966 bit in the IPv4 header from the IESG failed and the proponents became 967 discouraged. 3) was abandoned because it was perceived as too 968 complicated, and 4) never received any apparent serious 969 consideration. Proposal 5) was a late entry into the discussion from 970 Steve Deering on Feb. 24th, 1990. The discussion group soon 971 thereafter seemingly lost track of all other proposals and adopted 972 5), which eventually evolved into [RFC1191] and later [RFC1981]. 974 In retrospect, the "RF" bit postulated in 2) is not needed if a 975 "contract" is first established between the peers, as in proposal 4) 976 and a message to the MTUDWG mailing list from jrd@PTT.LCS.MIT.EDU on 977 Feb 19. 1990. These proposals saw little discussion or rebuttal, and 978 were dismissed based on the following the assertions: 980 o routers upgrade their software faster than hosts 982 o PCs could not reassemble fragmented packets 984 o Proteon and Wellfleet routers did not reproduce the "RF" bit 985 properly in fragmented packets 987 o Ethernet-FDDI bridges would need to perform fragmentation (i.e., 988 "translucent" not "transparent" bridging) 990 o the 16-bit IP_ID field could wrap around and disrupt reassembly at 991 high packet arrival rates 993 The first four assertions, although perhaps valid at the time, have 994 been overcome by historical events leaving only the final to 995 consider. But, [FOLK] has shown that IP_ID wraparound simply does 996 not occur within several orders of magnitude the reassembly timeout 997 window on high-bandwidth networks. 999 (Authors 2/11/08 note: this final point was based on a loose 1000 interpretation of [FOLK], and is more accurately addressed in 1001 [RFC4963].) 1003 Author's Address 1005 Fred L. Templin (editor) 1006 Boeing Phantom Works 1007 P.O. Box 3707 1008 Seattle, WA 98124 1009 USA 1011 Email: fltemplin@acm.org 1013 Full Copyright Statement 1015 Copyright (C) The IETF Trust (2008). 1017 This document is subject to the rights, licenses and restrictions 1018 contained in BCP 78, and except as set forth therein, the authors 1019 retain all their rights. 1021 This document and the information contained herein are provided on an 1022 "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS 1023 OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY, THE IETF TRUST AND 1024 THE INTERNET ENGINEERING TASK FORCE DISCLAIM ALL WARRANTIES, EXPRESS 1025 OR IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF 1026 THE INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED 1027 WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. 1029 Intellectual Property 1031 The IETF takes no position regarding the validity or scope of any 1032 Intellectual Property Rights or other rights that might be claimed to 1033 pertain to the implementation or use of the technology described in 1034 this document or the extent to which any license under such rights 1035 might or might not be available; nor does it represent that it has 1036 made any independent effort to identify any such rights. Information 1037 on the procedures with respect to rights in RFC documents can be 1038 found in BCP 78 and BCP 79. 1040 Copies of IPR disclosures made to the IETF Secretariat and any 1041 assurances of licenses to be made available, or the result of an 1042 attempt made to obtain a general license or permission for the use of 1043 such proprietary rights by implementers or users of this 1044 specification can be obtained from the IETF on-line IPR repository at 1045 http://www.ietf.org/ipr. 1047 The IETF invites any interested party to bring to its attention any 1048 copyrights, patents or patent applications, or other proprietary 1049 rights that may cover technology that may be required to implement 1050 this standard. Please address the information to the IETF at 1051 ietf-ipr@ietf.org.