idnits 2.17.1 draft-ietf-6lo-fragment-recovery-02.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- 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 23, 2019) is 1921 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 'I-D.watteyne-6lo-minimal-fragment') == Outdated reference: A later version (-30) exists of draft-ietf-6tisch-architecture-19 Summary: 1 error (**), 0 flaws (~~), 2 warnings (==), 2 comments (--).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 23, 2019 5 Intended status: Standards Track 6 Expires: July 27, 2019 8 6LoWPAN Selective Fragment Recovery 9 draft-ietf-6lo-fragment-recovery-02 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 27, 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. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 3 53 2.1. BCP 14 . . . . . . . . . . . . . . . . . . . . . . . . . 3 54 2.2. References . . . . . . . . . . . . . . . . . . . . . . . 4 55 2.3. 6LoWPAN Acronyms . . . . . . . . . . . . . . . . . . . . 4 56 2.4. Referenced Work . . . . . . . . . . . . . . . . . . . . . 4 57 2.5. New Terms . . . . . . . . . . . . . . . . . . . . . . . . 5 58 3. Updating RFC 4944 . . . . . . . . . . . . . . . . . . . . . . 5 59 4. Updating draft-watteyne-6lo-minimal-fragment . . . . . . . . 6 60 4.1. Slack in the First Fragment . . . . . . . . . . . . . . . 6 61 4.2. Modifying the First Fragment . . . . . . . . . . . . . . 6 62 5. New Dispatch types and headers . . . . . . . . . . . . . . . 7 63 5.1. Recoverable Fragment Dispatch type and Header . . . . . . 8 64 5.2. RFRAG Acknowledgment Dispatch type and Header . . . . . . 9 65 6. Fragments Recovery . . . . . . . . . . . . . . . . . . . . . 11 66 7. Forwarding Fragments . . . . . . . . . . . . . . . . . . . . 13 67 7.1. Upon the first fragment . . . . . . . . . . . . . . . . . 13 68 7.2. Upon the next fragments . . . . . . . . . . . . . . . . . 13 69 7.3. Upon the RFRAG Acknowledgments . . . . . . . . . . . . . 14 70 8. Security Considerations . . . . . . . . . . . . . . . . . . . 14 71 9. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 15 72 10. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 15 73 11. References . . . . . . . . . . . . . . . . . . . . . . . . . 15 74 11.1. Normative References . . . . . . . . . . . . . . . . . . 15 75 11.2. 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. Terminology 140 2.1. BCP 14 142 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 143 "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and 144 "OPTIONAL" in this document are to be interpreted as described in BCP 145 14 [RFC2119][RFC8174] when, and only when, they appear in all 146 capitals, as shown here. 148 2.2. References 150 In this document, readers will encounter terms and concepts that are 151 discussed in the following documents: 153 o "Problem Statement and Requirements for IPv6 over Low-Power 154 Wireless Personal Area Network (6LoWPAN) Routing" [RFC6606] 156 2.3. 6LoWPAN Acronyms 158 This document uses the following acronyms: 160 6BBR: 6LoWPAN Backbone Router 162 6LBR: 6LoWPAN Border Router 164 6LN: 6LoWPAN Node 166 6LR: 6LoWPAN Router 168 LLN: Low-Power and Lossy Network 170 2.4. Referenced Work 172 Past experience with fragmentation has shown that miss-associated or 173 lost fragments can lead to poor network behavior and, occasionally, 174 trouble at application layer. The reader is encouraged to read "IPv4 175 Reassembly Errors at High Data Rates" [RFC4963] and follow the 176 references for more information. 178 That experience led to the definition of "Path MTU discovery" 179 [RFC8201] (PMTUD) protocol that limits fragmentation over the 180 Internet. 182 Specifically in the case of UDP, valuable additional information can 183 be found in "UDP Usage Guidelines for Application Designers" 184 [RFC8085]. 186 Readers are expected to be familiar with all the terms and concepts 187 that are discussed in "IPv6 over Low-Power Wireless Personal Area 188 Networks (6LoWPANs): Overview, Assumptions, Problem Statement, and 189 Goals" [RFC4919] and "Transmission of IPv6 Packets over IEEE 802.15.4 190 Networks" [RFC4944]. 192 "The Benefits of Using Explicit Congestion Notification (ECN)" 193 [RFC8087] provides useful information on the potential benefits and 194 pitfalls of using ECN. 196 Quoting the "Multiprotocol Label Switching (MPLS) Architecture" 197 [RFC3031]: with MPLS, "packets are "labeled" before they are 198 forwarded. At subsequent hops, there is no further analysis of the 199 packet's network layer header. Rather, the label is used as an index 200 into a table which specifies the next hop, and a new label". The 201 MPLS technique is leveraged in the present specification to forward 202 fragments that actually do not have a network layer header, since the 203 fragmentation occurs below IP. 205 "LLN Minimal Fragment Forwarding" [I-D.watteyne-6lo-minimal-fragment] 206 introduces the concept of a Virtual Reassembly Buffer (VRB) and an 207 associated technique to forward fragments as they come, using the 208 datagram_tag as a label in a fashion similar to MLPS. This 209 specification reuses that technique with slightly modified controls. 