idnits 2.17.00 (12 Aug 2021) /tmp/idnits23286/draft-ietf-6lo-fragment-recovery-01.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 (Using the creation date from RFC4944, updated by this document, for RFC5378 checks: 2005-07-13) -- 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 (January 22, 2019) is 1215 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) ** Downref: Normative reference to an Informational draft: draft-watteyne-6lo-minimal-fragment (ref. 'I-D.watteyne-6lo-minimal-fragment') == Outdated reference: draft-ietf-6tisch-architecture has been published as RFC 9030 Summary: 1 error (**), 0 flaws (~~), 2 warnings (==), 2 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 6lo P. Thubert, Ed. 3 Internet-Draft Cisco Systems 4 Updates: 4944 (if approved) January 22, 2019 5 Intended status: Standards Track 6 Expires: July 26, 2019 8 6LoWPAN Selective Fragment Recovery 9 draft-ietf-6lo-fragment-recovery-01 11 Abstract 13 This draft updates RFC 4944 with a simple protocol to recover 14 individual fragments across a route-over mesh network, with a minimal 15 flow control to protect the network against bloat. 17 Status of This Memo 19 This Internet-Draft is submitted in full conformance with the 20 provisions of BCP 78 and BCP 79. 22 Internet-Drafts are working documents of the Internet Engineering 23 Task Force (IETF). Note that other groups may also distribute 24 working documents as Internet-Drafts. The list of current Internet- 25 Drafts is at https://datatracker.ietf.org/drafts/current/. 27 Internet-Drafts are draft documents valid for a maximum of six months 28 and may be updated, replaced, or obsoleted by other documents at any 29 time. It is inappropriate to use Internet-Drafts as reference 30 material or to cite them other than as "work in progress." 32 This Internet-Draft will expire on July 26, 2019. 34 Copyright Notice 36 Copyright (c) 2019 IETF Trust and the persons identified as the 37 document authors. All rights reserved. 39 This document is subject to BCP 78 and the IETF Trust's Legal 40 Provisions Relating to IETF Documents 41 (https://trustee.ietf.org/license-info) in effect on the date of 42 publication of this document. Please review these documents 43 carefully, as they describe your rights and restrictions with respect 44 to this document. Code Components extracted from this document must 45 include Simplified BSD License text as described in Section 4.e of 46 the Trust Legal Provisions and are provided without warranty as 47 described in the Simplified BSD License. 49 Table of Contents 51 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2 52 2. Updating RFC 4944 . . . . . . . . . . . . . . . . . . . . . . 3 53 2.1. Updating draft-watteyne-6lo-minimal-fragment . . . . . . 4 54 2.2. Slack in the First Fragment . . . . . . . . . . . . . . . 4 55 2.3. Modifying the First Fragment . . . . . . . . . . . . . . 5 56 3. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 5 57 3.1. BCP 14 . . . . . . . . . . . . . . . . . . . . . . . . . 5 58 3.2. References . . . . . . . . . . . . . . . . . . . . . . . 5 59 3.3. 6LoWPAN Acronyms . . . . . . . . . . . . . . . . . . . . 5 60 3.4. Referenced Work . . . . . . . . . . . . . . . . . . . . . 6 61 3.5. New Terms . . . . . . . . . . . . . . . . . . . . . . . . 6 62 4. New Dispatch types and headers . . . . . . . . . . . . . . . 7 63 4.1. Recoverable Fragment Dispatch type and Header . . . . . . 8 64 4.2. RFRAG Acknowledgment Dispatch type and Header . . . . . . 9 65 5. Fragments Recovery . . . . . . . . . . . . . . . . . . . . . 11 66 6. Forwarding Fragments . . . . . . . . . . . . . . . . . . . . 13 67 6.1. Upon the first fragment . . . . . . . . . . . . . . . . . 13 68 6.2. Upon the next fragments . . . . . . . . . . . . . . . . . 13 69 6.3. Upon the RFRAG Acknowledgments . . . . . . . . . . . . . 14 70 7. Security Considerations . . . . . . . . . . . . . . . . . . . 14 71 8. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 15 72 9. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 15 73 10. References . . . . . . . . . . . . . . . . . . . . . . . . . 15 74 10.1. Normative References . . . . . . . . . . . . . . . . . . 15 75 10.2. Informative References . . . . . . . . . . . . . . . . . 16 76 Appendix A. Rationale . . . . . . . . . . . . . . . . . . . . . 18 77 Appendix B. Requirements . . . . . . . . . . . . . . . . . . . . 19 78 Appendix C. Considerations On Flow Control . . . . . . . . . . . 20 79 Author's Address . . . . . . . . . . . . . . . . . . . . . . . . 21 81 1. Introduction 83 In most Low Power and Lossy Network (LLN) applications, the bulk of 84 the traffic consists of small chunks of data (in the order few bytes 85 to a few tens of bytes) at a time. Given that an IEEE Std. 802.15.4 86 [IEEE.802.15.4] frame can carry 74 bytes or more in all cases, 87 fragmentation is usually not required. However, and though this 88 happens only occasionally, a number of mission critical applications 89 do require the capability to transfer larger chunks of data, for 90 instance to support a firmware upgrades of the LLN nodes or an 91 extraction of logs from LLN nodes. In the former case, the large 92 chunk of data is transferred to the LLN node, whereas in the latter, 93 the large chunk flows away from the LLN node. In both cases, the 94 size can be on the order of 10Kbytes or more and an end-to-end 95 reliable transport is required. 