idnits 2.17.1 draft-ietf-6lo-fragment-recovery-00.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 (September 20, 2018) is 2044 days in the past. Is this intentional? Checking references for intended status: Proposed Standard ---------------------------------------------------------------------------- (See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) == Outdated reference: A later version (-02) exists of draft-watteyne-6lo-minimal-fragment-01 ** Downref: Normative reference to an Informational draft: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 'I-D.watteyne-6lo-minimal-fragment') == Outdated reference: A later version (-30) exists of draft-ietf-6tisch-architecture-14 Summary: 1 error (**), 0 flaws (~~), 3 warnings (==), 2 comments (--).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) September 20, 2018 5 Intended status: Standards Track 6 Expires: March 29, 2019 8 6LoWPAN Selective Fragment Recovery 9 draft-ietf-6lo-fragment-recovery-00 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 December 29, 2018. 34 Copyright Notice 36 Copyright (c) 2018 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 3. Updating draft-watteyne-6lo-minimal-fragment . . . . . . . . 4 54 4. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 4 55 4.1. BCP 14 . . . . . . . . . . . . . . . . . . . . . . . . . 4 56 4.2. References . . . . . . . . . . . . . . . . . . . . . . . 4 57 4.3. 6LoWPAN Acronyms . . . . . . . . . . . . . . . . . . . . 4 58 4.4. Referenced Work . . . . . . . . . . . . . . . . . . . . . 5 59 4.5. New Terms . . . . . . . . . . . . . . . . . . . . . . . . 6 60 5. New Dispatch types and headers . . . . . . . . . . . . . . . 6 61 5.1. Recoverable Fragment Dispatch type and Header . . . . . . 7 62 5.2. RFRAG Acknowledgment Dispatch type and Header . . . . . . 9 63 6. Fragments Recovery . . . . . . . . . . . . . . . . . . . . . 10 64 7. Forwarding Fragments . . . . . . . . . . . . . . . . . . . . 12 65 7.1. Upon the first fragment . . . . . . . . . . . . . . . . . 12 66 7.2. Upon the next fragments . . . . . . . . . . . . . . . . . 13 67 7.3. Upon the RFRAG Acknowledgments . . . . . . . . . . . . . 13 68 8. Security Considerations . . . . . . . . . . . . . . . . . . . 14 69 9. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 14 70 10. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 14 71 11. References . . . . . . . . . . . . . . . . . . . . . . . . . 14 72 11.1. Normative References . . . . . . . . . . . . . . . . . . 14 73 11.2. Informative References . . . . . . . . . . . . . . . . . 15 74 Appendix A. Rationale . . . . . . . . . . . . . . . . . . . . . 17 75 Appendix B. Requirements . . . . . . . . . . . . . . . . . . . . 18 76 Appendix C. Considerations On Flow Control . . . . . . . . . . . 19 77 Author's Address . . . . . . . . . . . . . . . . . . . . . . . . 20 79 1. Introduction 81 In most Low Power and Lossy Network (LLN) applications, the bulk of 82 the traffic consists of small chunks of data (in the order few bytes 83 to a few tens of bytes) at a time. Given that an IEEE Std. 802.15.4 84 [IEEE.802.15.4] frame can carry 74 bytes or more in all cases, 85 fragmentation is usually not required. However, and though this 86 happens only occasionally, a number of mission critical applications 87 do require the capability to transfer larger chunks of data, for 88 instance to support a firmware upgrades of the LLN nodes or an 89 extraction of logs from LLN nodes. In the former case, the large 90 chunk of data is transferred to the LLN node, whereas in the latter, 91 the large chunk flows away from the LLN node. In both cases, the 92 size can be on the order of 10Kbytes or more and an end-to-end 93 reliable transport is required. 95 "Transmission of IPv6 Packets over IEEE 802.15.4 Networks" [RFC4944] 96 defines the original 6LoWPAN datagram fragmentation mechanism for 97 LLNs. One critical issue with this original design is that routing 98 an IPv6 [RFC8200] packet across a route-over mesh requires to 99 reassemble the full packet at each hop, which may cause latency along 100 a path and an overall buffer bloat in the network. The "6TiSCH 101 Architecture" [I-D.ietf-6tisch-architecture] recommends to use a hop- 102 by-hop fragment forwarding technique to alleviate those undesirable 103 effects. "LLN Minimal Fragment Forwarding" 104 [I-D.watteyne-6lo-minimal-fragment] proposes such a technique, in a 105 fashion that is compatible with [RFC4944] without the need to define 106 a new protocol. However, adding that capability alone to the local 107 implementation of the original 6LoWPAN fragmentation would not 108 address the bulk of the issues raised against it, and may create new 109 issues like remnant state in the network. 