211 2.5. New Terms 213 This specification uses the following terms: 215 6LoWPAN endpoints 217 The LLN nodes in charge of generating or expanding a 6LoWPAN 218 header from/to a full IPv6 packet. The 6LoWPAN endpoints are the 219 points where fragmentation and reassembly take place. 221 3. Updating RFC 4944 223 This specification updates the fragmentation mechanism that is 224 specified in "Transmission of IPv6 Packets over IEEE 802.15.4 225 Networks" [RFC4944] for use in route-over LLNs by providing a model 226 where fragments can be forwarded end-to-end across a 6LoWPAN LLN, and 227 where fragments that are lost on the way can be recovered 228 individually. A new format for fragment is introduces and new 229 dispatch types are defined in Section 5. 231 [RFC8138] allows to modifies the size of a packet en-route by 232 removing the consumed hops in a compressed Routing Header. It 233 results that the fragment_offset and datagram_size cannot be signaled 234 in the uncompressed form. This specification expresses those fields 235 in the compressed form and allows to modify them en-route (see 236 Section 4.2. 238 Note that consistantly with in Section 2 of [RFC6282] for the 239 fragmentation mechanism described in Section 5.3 of [RFC4944], any 240 header that cannot fit within the first fragment MUST NOT be 241 compressed when using the fragmentation mechanism described in this 242 specification. 244 4. Updating draft-watteyne-6lo-minimal-fragment 246 This specification updates the fragment forwarding mechanism 247 specified in "LLN Minimal Fragment Forwarding" 248 [I-D.watteyne-6lo-minimal-fragment] by providing additional 249 operations to improve the management of the Virtual Reassembly Buffer 250 (VRB). 252 4.1. Slack in the First Fragment 254 At the time of this writing, [I-D.watteyne-6lo-minimal-fragment] 255 allows for refragmenting in intermediate nodes, meaning that some 256 bytes from a given fragment may be left in the VRB to be added to the 257 next fragment. The reason for this to happen would be the need for 258 space in the outgoing fragment that was not needed in the incoming 259 fragment, for instance because the 6LoWPAN Header Compression is not 260 as efficient on the outgoing link, e.g., if the Interface ID (IID) of 261 the source IPv6 address is elided by the originator on the first hop 262 because it matches the source MAC address, but cannot be on the next 263 hops because the source MAC address changes. 265 This specification cannot allow this operation since fragments are 266 recovered end-to-end based on the fragment number. This means that 267 the fragments that contain a 6LoWPAN-compressed header MUST have 268 enough slack to enable a less efficient compression in the next hops 269 that still fits in one MAC frame. For instance, if the IID of the 270 source IPv6 address is elided by the originator, then it MUST compute 271 the fragment_size as if the MTU was 8 bytes less. This way, the next 272 hop can restore the source IID to the first fragment without 273 impacting the second fragment. 275 4.2. Modifying the First Fragment 277 The compression of the Hop Limit, of the source and destination 278 addresses, and of the Routing Header may change en route in a Route- 279 Over mesh LLN. If the size of the first fragment is modified, then 280 the intermediate node MUST adapt the datagram_size to reflect that 281 difference. 283 The intermediate node MUST also save the difference of datagram_size 284 of the first fragment in the VRB, and add it to the datagram_size and 285 to the fragment_offset of all the subsequent fragments for that 286 datagram. 288 5. New Dispatch types and headers 290 This specification enables the 6LoWPAN fragmentation sublayer to 291 provide an MTU up to 2048 bytes to the upper layer, which can be the 292 6LoWPAN Header Compression sublayer that is defined in the 293 "Compression Format for IPv6 Datagrams" [RFC6282] specification. In 294 order to achieve this, this specification enables the fragmentation 295 and the reliable transmission of fragments over a multihop 6LoWPAN 296 mesh network. 298 This specification provides a technique that is derived from MPLS in 299 order to forward individual fragments across a 6LoWPAN route-over 300 mesh. The datagram_tag is used as a label; it is locally unique to 301 the node that is the source MAC address of the fragment, so together 302 the MAC address and the label can identify the fragment globally. A 303 node may build the datagram_tag in its own locally-significant way, 304 as long as the selected tag stays unique to the particular datagram 305 for the lifetime of that datagram. It results that the label does 306 not need to be globally unique but also that it must be swapped at 307 each hop as the source MAC address changes. 309 This specification extends RFC 4944 [RFC4944] with 4 new Dispatch 310 types, for Recoverable Fragment (RFRAG) headers with or without 311 Acknowledgment Request (RFRAG vs. RFRAG-ARQ), and for the RFRAG 312 Acknowledgment back, with or without ECN Echo (RFRAG-ACK vs. RFRAG- 313 ECHO). 