97 "Transmission of IPv6 Packets over IEEE 802.15.4 Networks" [RFC4944] 98 defines the original 6LoWPAN datagram fragmentation mechanism for 99 LLNs. One critical issue with this original design is that routing 100 an IPv6 [RFC8200] packet across a route-over mesh requires to 101 reassemble the full packet at each hop, which may cause latency along 102 a path and an overall buffer bloat in the network. The "6TiSCH 103 Architecture" [I-D.ietf-6tisch-architecture] recommends to use a hop- 104 by-hop fragment forwarding technique to alleviate those undesirable 105 effects. "LLN Minimal Fragment Forwarding" 106 [I-D.watteyne-6lo-minimal-fragment] proposes such a technique, in a 107 fashion that is compatible with [RFC4944] without the need to define 108 a new protocol. However, adding that capability alone to the local 109 implementation of the original 6LoWPAN fragmentation would not 110 address the bulk of the issues raised against it, and may create new 111 issues like remnant state in the network. 113 Another issue against [RFC4944] is that it does not define a 114 mechanism to first discover the loss of a fragment along a multi-hop 115 path (e.g. having exhausted the link-layer retries at some hop on the 116 way), and then to recover that loss. With RFC 4944, the forwarding 117 of a whole datagram fails when one fragment is not delivered properly 118 to the destination 6LoWPAN endpoint. End-to-end transport or 119 application-level mechanisms may require a full retransmission of the 120 datagram, wasting resources in an already constrained network. 122 In that situation, the source 6LoWPAN endpoint will not be aware that 123 a loss occurred and will continue sending all fragments for a 124 datagram that is already doomed. The original support is missing 125 signaling to abort a multi-fragment transmission at any time and from 126 either end, and, if the capability to forward fragments is 127 implemented, clean up the related state in the network. It is also 128 lacking flow control capabilities to avoid participating to a 129 congestion that may in turn cause the loss of a fragment and trigger 130 the retransmission of the full datagram. 132 This specification proposes a method to forward fragments across a 133 multi-hop route-over mesh, and to recover individual fragments 134 between LLN endpoints. The method is designed to limit congestion 135 loss in the network and addresses the requirements that are detailed 136 in Appendix B. 138 2. Updating RFC 4944 140 This specification updates the fragmentation mechanism that is 141 specified in "Transmission of IPv6 Packets over IEEE 802.15.4 142 Networks" [RFC4944] for use in route-over LLNs by providing a model 143 where fragments can be forwarded end-to-end across a 6LoWPAN LLN, and 144 where fragments that are lost on the way can be recovered 145 individually. A new format for fragment is introduces and new 146 dispatch types are defined in Section 4. 148 [RFC8138] allows to modifies the size of a packet en-route by 149 removing the consumed hops in a compressed Routing Header. It 150 results that the fragment_offset and datagram_size cannot be signaled 151 in the uncompressed form. This specification expresses those fields 152 in the compressed form and allows to modify them en-route (see 153 Section 2.3. 155 Note that consistantly with in Section 2 of [RFC6282] for the 156 fragmentation mechanism described in Section 5.3 of [RFC4944], any 157 header that cannot fit within the first fragment MUST NOT be 158 compressed when using the fragmentation mechanism described in this 159 specification. 161 2.1. Updating draft-watteyne-6lo-minimal-fragment 163 This specification updates the fragment forwarding mechanism 164 specified in "LLN Minimal Fragment Forwarding" 165 [I-D.watteyne-6lo-minimal-fragment] by providing additional 166 operations to improve the management of the Virtual Reassembly Buffer 167 (VRB). 169 2.2. Slack in the First Fragment 171 At the time of this writing, [I-D.watteyne-6lo-minimal-fragment] 172 allows for refragmenting in intermediate nodes, meaning that some 173 bytes from a given fragment may be left in the VRB to be added to the 174 next fragment. The reason for this to happen would be the need for 175 space in the outgoing fragment that was not needed in the incoming 176 fragment, for instance because the 6LoWPAN Header Compression is not 177 as efficient on the outgoing link, e.g., if the Interface ID (IID) of 178 the source IPv6 address is elided by the originator on the first hop 179 because it matches the source MAC address, but cannot be on the next 180 hops because the source MAC address changes. 182 This specification cannot allow this operation since fragments are 183 recovered end-to-end based on the fragment number. This means that 184 the fragments that contain a 6LoWPAN-compressed header MUST have 185 enough slack to enable a less efficient compression in the next hops 186 that still fits in one MAC frame. For instance, if the IID of the 187 source IPv6 address is elided by the originator, then it MUST compute 188 the fragment_size as if the MTU was 8 bytes less. This way, the next 189 hop can restore the source IID to the first fragment without 190 impacting the second fragment. 192 2.3. Modifying the First Fragment 194 The compression of the Hop Limit, of the source and destination 195 addresses, and of the Routing Header may change en route in a Route- 196 Over mesh network. If the size of the first fragment is modified, 197 then the intermediate node MUST adapt te datagram_size to reflect 198 that difference. 200 The intermediate node MUSt also save the difference of datagram_size 201 of the first fragment in the VRB, and add it to the datagram_size and 202 to the fragment_offset of all the subsequent fragments for that 203 datagram. 205 3. Terminology 207 3.1. BCP 14 209 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 210 "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and 211 "OPTIONAL" in this document are to be interpreted as described in BCP 212 14 [RFC2119][RFC8174] when, and only when, they appear in all 213 capitals, as shown here. 215 3.2. References 217 In this document, readers will encounter terms and concepts that are 218 discussed in the following documents: 220 o "Problem Statement and Requirements for IPv6 over Low-Power 221 Wireless Personal Area Network (6LoWPAN) Routing" [RFC6606] 223 3.3. 