111 Another issue against [RFC4944] is that it does not define a 112 mechanism to first discover the loss of a fragment along a multi-hop 113 path (e.g. having exhausted the link-layer retries at some hop on the 114 way), and then to recover that loss. With RFC 4944, the forwarding 115 of a whole datagram fails when one fragment is not delivered properly 116 to the destination 6LoWPAN endpoint. End-to-end transport or 117 application-level mechanisms may require a full retransmission of the 118 datagram, wasting resources in an already constrained network. 120 In that situation, the source 6LoWPAN endpoint will not be aware that 121 a loss occurred and will continue sending all fragments for a 122 datagram that is already doomed. The original support is missing 123 signaling to abort a multi-fragment transmission at any time and from 124 either end, and, if the capability to forward fragments is 125 implemented, clean up the related state in the network. It is also 126 lacking flow control capabilities to avoid participating to a 127 congestion that may in turn cause the loss of a fragment and trigger 128 the retransmission of the full datagram. 130 This specification proposes a method to forward fragments across a 131 multi-hop route-over mesh, and to recover individual fragments 132 between LLN endpoints. The method is designed to limit congestion 133 loss in the network and addresses the requirements that are detailed 134 in Appendix B. 136 2. Updating RFC 4944 138 This specification updates the fragmentation mechanism that is 139 specified in "Transmission of IPv6 Packets over IEEE 802.15.4 140 Networks" [RFC4944] for use in route-over LLNs by providing a model 141 where fragments can be forwarded end-to-end across a 6LoWPAN LLN, and 142 where fragments that are lost on the way can be recovered 143 individually. A new format for fragment is introduces and new 144 dispatch types are defined in Section 5. 146 3. Updating draft-watteyne-6lo-minimal-fragment 148 This specification updates the fragment forwarding mechanism 149 specified in "LLN Minimal Fragment Forwarding" 150 [I-D.watteyne-6lo-minimal-fragment] by providing additional events to 151 iprove the management of the Virtual Reassembly Buffer (VRB). 153 At the time of this writing, [I-D.watteyne-6lo-minimal-fragment] 154 allows for refragmenting in intermediate nodes, meaning that some 155 bytes from a given fragment may be left in the VRB to be added to the 156 next fragment. The reason for this to happen would be the need for 157 space in the outgoing fragment that was not needed in the incoming 158 fragment, for instance because the 6LoWPAN Header Compression is not 159 as efficient on the outgoing link, e.g., if the IID of the source 160 IPv6 address is elided on the first hop because it matches the MAC 161 address, but cannot be on the next hops. This specification does not 162 allow this since fragments are recovered end-to-end. This means that 163 the fragments that contain 6LoWPAN-compressed data must have enough 164 slack in them to enable a lesser efficient compression in the next 165 hops to still fit in one frame. 167 4. Terminology 169 4.1. BCP 14 171 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 172 "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and 173 "OPTIONAL" in this document are to be interpreted as described in BCP 174 14 [RFC2119][RFC8174] when, and only when, they appear in all 175 capitals, as shown here. 177 4.2. References 179 In this document, readers will encounter terms and concepts that are 180 discussed in the following documents: 182 o "Problem Statement and Requirements for IPv6 over Low-Power 183 Wireless Personal Area Network (6LoWPAN) Routing" [RFC6606] 185 4.3. 6LoWPAN Acronyms 187 This document uses the following acronyms: 189 6BBR: 6LoWPAN Backbone Router 191 6LBR: 6LoWPAN Border Router 193 6LN: 6LoWPAN Node 194 6LR: 6LoWPAN Router 196 LLN: Low-Power and Lossy Network 198 4.4. Referenced Work 200 Past experience with fragmentation has shown that miss-associated or 201 lost fragments can lead to poor network behavior and, occasionally, 202 trouble at application layer. The reader is encouraged to read "IPv4 203 Reassembly Errors at High Data Rates" [RFC4963] and follow the 204 references for more information. 