315 (to be confirmed by IANA) The new 6LoWPAN Dispatch types use the 316 Value Bit Pattern of 11 1010xx from page 0 [RFC8025], as follows: 318 Pattern Header Type 319 +------------+------------------------------------------+ 320 | 11 101000 | RFRAG - Recoverable Fragment | 321 | 11 101001 | RFRAG-ARQ - RFRAG with Ack Request | 322 | 11 101010 | RFRAG-ACK - RFRAG Acknowledgment | 323 | 11 101011 | RFRAG-ECHO - RFRAG Ack with ECN Echo | 324 +------------+------------------------------------------+ 326 Figure 1: Additional Dispatch Value Bit Patterns 328 In the following sections, the semantics of "datagram_tag" are 329 unchanged from [RFC4944] Section 5.3. "Fragmentation Type and 330 Header." and is compatible with the fragment forwarding operation 331 described in [I-D.watteyne-6lo-minimal-fragment]. 333 5.1. Recoverable Fragment Dispatch type and Header 335 In this specification, the size and offset of the fragments are 336 expressed on the compressed packet form as opposed to the 337 uncompressed - native - packet form. 339 The first fragment is recognized by a sequence of 0; it carries its 340 fragment_size and the datagram_size of the compressed packet, whereas 341 the other fragments carry their fragment_size and fragment_offset. 342 The last fragment for a datagram is recognized when its 343 fragment_offset and its fragment_size add up to the datagram_size. 345 Recoverable Fragments are sequenced and a bitmap is used in the RFRAG 346 Acknowledgment to indicate the received fragments by setting the 347 individual bits that correspond to their sequence. 349 1 2 3 350 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 1 351 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 352 |1 1 1 0 1 0 0|E| datagram_tag | 353 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 354 |X| sequence| fragment_size | fragment_offset | 355 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 357 X set == Ack Requested 359 Figure 2: RFRAG Dispatch type and Header 361 E: 1 bit; Explicit Congestion Notification; the "E" flag is reset by 362 the source of the fragment and set by intermediate routers to 363 signal that this fragment experienced congestion along its path. 365 Fragment_size: 10 bit unsigned integer; the size of this fragment in 366 a unit that depends on the MAC layer technology. For IEEE Std. 367 802.15.4, the unit is octet, and the maximum fragment size, which 368 is constrained by the maximum frame size of 128 octet minus the 369 overheads of the MAC and Fragment Headers, is not limited by this 370 encoding. 372 X: 1 bit; Ack Requested: when set, the sender requires an RFRAG 373 Acknowledgment from the receiver. 375 Sequence: 5 bit unsigned integer; the sequence number of the 376 fragment. Fragments are sequence numbered [0..N] where N is in 377 [0..31]. A sequence of 0 indicates the first fragment in a 378 datagram. For IEEE Std. 802.15.4, as long as the overheads enable 379 a fragment size of 64 octets or more, this enables to fragment a 380 packet of 2047 octets. 382 Fragment_offset: 16 bit unsigned integer; 384 * When set to a non-0 value, the semantics of the Fragment_offset 385 depends on the value of the Sequence. 387 + When the Sequence is not 0, this field indicates the offset 388 of the fragment in the compressed form. The fragment should 389 be forwarded based on an existing VRB as described in 390 Section 7.2, or silently dropped if none is found. 392 + For a first fragment (i.e. with a sequence of 0), this field 393 is overloaded to indicate the total_size of the compressed 394 packet, to help the receiver allocate an adapted buffer for 395 the reception and reassembly operations. This format limits 396 the maximum MTU on a 6LoWPAN link to 2047 bytes, but 1280 397 bytes is the recommended value to avoid issues with IPV6 398 Path MTU Discovery [RFC8201]. The fragment should be routed 399 based on the destination IPv6 address, and an VRB state 400 should be installed as described in Section 7.1. 402 * When set to 0, this field indicates an abort condition and all 403 state regarding the datagram should be cleaned up once the 404 processing of the fragment is complete; the processing of the 405 fragment depends on whether there is a VRB already established 406 for this datagram, and the next hop is still reachable: 408 + if a VRB already exists and is not broken, the fragment is 409 to be forwarded along the associated Label Switched Path 410 (LSP) as described in Section 7.2, but regardless of the 411 value of the Sequence field; 413 + else, if the Sequence is 0, then the fragment is to be 414 routed as described in Section 7.1 but no state is conserved 415 afterwards. 417 If the fragment cannot be forwarded or routed, then an abort 418 RFRAG-ACK is sent back to the source. 420 5.2. RFRAG Acknowledgment Dispatch type and Header 422 This specification also defines a 4-octet RFRAG Acknowledgment bitmap 423 that is used by the reassembling end point to confirm selectively the 424 reception of individual fragments. A given offset in the bitmap maps 425 one to one with a given sequence number. 