6LoWPAN Acronyms 225 This document uses the following acronyms: 227 6BBR: 6LoWPAN Backbone Router 229 6LBR: 6LoWPAN Border Router 231 6LN: 6LoWPAN Node 233 6LR: 6LoWPAN Router 235 LLN: Low-Power and Lossy Network 237 3.4. Referenced Work 239 Past experience with fragmentation has shown that miss-associated or 240 lost fragments can lead to poor network behavior and, occasionally, 241 trouble at application layer. The reader is encouraged to read "IPv4 242 Reassembly Errors at High Data Rates" [RFC4963] and follow the 243 references for more information. 245 That experience led to the definition of "Path MTU discovery" 246 [RFC8201] (PMTUD) protocol that limits fragmentation over the 247 Internet. 249 Specifically in the case of UDP, valuable additional information can 250 be found in "UDP Usage Guidelines for Application Designers" 251 [RFC8085]. 253 Readers are expected to be familiar with all the terms and concepts 254 that are discussed in "IPv6 over Low-Power Wireless Personal Area 255 Networks (6LoWPANs): Overview, Assumptions, Problem Statement, and 256 Goals" [RFC4919] and "Transmission of IPv6 Packets over IEEE 802.15.4 257 Networks" [RFC4944]. 259 "The Benefits of Using Explicit Congestion Notification (ECN)" 260 [RFC8087] provides useful information on the potential benefits and 261 pitfalls of using ECN. 263 Quoting the "Multiprotocol Label Switching (MPLS) Architecture" 264 [RFC3031]: with MPLS, "packets are "labeled" before they are 265 forwarded. At subsequent hops, there is no further analysis of the 266 packet's network layer header. Rather, the label is used as an index 267 into a table which specifies the next hop, and a new label". The 268 MPLS technique is leveraged in the present specification to forward 269 fragments that actually do not have a network layer header, since the 270 fragmentation occurs below IP. 272 "LLN Minimal Fragment Forwarding" [I-D.watteyne-6lo-minimal-fragment] 273 introduces the concept of a Virtual Reassembly Buffer (VRB) and an 274 associated technique to forward fragments as they come, using the 275 datagram_tag as a label in a fashion similar to MLPS. This 276 specification reuses that technique with slightly modified controls. 278 3.5. New Terms 280 This specification uses the following terms: 282 6LoWPAN endpoints 283 The LLN nodes in charge of generating or expanding a 6LoWPAN 284 header from/to a full IPv6 packet. The 6LoWPAN endpoints are the 285 points where fragmentation and reassembly take place. 287 4. New Dispatch types and headers 289 This specification enables the 6LoWPAN fragmentation sublayer to 290 provide an MTU up to 2048 bytes to the upper layer, which can be the 291 6LoWPAN Header Compression sublayer that is defined in the 292 "Compression Format for IPv6 Datagrams" [RFC6282] specification. In 293 order to achieve this, this specification enables the fragmentation 294 and the reliable transmission of fragments over a multihop 6LoWPAN 295 mesh network. 297 This specification provides a technique that is derived from MPLS in 298 order to forward individual fragments across a 6LoWPAN route-over 299 mesh. The datagram_tag is used as a label; it is locally unique to 300 the node that is the source MAC address of the fragment, so together 301 the MAC address and the label can identify the fragment globally. A 302 node may build the datagram_tag in its own locally-significant way, 303 as long as the selected tag stays unique to the particular datagram 304 for the lifetime of that datagram. It results that the label does 305 not need to be globally unique but also that it must be swapped at 306 each hop as the source MAC address changes. 308 This specification extends RFC 4944 [RFC4944] with 4 new Dispatch 309 types, for Recoverable Fragment (RFRAG) headers with or without 310 Acknowledgment Request (RFRAG vs. RFRAG-ARQ), and for the RFRAG 311 Acknowledgment back, with or without ECN Echo (RFRAG-ACK vs. RFRAG- 312 ECHO). 314 (to be confirmed by IANA) The new 6LoWPAN Dispatch types use the 315 Value Bit Pattern of 11 1010xx from page 0 [RFC8025], as follows: 317 Pattern Header Type 318 +------------+------------------------------------------+ 319 | 11 101000 | RFRAG - Recoverable Fragment | 320 | 11 101001 | RFRAG-ARQ - RFRAG with Ack Request | 321 | 11 101010 | RFRAG-ACK - RFRAG Acknowledgment | 322 | 11 101011 | RFRAG-ECHO - RFRAG Ack with ECN Echo | 323 +------------+------------------------------------------+ 325 Figure 1: Additional Dispatch Value Bit Patterns 327 In the following sections, the semantics of "datagram_tag" are 328 unchanged from [RFC4944] Section 5.3. "Fragmentation Type and 329 Header." and is compatible with the fragment forwarding operation 330 described in [I-D.watteyne-6lo-minimal-fragment]. 332 4.1. Recoverable Fragment Dispatch type and Header 334 In this specification, the size and offset of the fragments are 335 expressed on the compressed packet form as opposed to the 336 uncompressed - native - packet form. 338 The first fragment is recognized by a sequence of 0; it carries its 339 fragment_size and the datagram_size of the compressed packet, whereas 340 the other fragments carry their fragment_size and fragment_offset. 341 The last fragment for a datagram is recognized when its 342 fragment_offset and its fragment_size add up to the datagram_size. 344 Recoverable Fragments are sequenced and a bitmap is used in the RFRAG 345 Acknowledgment to indicate the received fragments by setting the 346 individual bits that correspond to their sequence. 