206 That experience led to the definition of "Path MTU discovery" 207 [RFC8201] (PMTUD) protocol that limits fragmentation over the 208 Internet. 210 Specifically in the case of UDP, valuable additional information can 211 be found in "UDP Usage Guidelines for Application Designers" 212 [RFC8085]. 214 Readers are expected to be familiar with all the terms and concepts 215 that are discussed in "IPv6 over Low-Power Wireless Personal Area 216 Networks (6LoWPANs): Overview, Assumptions, Problem Statement, and 217 Goals" [RFC4919] and "Transmission of IPv6 Packets over IEEE 802.15.4 218 Networks" [RFC4944]. 220 "The Benefits of Using Explicit Congestion Notification (ECN)" 221 [RFC8087] provides useful information on the potential benefits and 222 pitfalls of using ECN. 224 Quoting the "Multiprotocol Label Switching (MPLS) Architecture" 225 [RFC3031]: with MPLS, "packets are "labeled" before they are 226 forwarded. At subsequent hops, there is no further analysis of the 227 packet's network layer header. Rather, the label is used as an index 228 into a table which specifies the next hop, and a new label". The 229 MPLS technique is leveraged in the present specification to forward 230 fragments that actually do not have a network layer header, since the 231 fragmentation occurs below IP. 233 "LLN Minimal Fragment Forwarding" [I-D.watteyne-6lo-minimal-fragment] 234 introduces the concept of a Virtual Reassembly Buffer (VRB) and an 235 associated technique to forward fragments as they come, using the 236 datagram_tag as a label in a fashion similar to MLPS. This 237 specification reuses that technique with slightly modified controls. 239 4.5. New Terms 241 This specification uses the following terms: 243 6LoWPAN endpoints 245 The LLN nodes in charge of generating or expanding a 6LoWPAN 246 header from/to a full IPv6 packet. The 6LoWPAN endpoints are the 247 points where fragmentation and reassembly take place. 249 5. New Dispatch types and headers 251 This specification enables the 6LoWPAN fragmentation sublayer to 252 provide an MTU up to 2048 bytes to the upper layer, which can be the 253 6LoWPAN Header Compression sublayer that is defined in the 254 "Compression Format for IPv6 Datagrams" [RFC6282] specification. In 255 order to achieve this, this specification enables the fragmentation 256 and the reliable transmission of fragments over a multihop 6LoWPAN 257 mesh network. 259 This specification provides a technique that is derived from MPLS in 260 order to forward individual fragments across a 6LoWPAN route-over 261 mesh. The datagram_tag is used as a label; it is locally unique to 262 the node that is the source MAC address of the fragment, so together 263 the MAC address and the label can identify the fragment globally. A 264 node may build the datagram_tag in its own locally-significant way, 265 as long as the selected tag stays unique to the particular datagram 266 for the lifetime of that datagram. It results that the label does 267 not need to be globally unique but also that it must be swapped at 268 each hop as the source MAC address changes. 270 This specification extends RFC 4944 [RFC4944] with 4 new Dispatch 271 types, for Recoverable Fragment (RFRAG) headers with or without 272 Acknowledgment Request (RFRAG vs. RFRAG-ARQ), and for the RFRAG 273 Acknowledgment back, with or without ECN Echo (RFRAG-ACK vs. RFRAG- 274 ECHO). 276 (to be confirmed by IANA) The new 6LoWPAN Dispatch types use the 277 Value Bit Pattern of 11 1010xx from page 0 [RFC8025], as follows: 279 Pattern Header Type 280 +------------+------------------------------------------+ 281 | 11 101000 | RFRAG - Recoverable Fragment | 282 | 11 101001 | RFRAG-ARQ - RFRAG with Ack Request | 283 | 11 101010 | RFRAG-ACK - RFRAG Acknowledgment | 284 | 11 101011 | RFRAG-ECHO - RFRAG Ack with ECN Echo | 285 +------------+------------------------------------------+ 287 Figure 1: Additional Dispatch Value Bit Patterns 289 In the following sections, the semantics of "datagram_tag" are 290 unchanged from [RFC4944] Section 5.3. "Fragmentation Type and 291 Header." and is compatible with the fragment forwarding operation 292 described in [I-D.watteyne-6lo-minimal-fragment]. 