427 The offset of the bit in the bitmap indicates which fragment is 428 acknowledged as follows: 430 1 2 3 431 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 1 432 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 433 | RFRAG Acknowledgment Bitmap | 434 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 435 ^ ^ 436 | | bitmap indicating whether: 437 | +--- Fragment with sequence 10 was received 438 +----------------------- Fragment with sequence 00 was received 440 Figure 3: RFRAG Acknowledgment bitmap encoding 442 Figure 4 shows an example Acknowledgment bitmap which indicates that 443 all fragments from sequence 0 to 20 were received, except for 444 fragments 1, 2 and 16 that were either lost or are still in the 445 network over a slower path. 447 1 2 3 448 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 1 449 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 450 |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| 451 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 453 Figure 4: Expanding 3 octets encoding 455 The RFRAG Acknowledgment Bitmap is included in a RFRAG Acknowledgment 456 header, as follows: 458 1 2 3 459 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 1 460 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 461 |1 1 1 0 1 0 1 Y| datagram_tag | 462 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 463 | RFRAG Acknowledgment Bitmap (32 bits) | 464 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 466 Figure 5: RFRAG Acknowledgment Dispatch type and Header 468 Y: 1 bit; Explicit Congestion Notification Echo 470 When set, the sender indicates that at least one of the 471 acknowledged fragments was received with an Explicit Congestion 472 Notification, indicating that the path followed by the fragments 473 is subject to congestion. 475 RFRAG Acknowledgment Bitmap 477 An RFRAG Acknowledgment Bitmap, whereby setting the bit at offset 478 x indicates that fragment x was received, as shown in Figure 3. 479 All 0's is a NULL bitmap that indicates that the fragmentation 480 process is aborted. All 1's is a FULL bitmap that indicates that 481 the fragmentation process is complete, all fragments were received 482 at the reassembly end point. 484 6. Fragments Recovery 486 The Recoverable Fragment headers RFRAG and RFRAG-ARQ are used to 487 transport a fragment and optionally request an RFRAG Acknowledgment 488 that will confirm the good reception of a one or more fragments. An 489 RFRAG Acknowledgment can optionally carry an ECN indication; it is 490 carried as a standalone header in a message that is sent back to the 491 6LoWPAN endpoint that was the source of the fragments, as known by 492 its MAC address. The process ensures that at every hop, the source 493 MAC address and the datagram_tag in the received fragment are enough 494 information to send the RFRAG Acknowledgment back towards the source 495 6LoWPAN endpoint by reversing the MPLS operation. 497 The 6LoWPAN endpoint that fragments the packets at 6LoWPAN level (the 498 sender) also controls when the reassembling end point sends the RFRAG 499 Acknowledgments by setting the Ack Requested flag in the RFRAG 500 packets. It may set the Ack Requested flag on any fragment to 501 perform congestion control by limiting the number of outstanding 502 fragments, which are the fragments that have been sent but for which 503 reception or loss was not positively confirmed by the reassembling 504 endpoint. When the sender of the fragment knows that an underlying 505 link-layer mechanism protects the Fragments, it may refrain from 506 using the RFRAG Acknowledgment mechanism, and never set the Ack 507 Requested bit. When it receives a fragment with the ACK Request flag 508 set, the 6LoWPAN endpoint that reassembles the packets at 6LoWPAN 509 level (the receiver) sends back an RFRAG Acknowledgment to confirm 510 reception of all the fragments it has received so far. 512 The sender transfers a controlled number of fragments and MAY flag 513 the last fragment of a series with an RFRAG Acknowledgment Request. 514 The received MUST acknowledge a fragment with the acknowledgment 515 request bit set. If any fragment immediately preceding an 516 acknowledgment request is still missing, the receiver MAY 517 intentionally delay its acknowledgment to allow in-transit fragments 518 to arrive. Delaying the acknowledgment might defeat the round trip 519 delay computation so it should be configurable and not enabled by 520 default. 522 The receiver MAY issue unsolicited acknowledgments. An unsolicited 523 acknowledgment signals to the sender endpoint that it can resume 524 sending if it had reached its maximum number of outstanding 525 fragments. Another use is to inform that the reassembling endpoint 526 has canceled the process of an individual datagram. Note that 527 acknowledgments might consume precious resources so the use of 528 unsolicited acknowledgments should be configurable and not enabled by 529 default. 