348 1 2 3 349 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 1 350 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 351 |1 1 1 0 1 0 0|E| datagram_tag | 352 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 353 |X| sequence| fragment_size | fragment_offset | 354 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 356 X set == Ack Requested 358 Figure 2: RFRAG Dispatch type and Header 360 E: 1 bit; Explicit Congestion Notification; the "E" flag is reset by 361 the source of the fragment and set by intermediate routers to 362 signal that this fragment experienced congestion along its path. 364 Fragment_size: 10 bit unsigned integer; the size of this fragment in 365 a unit that depends on the MAC layer technology. For IEEE Std. 366 802.15.4, the unit is octet, and the maximum fragment size, which 367 is constrained by the maximum frame size of 128 octet minus the 368 overheads of the MAC and Fragment Headers, is not limited by this 369 encoding. 371 X: 1 bit; Ack Requested: when set, the sender requires an RFRAG 372 Acknowledgment from the receiver. 374 Sequence: 5 bit unsigned integer; the sequence number of the 375 fragment. Fragments are sequence numbered [0..N] where N is in 376 [0..31]. A sequence of 0 indicates the first fragment in a 377 datagram. For IEEE Std. 802.15.4, as long as the overheads enable 378 a fragment size of 64 octets or more, this enables to fragment a 379 packet of 2047 octets. 381 Fragment_offset: 16 bit unsigned integer; 383 * When set to a non-0 value, the semantics of the Fragment_offset 384 depends on the value of the Sequence. 386 + When the Sequence is not 0, this field indicates the offset 387 of the fragment in the compressed form. The fragment should 388 be forwarded based on an existing VRB as described in 389 Section 6.2, or silently dropped if none is found. 391 + For a first fragment (i.e. with a sequence of 0), this field 392 is overloaded to indicate the total_size of the compressed 393 packet, to help the receiver allocate an adapted buffer for 394 the reception and reassembly operations. This format limits 395 the maximum MTU on a 6LoWPAN link to 2047 bytes, but 1280 396 bytes is the recommended value to avoid issues with IPV6 397 Path MTU Discovery [RFC8201]. The fragment should be routed 398 based on the destination IPv6 address, and an VRB state 399 should be installed as described in Section 6.1. 401 * When set to 0, this field indicates an abort condition and all 402 state regarding the datagram should be cleaned up once the 403 processing of the fragment is complete; the processing of the 404 fragment depends on whether there is a VRB already established 405 for this datagram, and the next hop is still reachable: 407 + if a VRB already exists and is not broken, the fragment is 408 to be forwarded along the associated Label Switched Path 409 (LSP) as described in Section 6.2, but regardless of the 410 value of the Sequence field; 412 + else, if the Sequence is 0, then the fragment is to be 413 routed as described in Section 6.1 but no state is conserved 414 afterwards. 416 If the fragment cannot be forwarded or routed, then an abort 417 RFRAG-ACK is sent back to the source. 419 4.2. RFRAG Acknowledgment Dispatch type and Header 421 This specification also defines a 4-octet RFRAG Acknowledgment bitmap 422 that is used by the reassembling end point to confirm selectively the 423 reception of individual fragments. A given offset in the bitmap maps 424 one to one with a given sequence number. 426 The offset of the bit in the bitmap indicates which fragment is 427 acknowledged as follows: 429 1 2 3 430 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 1 431 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 432 | RFRAG Acknowledgment Bitmap | 433 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 434 ^ ^ 435 | | bitmap indicating whether: 436 | +--- Fragment with sequence 10 was received 437 +----------------------- Fragment with sequence 00 was received 439 Figure 3: RFRAG Acknowledgment bitmap encoding 441 Figure 4 shows an example Acknowledgment bitmap which indicates that 442 all fragments from sequence 0 to 20 were received, except for 443 fragments 1, 2 and 16 that were either lost or are still in the 444 network over a slower path. 446 1 2 3 447 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 1 448 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 449 |1|0|0|1|1|1|1|1|1|1|1|1|1|1|1|1|0|1|1|1|1|0|0|0|0|0|0|0|0|0|0|0| 450 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 452 Figure 4: Expanding 3 octets encoding 454 The RFRAG Acknowledgment Bitmap is included in a RFRAG Acknowledgment 455 header, as follows: 457 1 2 3 458 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 1 459 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 460 |1 1 1 0 1 0 1 Y| datagram_tag | 461 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 462 | RFRAG Acknowledgment Bitmap (32 bits) | 463 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 465 Figure 5: RFRAG Acknowledgment Dispatch type and Header 467 Y: 1 bit; Explicit Congestion Notification Echo 469 When set, the sender indicates that at least one of the 470 acknowledged fragments was received with an Explicit Congestion 471 Notification, indicating that the path followed by the fragments 472 is subject to congestion. 474 RFRAG Acknowledgment Bitmap 476 An RFRAG Acknowledgment Bitmap, whereby setting the bit at offset 477 x indicates that fragment x was received, as shown in Figure 3. 478 All 0's is a NULL bitmap that indicates that the fragmentation 479 process is aborted. All 1's is a FULL bitmap that indicates that 480 the fragmentation process is complete, all fragments were received 481 at the reassembly end point. 