294 5.1. Recoverable Fragment Dispatch type and Header 296 In this specification, the size and offset of the fragments are 297 expressed on the compressed packet form as opposed to the 298 uncompressed - native - packet form. 300 The first fragment is recognized by a sequence of 0; it carries its 301 fragment_size and the datagram_size of the compressed packet, whereas 302 the other fragments carry their fragment_size and fragment_offset. 303 The last fragment for a datagram is recognized when its 304 fragment_offset and its fragment_size add up to the datagram_size. 306 Recoverable Fragments are sequenced and a bitmap is used in the RFRAG 307 Acknowledgment to indicate the received fragments by setting the 308 individual bits that correspond to their sequence. 310 1 2 3 311 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 1 312 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 313 |1 1 1 0 1 0 0 X|E|fragment_size| datagram_tag | 314 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 315 |sequence | fragment_offset | 316 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 317 X set == Ack Requested 319 Figure 2: RFRAG Dispatch type and Header 321 X: 1 bit; Ack Requested: when set, the sender requires an RFRAG 322 Acknowledgment from the receiver. 324 E: 1 bit; Explicit Congestion Notification; the "E" flag is reset by 325 the source of the fragment and set by intermediate routers to 326 signal that this fragment experienced congestion along its path. 328 Fragment_size: 7 bit unsigned integer; the size of this fragment in 329 a unit that depends on the MAC layer technology. For IEEE Std. 330 802.15.4, the unit is octet, and the maximum fragment size, which 331 is constrained by the maximum frame size of 128 octet minus the 332 overheads of the MAC and Fragment Headers, is not limited by this 333 encoding. 335 Sequence: 5 bit unsigned integer; the sequence number of the 336 fragment. Fragments are sequence numbered [0..N] where N is in 337 [0..31]. A sequence of 0 indicates the first fragment in a 338 datagram. For IEEE Std. 802.15.4, as long as the overheads enable 339 a fragment size of 64 octets or more, this enables to fragment a 340 packet of 2047 octets. 342 Fragment_offset: 11 bit unsigned integer; 344 * When set to a non-0 value, the semantics of the Fragment_offset 345 depends on the value of the Sequence. 347 + When the Sequence is not 0, this field indicates the offset 348 of the fragment in the compressed form. The fragment should 349 be forwarded based on an existing VRB as described in 350 Section 7.2, or silently dropped if none is found. 352 + For a first fragment (i.e. with a sequence of 0), this field 353 is overloaded to indicate the total_size of the compressed 354 packet, to help the receiver allocate an adapted buffer for 355 the reception and reassembly operations. This format limits 356 the maximum MTU on a 6LoWPAN link to 2047 bytes, but 1280 357 bytes is the recommended value to avoid issues with IPV6 358 Path MTU Discovery [RFC8201]. The fragment should be routed 359 based on the destination IPv6 address, and an VRB state 360 should be installed as described in Section 7.1. 362 * When set to 0, this field indicates an abort condition and all 363 state regarding the datagram should be cleaned up once the 364 processing of the fragment is complete; the processing of the 365 fragment depends on whether there is a VRB already established 366 for this datagram, and the next hop is still reachable: 368 + if a VRB already exists and is not broken, the fragment is 369 to be forwarded along the associated Label Switched Path 370 (LSP) as described in Section 7.2, but regardless of the 371 value of the Sequence field; 373 + else, if the Sequence is 0, then the fragment is to be 374 routed as described in Section 7.1 but no state is conserved 375 afterwards. 377 If the fragment cannot be forwarded or routed, then an abort 378 RFRAG-ACK is sent back to the source. 380 5.2. RFRAG Acknowledgment Dispatch type and Header 382 This specification also defines a 4-octet RFRAG Acknowledgment bitmap 383 that is used by the reassembling end point to confirm selectively the 384 reception of individual fragments. A given offset in the bitmap maps 385 one to one with a given sequence number. 