531 An observation is that streamlining forwarding of fragments generally 532 reduces the latency over the LLN mesh, providing room for retries 533 within existing upper-layer reliability mechanisms. The sender 534 protects the transmission over the LLN mesh with a retry timer that 535 is computed according to the method detailed in [RFC6298]. It is 536 expected that the upper layer retries obey the recommendations in 537 "UDP Usage Guidelines" [RFC8085], in which case a single round of 538 fragment recovery should fit within the upper layer recovery timers. 540 Fragments are sent in a round robin fashion: the sender sends all the 541 fragments for a first time before it retries any lost fragment; lost 542 fragments are retried in sequence, oldest first. This mechanism 543 enables the receiver to acknowledge fragments that were delayed in 544 the network before they are actually retried. 546 When a single frequency is used by contiguous hops, the sender should 547 wait a reasonable amount of time between fragments so as to let a 548 fragment progress a few hops and avoid hidden terminal issues. This 549 precaution is not required on channel hopping technologies such as 550 Time Slotted CHannel Hopping (TSCH) [RFC6554] 552 When the sender decides that a packet should be dropped and the 553 fragmentation process canceled, it sends a pseudo fragment with the 554 fragment_offset, sequence and fragment_size all set to 0, and no 555 data. Upon reception of this message, the receiver should clean up 556 all resources for the packet associated to the datagram_tag. If an 557 acknowledgment is requested, the receiver responds with a NULL 558 bitmap. 560 The receiver might need to cancel the process of a fragmented packet 561 for internal reasons, for instance if it is out of reassembly 562 buffers, or considers that this packet is already fully reassembled 563 and passed to the upper layer. In that case, the receiver SHOULD 564 indicate so to the sender with a NULL bitmap in a RFRAG 565 Acknowledgment. Upon an acknowledgment with a NULL bitmap, the 566 sender endpoint MUST abort the transmission of the fragmented 567 datagram. 569 7. Forwarding Fragments 571 It is assumed that the first Fragment is large enough to carry the 572 IPv6 header and make routing decisions. If that is not so, then this 573 specification MUST NOT be used. 575 This specification extends the Virtual Reassembly Buffer (VRB) 576 technique to forward fragments with no intermediate reconstruction of 577 the entire packet. The first fragment carries the IP header and it 578 is routed all the way from the fragmenting end point to the 579 reassembling end point. Upon the first fragment, the routers along 580 the path install a label-switched path (LSP), and the following 581 fragments are label-switched along that path. As a consequence, 582 alternate routes not possible for individual fragments. The 583 datagram_tag is used to carry the label, that is swapped at each hop. 584 All fragments follow the same path and fragments are delivered in the 585 order at which they are sent. 587 7.1. Upon the first fragment 589 In Route-Over mode, the source and destination MAC addressed in a 590 frame change at each hop. The label that is formed and placed in the 591 datagram_tag is associated to the source MAC and only valid (and 592 unique) for that source MAC. Upon a first fragment (i.e. with a 593 sequence of zero), a VRB and the associated LSP state are created for 594 the tuple (source MAC address, datagram_tag) and the fragment is 595 forwarded along the IPv6 route that matches the destination IPv6 596 address in the IPv6 header as prescribed by 597 [I-D.watteyne-6lo-minimal-fragment]. The LSP state enables to match 598 the (previous MAC address, datagram_tag) in an incoming fragment to 599 the tuple (next MAC address, swapped datagram_tag) used in the 600 forwarded fragment and points at the VRB. In addition, the router 601 also forms a Reverse LSP state indexed by the MAC address of the next 602 hop and the swapped datagram_tag. This reverse LSP state also points 603 at the VRB and enables to match the (next MAC address, 604 swapped_datagram_tag) found in an RFRAG Acknowledgment to the tuple 605 (previous MAC address, datagram_tag) used when forwarding a Fragment 606 Acknowledgment (RFRAG-ACK) back to the sender endpoint. 608 7.2. Upon the next fragments 610 Upon a next fragment (i.e. with a non-zero sequence), the router 611 looks up a LSP indexed by the tuple (MAC address, datagram_tag) found 612 in the fragment. If it is found, the router forwards the fragment 613 using the associated VRB as prescribed by 614 [I-D.watteyne-6lo-minimal-fragment]. 616 if the VRB for the tuple is not found, the router builds an RFRAG-ACK 617 to abort the transmission of the packet. The resulting message has 618 the following information: 620 o The source and destination MAC addresses are swapped from those 621 found in the fragment 623 o The datagram_tag set to the datagram_tag found in the fragment 625 o A null bitmap is used to signal the abort condition 627 At this point the router is all set and can send the RFRAG-ACK back 628 to the previous router. The RFRAG-ACK should normally be forwarded 629 all the way to the source using the reverse LSP state in the VRBs in 630 the intermediate routers as described in the next section. 