483 5. Fragments Recovery 485 The Recoverable Fragment headers RFRAG and RFRAG-ARQ are used to 486 transport a fragment and optionally request an RFRAG Acknowledgment 487 that will confirm the good reception of a one or more fragments. An 488 RFRAG Acknowledgment can optionally carry an ECN indication; it is 489 carried as a standalone header in a message that is sent back to the 490 6LoWPAN endpoint that was the source of the fragments, as known by 491 its MAC address. The process ensures that at every hop, the source 492 MAC address and the datagram_tag in the received fragment are enough 493 information to send the RFRAG Acknowledgment back towards the source 494 6LoWPAN endpoint by reversing the MPLS operation. 496 The 6LoWPAN endpoint that fragments the packets at 6LoWPAN level (the 497 sender) also controls when the reassembling end point sends the RFRAG 498 Acknowledgments by setting the Ack Requested flag in the RFRAG 499 packets. It may set the Ack Requested flag on any fragment to 500 perform congestion control by limiting the number of outstanding 501 fragments, which are the fragments that have been sent but for which 502 reception or loss was not positively confirmed by the reassembling 503 endpoint. When the sender of the fragment knows that an underlying 504 link-layer mechanism protects the Fragments, it may refrain from 505 using the RFRAG Acknowledgment mechanism, and never set the Ack 506 Requested bit. When it receives a fragment with the ACK Request flag 507 set, the 6LoWPAN endpoint that reassembles the packets at 6LoWPAN 508 level (the receiver) sends back an RFRAG Acknowledgment to confirm 509 reception of all the fragments it has received so far. 511 The sender transfers a controlled number of fragments and MAY flag 512 the last fragment of a series with an RFRAG Acknowledgment Request. 513 The received MUST acknowledge a fragment with the acknowledgment 514 request bit set. If any fragment immediately preceding an 515 acknowledgment request is still missing, the receiver MAY 516 intentionally delay its acknowledgment to allow in-transit fragments 517 to arrive. Delaying the acknowledgment might defeat the round trip 518 delay computation so it should be configurable and not enabled by 519 default. 521 The receiver MAY issue unsolicited acknowledgments. An unsolicited 522 acknowledgment signals to the sender endpoint that it can resume 523 sending if it had reached its maximum number of outstanding 524 fragments. Another use is to inform that the reassembling endpoint 525 has canceled the process of an individual datagram. Note that 526 acknowledgments might consume precious resources so the use of 527 unsolicited acknowledgments should be configurable and not enabled by 528 default. 530 An observation is that streamlining forwarding of fragments generally 531 reduces the latency over the LLN mesh, providing room for retries 532 within existing upper-layer reliability mechanisms. The sender 533 protects the transmission over the LLN mesh with a retry timer that 534 is computed according to the method detailed in [RFC6298]. It is 535 expected that the upper layer retries obey the recommendations in 536 "UDP Usage Guidelines" [RFC8085], in which case a single round of 537 fragment recovery should fit within the upper layer recovery timers. 539 Fragments are sent in a round robin fashion: the sender sends all the 540 fragments for a first time before it retries any lost fragment; lost 541 fragments are retried in sequence, oldest first. This mechanism 542 enables the receiver to acknowledge fragments that were delayed in 543 the network before they are actually retried. 545 When a single frequency is used by contiguous hops, the sender should 546 wait a reasonable amount of time between fragments so as to let a 547 fragment progress a few hops and avoid hidden terminal issues. This 548 precaution is not required on channel hopping technologies such as 549 Time Slotted CHannel Hopping (TSCH) [RFC6554] 551 When the sender decides that a packet should be dropped and the 552 fragmentation process canceled, it sends a pseudo fragment with the 553 fragment_offset, sequence and fragment_size all set to 0, and no 554 data. Upon reception of this message, the receiver should clean up 555 all resources for the packet associated to the datagram_tag. If an 556 acknowledgment is requested, the receiver responds with a NULL 557 bitmap. 559 The receiver might need to cancel the process of a fragmented packet 560 for internal reasons, for instance if it is out of reassembly 561 buffers, or considers that this packet is already fully reassembled 562 and passed to the upper layer. In that case, the receiver SHOULD 563 indicate so to the sender with a NULL bitmap in a RFRAG 564 Acknowledgment. Upon an acknowledgment with a NULL bitmap, the 565 sender endpoint MUST abort the transmission of the fragmented 566 datagram. 568 6. Forwarding Fragments 570 It is assumed that the first Fragment is large enough to carry the 571 IPv6 header and make routing decisions. If that is not so, then this 572 specification MUST NOT be used. 574 This specification extends the Virtual Reassembly Buffer (VRB) 575 technique to forward fragments with no intermediate reconstruction of 576 the entire packet. The first fragment carries the IP header and it 577 is routed all the way from the fragmenting end point to the 578 reassembling end point. Upon the first fragment, the routers along 579 the path install a label-switched path (LSP), and the following 580 fragments are label-switched along that path. As a consequence, 581 alternate routes not possible for individual fragments. The 582 datagram_tag is used to carry the label, that is swapped at each hop. 583 All fragments follow the same path and fragments are delivered in the 584 order at which they are sent. 586 6.1. Upon the first fragment 588 In Route-Over mode, the source and destination