387 The offset of the bit in the bitmap indicates which fragment is 388 acknowledged as follows: 390 1 2 3 391 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 1 392 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 393 | RFRAG Acknowledgment Bitmap | 394 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 395 ^ ^ 396 | | bitmap indicating whether: 397 | +--- Fragment with sequence 10 was received 398 +----------------------- Fragment with sequence 00 was received 400 Figure 3: RFRAG Acknowledgment bitmap encoding 402 Figure 4 shows an example Acknowledgment bitmap which indicates that 403 all fragments from sequence 0 to 20 were received, except for 404 fragments 1, 2 and 16 that were either lost or are still in the 405 network over a slower path. 407 1 2 3 408 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 1 409 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 410 |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| 411 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 413 Figure 4: Expanding 3 octets encoding 415 The RFRAG Acknowledgment Bitmap is included in a RFRAG Acknowledgment 416 header, as follows: 418 1 2 3 419 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 1 420 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 421 |1 1 1 0 1 0 1 Y| datagram_tag | 422 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 423 | RFRAG Acknowledgment Bitmap (32 bits) | 424 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 426 Figure 5: RFRAG Acknowledgment Dispatch type and Header 428 Y: 1 bit; Explicit Congestion Notification Echo 430 When set, the sender indicates that at least one of the 431 acknowledged fragments was received with an Explicit Congestion 432 Notification, indicating that the path followed by the fragments 433 is subject to congestion. 435 RFRAG Acknowledgment Bitmap 437 An RFRAG Acknowledgment Bitmap, whereby setting the bit at offset 438 x indicates that fragment x was received, as shown in Figure 3. 439 All 0's is a NULL bitmap that indicates that the fragmentation 440 process is aborted. All 1's is a FULL bitmap that indicates that 441 the fragmentation process is complete, all fragments were received 442 at the reassembly end point. 444 6. Fragments Recovery 446 The Recoverable Fragment headers RFRAG and RFRAG-ARQ are used to 447 transport a fragment and optionally request an RFRAG Acknowledgment 448 that will confirm the good reception of a one or more fragments. An 449 RFRAG Acknowledgment can optionally carry an ECN indication; it is 450 carried as a standalone header in a message that is sent back to the 451 6LoWPAN endpoint that was the source of the fragments, as known by 452 its MAC address. The process ensures that at every hop, the source 453 MAC address and the datagram_tag in the received fragment are enough 454 information to send the RFRAG Acknowledgment back towards the source 455 6LoWPAN endpoint by reversing the MPLS operation. 457 The 6LoWPAN endpoint that fragments the packets at 6LoWPAN level (the 458 sender) also controls when the reassembling end point sends the RFRAG 459 Acknowledgments by setting the Ack Requested flag in the RFRAG 460 packets. It may set the Ack Requested flag on any fragment to 461 perform congestion control by limiting the number of outstanding 462 fragments, which are the fragments that have been sent but for which 463 reception or loss was not positively confirmed by the reassembling 464 endpoint. When the sender of the fragment knows that an underlying 465 link-layer mechanism protects the Fragments, it may refrain from 466 using the RFRAG Acknowledgment mechanism, and never set the Ack 467 Requested bit. When it receives a fragment with the ACK Request flag 468 set, the 6LoWPAN endpoint that reassembles the packets at 6LoWPAN 469 level (the receiver) sends back an RFRAG Acknowledgment to confirm 470 reception of all the fragments it has received so far. 472 The sender transfers a controlled number of fragments and MAY flag 473 the last fragment of a series with an RFRAG Acknowledgment Request. 474 The received MUST acknowledge a fragment with the acknowledgment 475 request bit set. If any fragment immediately preceding an 476 acknowledgment request is still missing, the receiver MAY 477 intentionally delay its acknowledgment to allow in-transit fragments 478 to arrive. Delaying the acknowledgment might defeat the round trip 479 delay computation so it should be configurable and not enabled by 480 default. 