632 7.3. Upon the RFRAG Acknowledgments 634 Upon an RFRAG-ACK, the router looks up a Reverse LSP indexed by the 635 tuple (MAC address, datagram_tag), which are respectively the source 636 MAC address of the received frame and the received datagram_tag. If 637 it is found, the router forwards the fragment using the associated 638 VRB as prescribed by [I-D.watteyne-6lo-minimal-fragment], but using 639 the Reverse LSP so that the RFRAG-ACK flows back to the sender 640 endpoint. 642 If the Reverse LSP is not found, the router MUST silently drop the 643 RFRAG-ACK message. 645 Either way, if the RFRAG-ACK indicates either an error (NULL bitmap) 646 or that the fragment was entirely received (FULL bitmap), arms a 647 short timer, and upon timeout, the VRB and all associate state are 648 destroyed. During that time, fragments of that datagram may still be 649 received, e.g. if the RFRAG-ACK was lost on the way back and the 650 source retried the last fragment. In that case, the router sends an 651 abort RFRAG-ACK along the Reverse LSP to complete the clean up. 653 8. Security Considerations 655 The process of recovering fragments does not appear to create any 656 opening for new threat compared to "Transmission of IPv6 Packets over 657 IEEE 802.15.4 Networks" [RFC4944]. 659 9. IANA Considerations 661 Need extensions for formats defined in "Transmission of IPv6 Packets 662 over IEEE 802.15.4 Networks" [RFC4944]. 664 10. Acknowledgments 666 The author wishes to thank Thomas Watteyne and Michael Richardson for 667 in-depth reviews and comments. Also many thanks to Jonathan Hui, Jay 668 Werb, Christos Polyzois, Soumitri Kolavennu, Pat Kinney, Margaret 669 Wasserman, Richard Kelsey, Carsten Bormann and Harry Courtice for 670 their various contributions. 672 11. References 674 11.1. Normative References 676 [I-D.watteyne-6lo-minimal-fragment] 677 Watteyne, T., Bormann, C., and P. Thubert, "LLN Minimal 678 Fragment Forwarding", draft-watteyne-6lo-minimal- 679 fragment-02 (work in progress), July 2018. 681 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 682 Requirement Levels", BCP 14, RFC 2119, 683 DOI 10.17487/RFC2119, March 1997, 684 . 686 [RFC4944] Montenegro, G., Kushalnagar, N., Hui, J., and D. Culler, 687 "Transmission of IPv6 Packets over IEEE 802.15.4 688 Networks", RFC 4944, DOI 10.17487/RFC4944, September 2007, 689 . 691 [RFC6282] Hui, J., Ed. and P. Thubert, "Compression Format for IPv6 692 Datagrams over IEEE 802.15.4-Based Networks", RFC 6282, 693 DOI 10.17487/RFC6282, September 2011, 694 . 696 [RFC6554] Hui, J., Vasseur, JP., Culler, D., and V. Manral, "An IPv6 697 Routing Header for Source Routes with the Routing Protocol 698 for Low-Power and Lossy Networks (RPL)", RFC 6554, 699 DOI 10.17487/RFC6554, March 2012, 700 . 702 [RFC8025] Thubert, P., Ed. and R. Cragie, "IPv6 over Low-Power 703 Wireless Personal Area Network (6LoWPAN) Paging Dispatch", 704 RFC 8025, DOI 10.17487/RFC8025, November 2016, 705 . 707 [RFC8138] Thubert, P., Ed., Bormann, C., Toutain, L., and R. Cragie, 708 "IPv6 over Low-Power Wireless Personal Area Network 709 (6LoWPAN) Routing Header", RFC 8138, DOI 10.17487/RFC8138, 710 April 2017, . 712 [RFC8174] Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC 713 2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174, 714 May 2017, . 716 11.2. Informative References 718 [I-D.ietf-6tisch-architecture] 719 Thubert, P., "An Architecture for IPv6 over the TSCH mode 720 of IEEE 802.15.4", draft-ietf-6tisch-architecture-19 (work 721 in progress), December 2018. 723 [IEEE.802.15.4] 724 IEEE, "IEEE Standard for Low-Rate Wireless Networks", 725 IEEE Standard 802.15.4, DOI 10.1109/IEEE 726 P802.15.4-REVd/D01, 727 . 729 [RFC2914] Floyd, S., "Congestion Control Principles", BCP 41, 730 RFC 2914, DOI 10.17487/RFC2914, September 2000, 731 . 733 [RFC3031] Rosen, E., Viswanathan, A., and R. Callon, "Multiprotocol 734 Label Switching Architecture", RFC 3031, 735 DOI 10.17487/RFC3031, January 2001, 736 . 738 [RFC3168] Ramakrishnan, K., Floyd, S., and D. Black, "The Addition 739 of Explicit Congestion Notification (ECN) to IP", 740 RFC 3168, DOI 10.17487/RFC3168, September 2001, 741 . 743 [RFC4919] Kushalnagar, N., Montenegro, G., and C. Schumacher, "IPv6 744 over Low-Power Wireless Personal Area Networks (6LoWPANs): 745 Overview, Assumptions, Problem Statement, and Goals", 746 RFC 4919, DOI 10.17487/RFC4919, August 2007, 747 . 749 [RFC4963] Heffner, J., Mathis, M., and B. Chandler, "IPv4 Reassembly 750 Errors at High Data Rates", RFC 4963, 751 DOI 10.17487/RFC4963, July 2007, 752 . 754 [RFC5681] Allman, M., Paxson, V., and E. Blanton, "TCP Congestion 755 Control", RFC 5681, DOI 10.17487/RFC5681, September 2009, 756 . 758 [RFC6298] Paxson, V., Allman, M., Chu, J., and M. Sargent, 759 "Computing TCP's Retransmission Timer", RFC 6298, 760 DOI 10.17487/RFC6298, June 2011, 761 . 763 [RFC6606] Kim, E., Kaspar, D., Gomez, C., and C. Bormann, "Problem 764 Statement and Requirements for IPv6 over Low-Power 765 Wireless Personal Area Network (6LoWPAN) Routing", 766 RFC 6606, DOI 10.17487/RFC6606, May 2012, 767 . 