MAC addressed in a 589 frame change at each hop. The label that is formed and placed in the 590 datagram_tag is associated to the source MAC and only valid (and 591 unique) for that source MAC. Upon a first fragment (i.e. with a 592 sequence of zero), a VRB and the associated LSP state are created for 593 the tuple (source MAC address, datagram_tag) and the fragment is 594 forwarded along the IPv6 route that matches the destination IPv6 595 address in the IPv6 header as prescribed by 596 [I-D.watteyne-6lo-minimal-fragment]. The LSP state enables to match 597 the (previous MAC address, datagram_tag) in an incoming fragment to 598 the tuple (next MAC address, swapped datagram_tag) used in the 599 forwarded fragment and points at the VRB. In addition, the router 600 also forms a Reverse LSP state indexed by the MAC address of the next 601 hop and the swapped datagram_tag. This reverse LSP state also points 602 at the VRB and enables to match the (next MAC address, 603 swapped_datagram_tag) found in an RFRAG Acknowledgment to the tuple 604 (previous MAC address, datagram_tag) used when forwarding a Fragment 605 Acknowledgment (RFRAG-ACK) back to the sender endpoint. 607 6.2. Upon the next fragments 609 Upon a next fragment (i.e. with a non-zero sequence), the router 610 looks up a LSP indexed by the tuple (MAC address, datagram_tag) found 611 in the fragment. If it is found, the router forwards the fragment 612 using the associated VRB as prescribed by 613 [I-D.watteyne-6lo-minimal-fragment]. 615 if the VRB for the tuple is not found, the router builds an RFRAG-ACK 616 to abort the transmission of the packet. The resulting message has 617 the following information: 619 o The source and destination MAC addresses are swapped from those 620 found in the fragment 622 o The datagram_tag set to the datagram_tag found in the fragment 624 o A null bitmap is used to signal the abort condition 626 At this point the router is all set and can send the RFRAG-ACK back 627 to the previous router. The RFRAG-ACK should normally be forwarded 628 all the way to the source using the reverse LSP state in the VRBs in 629 the intermediate routers as described in the next section. 631 6.3. Upon the RFRAG Acknowledgments 633 Upon an RFRAG-ACK, the router looks up a Reverse LSP indexed by the 634 tuple (MAC address, datagram_tag), which are respectively the source 635 MAC address of the received frame and the received datagram_tag. If 636 it is found, the router forwards the fragment using the associated 637 VRB as prescribed by [I-D.watteyne-6lo-minimal-fragment], but using 638 the Reverse LSP so that the RFRAG-ACK flows back to the sender 639 endpoint. 641 If the Reverse LSP is not found, the router MUST silently drop the 642 RFRAG-ACK message. 644 Either way, if the RFRAG-ACK indicates either an error (NULL bitmap) 645 or that the fragment was entirely received (FULL bitmap), arms a 646 short timer, and upon timeout, the VRB and all associate state are 647 destroyed. During that time, fragments of that datagram may still be 648 received, e.g. if the RFRAG-ACK was lost on the way back and the 649 source retried the last fragment. In that case, the router sends an 650 abort RFRAG-ACK along the Reverse LSP to complete the clean up. 652 7. Security Considerations 654 The process of recovering fragments does not appear to create any 655 opening for new threat compared to "Transmission of IPv6 Packets over 656 IEEE 802.15.4 Networks" [RFC4944]. 658 8. IANA Considerations 660 Need extensions for formats defined in "Transmission of IPv6 Packets 661 over IEEE 802.15.4 Networks" [RFC4944]. 663 9. Acknowledgments 665 The author wishes to thank Thomas Watteyne and Michael Richardson for 666 in-depth reviews and comments. Also many thanks to Jonathan Hui, Jay 667 Werb, Christos Polyzois, Soumitri Kolavennu, Pat Kinney, Margaret 668 Wasserman, Richard Kelsey, Carsten Bormann and Harry Courtice for 669 their various contributions. 671 10. References 673 10.1. Normative References 675 [I-D.watteyne-6lo-minimal-fragment] 676 Watteyne, T., Bormann, C., and P. Thubert, "LLN Minimal 677 Fragment Forwarding", draft-watteyne-6lo-minimal- 678 fragment-02 (work in progress), July 2018. 680 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 681 Requirement Levels", BCP 14, RFC 2119, 682 DOI 10.17487/RFC2119, March 1997, 683 . 685 [RFC4944] Montenegro, G., Kushalnagar, N., Hui, J., and D. Culler, 686 "Transmission of IPv6 Packets over IEEE 802.15.4 687 Networks", RFC 4944, DOI 10.17487/RFC4944, September 2007, 688 . 690 [RFC6282] Hui, J., Ed. and P. Thubert, "Compression Format for IPv6 691 Datagrams over IEEE 802.15.4-Based Networks", RFC 6282, 692 DOI 10.17487/RFC6282, September 2011, 693 . 695 [RFC6554] Hui, J., Vasseur, JP., Culler, D., and V. Manral, "An IPv6 696 Routing Header for Source Routes with the Routing Protocol 697 for Low-Power and Lossy Networks (RPL)", RFC 6554, 698 DOI 10.17487/RFC6554, March 2012, 699 . 701 [RFC8025] Thubert, P., Ed. and R. Cragie, "IPv6 over Low-Power 702 Wireless Personal Area Network (6LoWPAN) Paging Dispatch", 703 RFC 8025, DOI 10.17487/RFC8025, November 2016, 704 . 706 [RFC8138] Thubert, P., Ed., Bormann, C., Toutain, L., and R. Cragie, 707 "IPv6 over Low-Power Wireless Personal Area Network 708 (6LoWPAN) Routing Header", RFC 8138, DOI 10.17487/RFC8138, 709 April 2017, . 711 [RFC8174] Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC 712 2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174, 713 May 2017, . 715 10.2. Informative References 717 [I-D.ietf-6tisch-architecture] 718 Thubert, P., "An Architecture for IPv6 over the TSCH mode 719 of IEEE 802.15.4", draft-ietf-6tisch-architecture-19 (work 720 in progress), December 2018. 722 [IEEE.802.15.4] 723 IEEE, "IEEE Standard for Low-Rate Wireless Networks", 724 IEEE Standard 802.15.4, DOI 10.1109/IEEE 725 P802.15.4-REVd/D01, 726 . 728 [RFC2914] Floyd, S., "Congestion Control Principles", BCP 41, 729 RFC 2914, DOI 10.17487/RFC2914, September 2000, 730 . 732 [RFC3031] Rosen, E., Viswanathan, A., and R. Callon, "Multiprotocol 733 Label Switching Architecture", RFC 3031, 734 DOI 10.17487/RFC3031, January 2001, 735 . 737 [RFC3168] Ramakrishnan, K., Floyd, S., and D. Black, "The Addition 738 of Explicit Congestion Notification (ECN) to IP", 739 RFC 3168, DOI 10.17487/RFC3168, September 2001, 740 . 742 [RFC4919] Kushalnagar, N., Montenegro, G., and C. Schumacher, "IPv6 743 over Low-Power Wireless Personal Area Networks (6LoWPANs): 744 Overview, Assumptions, Problem Statement, and Goals", 745 RFC 4919, DOI 10.17487/RFC4919, August 2007, 746 . 748 [RFC4963] Heffner, J., Mathis, M., and B. Chandler, "IPv4 Reassembly 749 Errors at High Data Rates", RFC 4963, 750 DOI 10.17487/RFC4963, July 2007, 751 . 753 [RFC5681] Allman, M., Paxson, V., and E. Blanton, "TCP Congestion 754 Control", RFC 5681, DOI 10.17487/RFC5681, September 2009, 755 . 757 [RFC6298] Paxson, V., Allman, M., Chu, J., and M. Sargent, 758 "Computing TCP's Retransmission Timer", RFC 6298, 759 DOI 10.17487/RFC6298, June 2011, 760 . 762 [RFC6606] Kim, E., Kaspar, D., Gomez, C., and C. Bormann, "Problem 763 Statement and Requirements for IPv6 over Low-Power 764 Wireless Personal Area Network (6LoWPAN) Routing", 765 RFC 6606, DOI 10.17487/RFC6606, May 2012, 766 . 768 [RFC7554] Watteyne, T., Ed., Palattella, M., and L. Grieco, "Using 769 IEEE 802.15.4e Time-Slotted Channel Hopping (TSCH) in the 770 Internet of Things (IoT): Problem Statement", RFC 7554, 771 DOI 10.17487/RFC7554, May 2015, 772 . 774 [RFC7567] Baker, F., Ed. and G. Fairhurst, Ed., "IETF 775 Recommendations Regarding Active Queue Management", 776 BCP 197, RFC 7567, DOI 10.17487/RFC7567, July 2015, 777 . 779 [RFC8085] Eggert, L., Fairhurst, G., and G. Shepherd, "UDP Usage 780 Guidelines", BCP 145, RFC 8085, DOI 10.17487/RFC8085, 781 March 2017, . 783 [RFC8087] Fairhurst, G. and M. Welzl, "The Benefits of Using 784 Explicit Congestion Notification (ECN)", RFC 8087, 785 DOI 10.17487/RFC8087, March 2017, 786 . 788 [RFC8200] Deering, S. and R. Hinden, "Internet Protocol, Version 6 789 (IPv6) Specification", STD 86, RFC 8200, 790 DOI 10.17487/RFC8200, July 2017, 791 . 793 [RFC8201] McCann, J., Deering, S., Mogul, J., and R. Hinden, Ed., 794 "Path MTU Discovery for IP version 6", STD 87, RFC 8201, 795 DOI 10.17487/RFC8201, July 2017, 796 . 798 Appendix A. Rationale 800 There are a number of uses for large packets in Wireless Sensor 801 Networks. Such usages may not be the most typical or represent the 802 largest amount of traffic over the LLN; however, the associated 803 functionality can be critical enough to justify extra care for 804 ensuring effective transport of large packets across the LLN. 806 The list of those usages includes: 808 Towards the LLN node: 810 Firmware update: For example, a new version of the LLN node 811 software is downloaded from a system manager over unicast or 812 multicast services. Such a reflashing operation typically 813 involves updating a large number of similar LLN nodes over a 814 relatively short period of time. 816 Packages of Commands: A number of commands or a full 817 configuration can be packaged as a single message to ensure 818 consistency and enable atomic execution or complete roll back. 819 Until such commands are fully received and interpreted, the 820 intended operation will not take effect. 822 From the LLN node: 824 Waveform captures: A number of consecutive samples are measured 825 at a high rate for a short time and then transferred from a 826 sensor to a gateway or an edge server as a single large report. 828 Data logs: LLN nodes may generate large logs of sampled data for 829 later extraction. LLN nodes may also generate system logs to 830 assist in diagnosing problems on the node or network. 832 Large data packets: Rich data types might require more than one 833 fragment. 835 Uncontrolled firmware download or waveform upload can easily result 836 in a massive increase of the traffic and saturate the network. 838 When a fragment is lost in transmission, the lack of recovery in the 839 original fragmentation system of RFC 4944 implies that all fragments 840 are resent, further contributing to the congestion that caused the 841 initial loss, and potentially leading to congestion collapse. 843 This saturation may lead to excessive radio interference, or random 844 early discard (leaky bucket) in relaying nodes. Additional queuing 845 and memory congestion may result while waiting for a low power next 846 hop to emerge from its sleeping state. 848 Considering that RFC 4944 defines an MTU is 1280 bytes and that in 849 most incarnations (but 802.15.4g) a IEEE Std. 802.15.4 frame can 850 limit the MAC payload to as few as 74 bytes, a packet might be 851 fragmented into at least 18 fragments at the 6LoWPAN shim layer. 852 Taking into account the worst-case header overhead for 6LoWPAN 853 Fragmentation and Mesh Addressing headers will increase the number of 854 required fragments to around 32. This level of fragmentation is much 855 higher than that traditionally experienced over the Internet with 856 IPv4 fragments. At the same time, the use of radios increases the 857 probability of transmission loss and Mesh-Under techniques compound 858 that risk over multiple hops. 860 Mechanisms such as TCP or application-layer segmentation could be 861 used to support end-to-end reliable transport. One option to support 862 bulk data transfer over a frame-size-constrained LLN is to set the 863 Maximum Segment Size to fit within the link maximum frame size. 864 Doing so, however, can add significant header overhead to each 865 802.15.4 frame. In addition, deploying such a mechanism requires 866 that the end-to-end transport is aware of the delivery properties of 867 the underlying LLN, which is a layer violation, and difficult to 868 achieve from the far end of the IPv6 network. 