769 [RFC7554] Watteyne, T., Ed., Palattella, M., and L. Grieco, "Using 770 IEEE 802.15.4e Time-Slotted Channel Hopping (TSCH) in the 771 Internet of Things (IoT): Problem Statement", RFC 7554, 772 DOI 10.17487/RFC7554, May 2015, 773 . 775 [RFC7567] Baker, F., Ed. and G. Fairhurst, Ed., "IETF 776 Recommendations Regarding Active Queue Management", 777 BCP 197, RFC 7567, DOI 10.17487/RFC7567, July 2015, 778 . 780 [RFC8085] Eggert, L., Fairhurst, G., and G. Shepherd, "UDP Usage 781 Guidelines", BCP 145, RFC 8085, DOI 10.17487/RFC8085, 782 March 2017, . 784 [RFC8087] Fairhurst, G. and M. Welzl, "The Benefits of Using 785 Explicit Congestion Notification (ECN)", RFC 8087, 786 DOI 10.17487/RFC8087, March 2017, 787 . 789 [RFC8200] Deering, S. and R. Hinden, "Internet Protocol, Version 6 790 (IPv6) Specification", STD 86, RFC 8200, 791 DOI 10.17487/RFC8200, July 2017, 792 . 794 [RFC8201] McCann, J., Deering, S., Mogul, J., and R. Hinden, Ed., 795 "Path MTU Discovery for IP version 6", STD 87, RFC 8201, 796 DOI 10.17487/RFC8201, July 2017, 797 . 799 Appendix A. Rationale 801 There are a number of uses for large packets in Wireless Sensor 802 Networks. Such usages may not be the most typical or represent the 803 largest amount of traffic over the LLN; however, the associated 804 functionality can be critical enough to justify extra care for 805 ensuring effective transport of large packets across the LLN. 807 The list of those usages includes: 809 Towards the LLN node: 811 Firmware update: For example, a new version of the LLN node 812 software is downloaded from a system manager over unicast or 813 multicast services. Such a reflashing operation typically 814 involves updating a large number of similar LLN nodes over a 815 relatively short period of time. 817 Packages of Commands: A number of commands or a full 818 configuration can be packaged as a single message to ensure 819 consistency and enable atomic execution or complete roll back. 820 Until such commands are fully received and interpreted, the 821 intended operation will not take effect. 823 From the LLN node: 825 Waveform captures: A number of consecutive samples are measured 826 at a high rate for a short time and then transferred from a 827 sensor to a gateway or an edge server as a single large report. 829 Data logs: LLN nodes may generate large logs of sampled data for 830 later extraction. LLN nodes may also generate system logs to 831 assist in diagnosing problems on the node or network. 833 Large data packets: Rich data types might require more than one 834 fragment. 836 Uncontrolled firmware download or waveform upload can easily result 837 in a massive increase of the traffic and saturate the network. 839 When a fragment is lost in transmission, the lack of recovery in the 840 original fragmentation system of RFC 4944 implies that all fragments 841 are resent, further contributing to the congestion that caused the 842 initial loss, and potentially leading to congestion collapse. 844 This saturation may lead to excessive radio interference, or random 845 early discard (leaky bucket) in relaying nodes. Additional queuing 846 and memory congestion may result while waiting for a low power next 847 hop to emerge from its sleeping state. 849 Considering that RFC 4944 defines an MTU is 1280 bytes and that in 850 most incarnations (but 802.15.4g) a IEEE Std. 802.15.4 frame can 851 limit the MAC payload to as few as 74 bytes, a packet might be 852 fragmented into at least 18 fragments at the 6LoWPAN shim layer. 853 Taking into account the worst-case header overhead for 6LoWPAN 854 Fragmentation and Mesh Addressing headers will increase the number of 855 required fragments to around 32. This level of fragmentation is much 856 higher than that traditionally experienced over the Internet with 857 IPv4 fragments. At the same time, the use of radios increases the 858 probability of transmission loss and Mesh-Under techniques compound 859 that risk over multiple hops. 861 Mechanisms such as TCP or application-layer segmentation could be 862 used to support end-to-end reliable transport. One option to support 863 bulk data transfer over a frame-size-constrained LLN is to set the 864 Maximum Segment Size to fit within the link maximum frame size. 865 Doing so, however, can add significant header overhead to each 866 802.15.4 frame. In addition, deploying such a mechanism requires 867 that the end-to-end transport is aware of the delivery properties of 868 the underlying LLN, which is a layer violation, and difficult to 869 achieve from the far end of the IPv6 network. 871 Appendix B. Requirements 873 For one-hop communications, a number of Low Power and Lossy Network 874 (LLN) link-layers propose a local acknowledgment mechanism that is 875 enough to detect and recover the loss of fragments. In a multihop 876 environment, an end-to-end fragment recovery mechanism might be a 877 good complement to a hop-by-hop MAC level recovery. This draft 878 introduces a simple protocol to recover individual fragments between 879 6LoWPAN endpoints that may be multiple hops away. The method 880 addresses the following requirements of a LLN: 882 Number of fragments 884 The recovery mechanism must support highly fragmented packets, 885 with a maximum of 32 fragments per packet. 887 Minimum acknowledgment overhead 889 Because the radio is half duplex, and because of silent time spent 890 in the various medium access mechanisms, an acknowledgment 891 consumes roughly as many resources as data fragment. 893 The new end-to-end fragment recovery mechanism should be able to 894 acknowledge multiple fragments in a single message and not require 895 an acknowledgment at all if fragments are already protected at a 896 lower layer. 898 Controlled latency 900 The recovery mechanism must succeed or give up within the time 901 boundary imposed by the recovery process of the Upper Layer 902 Protocols. 904 Optional congestion control 906 The aggregation of multiple concurrent flows may lead to the 907 saturation of the radio network and congestion collapse. 909 The recovery mechanism should provide means for controlling the 910 number of fragments in transit over the LLN. 912 Appendix C. Considerations On Flow Control 914 Considering that a multi-hop LLN can be a very sensitive environment 915 due to the limited queuing capabilities of a large population of its 916 nodes, this draft recommends a simple and conservative approach to 917 congestion control, based on TCP congestion avoidance. 919 Congestion on the forward path is assumed in case of packet loss, and 920 packet loss is assumed upon time out. The draft allows to control 921 the number of outstanding fragments, that have been transmitted but 922 for which an acknowledgment was not received yet. It must be noted 923 that the number of outstanding fragments should not exceed the number 924 of hops in the network, but the way to figure the number of hops is 925 out of scope for this document. 927 Congestion on the forward path can also be indicated by an Explicit 928 Congestion Notification (ECN) mechanism. Though whether and how ECN 929 [RFC3168] is carried out over the LoWPAN is out of scope, this draft 930 provides a way for the destination endpoint to echo an ECN indication 931 back to the source endpoint in an acknowledgment message as 932 represented in Figure 5 in Section 5.2. 934 It must be noted that congestion and collision are different topics. 935 In particular, when a mesh operates on a same channel over multiple 936 hops, then the forwarding of a fragment over a certain hop may 937 collide with the forwarding of a next fragment that is following over 938 a previous hop but in a same interference domain. This draft enables 939 an end-to-end flow control, but leaves it to the sender stack to pace 940 individual fragments within a transmit window, so that a given 941 fragment is sent only when the previous fragment has had a chance to 942 progress beyond the interference domain of this hop. In the case of 943 6TiSCH [I-D.ietf-6tisch-architecture], which operates over the 944 TimeSlotted Channel Hopping [RFC7554] (TSCH) mode of operation of 945 IEEE802.14.5, a fragment is forwarded over a different channel at a 946 different time and it makes full sense to transmit the next fragment 947 as soon as the previous fragment has had its chance to be forwarded 948 at the next hop. 950 From the standpoint of a source 6LoWPAN endpoint, an outstanding 951 fragment is a fragment that was sent but for which no explicit 952 acknowledgment was received yet. This means that the fragment might 953 be on the way, received but not yet acknowledged, or the 954 acknowledgment might be on the way back. It is also possible that 955 either the fragment or the acknowledgment was lost on the way. 957 From the sender standpoint, all outstanding fragments might still be 958 in the network and contribute to its congestion. There is an 959 assumption, though, that after a certain amount of time, a frame is 960 either received or lost, so it is not causing congestion anymore. 961 This amount of time can be estimated based on the round trip delay 962 between the 6LoWPAN endpoints. The method detailed in [RFC6298] is 963 recommended for that computation. 965 The reader is encouraged to read through "Congestion Control 966 Principles" [RFC2914]. Additionally [RFC7567] and [RFC5681] provide 967 deeper information on why this mechanism is needed and how TCP 968 handles Congestion Control. Basically, the goal here is to manage 969 the amount of fragments present in the network; this is achieved by 970 to reducing the number of outstanding fragments over a congested path 971 by throttling the sources. 973 Section 6 describes how the sender decides how many fragments are 974 (re)sent before an acknowledgment is required, and how the sender 975 adapts that number to the network conditions. 977 Author's Address 979 Pascal Thubert (editor) 980 Cisco Systems, Inc 981 Building D 982 45 Allee des Ormes - BP1200 983 MOUGINS - Sophia Antipolis 06254 984 FRANCE 986 Phone: +33 497 23 26 34 987 Email: pthubert@cisco.com