870 Appendix B. Requirements 872 For one-hop communications, a number of Low Power and Lossy Network 873 (LLN) link-layers propose a local acknowledgment mechanism that is 874 enough to detect and recover the loss of fragments. In a multihop 875 environment, an end-to-end fragment recovery mechanism might be a 876 good complement to a hop-by-hop MAC level recovery. This draft 877 introduces a simple protocol to recover individual fragments between 878 6LoWPAN endpoints that may be multiple hops away. The method 879 addresses the following requirements of a LLN: 881 Number of fragments 883 The recovery mechanism must support highly fragmented packets, 884 with a maximum of 32 fragments per packet. 886 Minimum acknowledgment overhead 888 Because the radio is half duplex, and because of silent time spent 889 in the various medium access mechanisms, an acknowledgment 890 consumes roughly as many resources as data fragment. 892 The new end-to-end fragment recovery mechanism should be able to 893 acknowledge multiple fragments in a single message and not require 894 an acknowledgment at all if fragments are already protected at a 895 lower layer. 897 Controlled latency 899 The recovery mechanism must succeed or give up within the time 900 boundary imposed by the recovery process of the Upper Layer 901 Protocols. 903 Optional congestion control 905 The aggregation of multiple concurrent flows may lead to the 906 saturation of the radio network and congestion collapse. 908 The recovery mechanism should provide means for controlling the 909 number of fragments in transit over the LLN. 911 Appendix C. Considerations On Flow Control 913 Considering that a multi-hop LLN can be a very sensitive environment 914 due to the limited queuing capabilities of a large population of its 915 nodes, this draft recommends a simple and conservative approach to 916 congestion control, based on TCP congestion avoidance. 918 Congestion on the forward path is assumed in case of packet loss, and 919 packet loss is assumed upon time out. The draft allows to control 920 the number of outstanding fragments, that have been transmitted but 921 for which an acknowledgment was not received yet. It must be noted 922 that the number of outstanding fragments should not exceed the number 923 of hops in the network, but the way to figure the number of hops is 924 out of scope for this document. 926 Congestion on the forward path can also be indicated by an Explicit 927 Congestion Notification (ECN) mechanism. Though whether and how ECN 928 [RFC3168] is carried out over the LoWPAN is out of scope, this draft 929 provides a way for the destination endpoint to echo an ECN indication 930 back to the source endpoint in an acknowledgment message as 931 represented in Figure 5 in Section 4.2. 933 It must be noted that congestion and collision are different topics. 934 In particular, when a mesh operates on a same channel over multiple 935 hops, then the forwarding of a fragment over a certain hop may 936 collide with the forwarding of a next fragment that is following over 937 a previous hop but in a same interference domain. This draft enables 938 an end-to-end flow control, but leaves it to the sender stack to pace 939 individual fragments within a transmit window, so that a given 940 fragment is sent only when the previous fragment has had a chance to 941 progress beyond the interference domain of this hop. In the case of 942 6TiSCH [I-D.ietf-6tisch-architecture], which operates over the 943 TimeSlotted Channel Hopping [RFC7554] (TSCH) mode of operation of 944 IEEE802.14.5, a fragment is forwarded over a different channel at a 945 different time and it makes full sense to transmit the next fragment 946 as soon as the previous fragment has had its chance to be forwarded 947 at the next hop. 949 From the standpoint of a source 6LoWPAN endpoint, an outstanding 950 fragment is a fragment that was sent but for which no explicit 951 acknowledgment was received yet. This means that the fragment might 952 be on the way, received but not yet acknowledged, or the 953 acknowledgment might be on the way back. It is also possible that 954 either the fragment or the acknowledgment was lost on the way. 956 From the sender standpoint, all outstanding fragments might still be 957 in the network and contribute to its congestion. There is an 958 assumption, though, that after a certain amount of time, a frame is 959 either received or lost, so it is not causing congestion anymore. 960 This amount of time can be estimated based on the round trip delay 961 between the 6LoWPAN endpoints. The method detailed in [RFC6298] is 962 recommended for that computation. 964 The reader is encouraged to read through "Congestion Control 965 Principles" [RFC2914]. Additionally [RFC7567] and [RFC5681] provide 966 deeper information on why this mechanism is needed and how TCP 967 handles Congestion Control. Basically, the goal here is to manage 968 the amount of fragments present in the network; this is achieved by 969 to reducing the number of outstanding fragments over a congested path 970 by throttling the sources. 972 Section 5 describes how the sender decides how many fragments are 973 (re)sent before an acknowledgment is required, and how the sender 974 adapts that number to the network conditions. 976 Author's Address 978 Pascal Thubert (editor) 979 Cisco Systems, Inc 980 Building D 981 45 Allee des Ormes - BP1200 982 MOUGINS - Sophia Antipolis 06254 983 FRANCE 985 Phone: +33 497 23 26 34 986 Email: pthubert@cisco.com