482 The receiver MAY issue unsolicited acknowledgments. An unsolicited 483 acknowledgment signals to the sender endpoint that it can resume 484 sending if it had reached its maximum number of outstanding 485 fragments. Another use is to inform that the reassembling endpoint 486 has canceled the process of an individual datagram. Note that 487 acknowledgments might consume precious resources so the use of 488 unsolicited acknowledgments should be configurable and not enabled by 489 default. 491 An observation is that streamlining forwarding of fragments generally 492 reduces the latency over the LLN mesh, providing room for retries 493 within existing upper-layer reliability mechanisms. The sender 494 protects the transmission over the LLN mesh with a retry timer that 495 is computed according to the method detailed in [RFC6298]. It is 496 expected that the upper layer retries obey the recommendations in 497 "UDP Usage Guidelines" [RFC8085], in which case a single round of 498 fragment recovery should fit within the upper layer recovery timers. 500 Fragments are sent in a round robin fashion: the sender sends all the 501 fragments for a first time before it retries any lost fragment; lost 502 fragments are retried in sequence, oldest first. This mechanism 503 enables the receiver to acknowledge fragments that were delayed in 504 the network before they are actually retried. 506 When a single frequency is used by contiguous hops, the sender should 507 wait a reasonable amount of time between fragments so as to let a 508 fragment progress a few hops and avoid hidden terminal issues. This 509 precaution is not required on channel hopping technologies such as 510 Time Slotted CHannel Hopping (TSCH) [RFC6554] 512 When the sender decides that a packet should be dropped and the 513 fragmentation process canceled, it sends a pseudo fragment with the 514 fragment_offset, sequence and fragment_size all set to 0, and no 515 data. Upon reception of this message, the receiver should clean up 516 all resources for the packet associated to the datagram_tag. If an 517 acknowledgment is requested, the receiver responds with a NULL 518 bitmap. 520 The receiver might need to cancel the process of a fragmented packet 521 for internal reasons, for instance if it is out of reassembly 522 buffers, or considers that this packet is already fully reassembled 523 and passed to the upper layer. In that case, the receiver SHOULD 524 indicate so to the sender with a NULL bitmap in a RFRAG 525 Acknowledgment. Upon an acknowledgment with a NULL bitmap, the 526 sender endpoint MUST abort the transmission of the fragmented 527 datagram. 529 7. Forwarding Fragments 531 It is assumed that the first Fragment is large enough to carry the 532 IPv6 header and make routing decisions. If that is not so, then this 533 specification MUST NOT be used. 535 This specification extends the Virtual Reassembly Buffer (VRB) 536 technique to forward fragments with no intermediate reconstruction of 537 the entire packet. The first fragment carries the IP header and it 538 is routed all the way from the fragmenting end point to the 539 reassembling end point. Upon the first fragment, the routers along 540 the path install a label-switched path (LSP), and the following 541 fragments are label-switched along that path. As a consequence, 542 alternate routes not possible for individual fragments. The 543 datagram_tag is used to carry the label, that is swapped at each hop. 544 All fragments follow the same path and fragments are delivered in the 545 order at which they are sent. 547 7.1. Upon the first fragment 549 In Route-Over mode, the source and destination MAC addressed in a 550 frame change at each hop. The label that is formed and placed in the 551 datagram_tag is associated to the source MAC and only valid (and 552 unique) for that source MAC. Upon a first fragment (i.e. with a 553 sequence of zero), a VRB and the associated LSP state are created for 554 the tuple (source MAC address, datagram_tag) and the fragment is 555 forwarded along the IPv6 route that matches the destination IPv6 556 address in the IPv6 header as prescribed by 557 [I-D.watteyne-6lo-minimal-fragment]. The LSP state enables to match 558 the (previous MAC address, datagram_tag) in an incoming fragment to 559 the tuple (next MAC address, swapped datagram_tag) used in the 560 forwarded fragment and points at the VRB. In addition, the router 561 also forms a Reverse LSP state indexed by the MAC address of the next 562 hop and the swapped datagram_tag. This reverse LSP state also points 563 at the VRB and enables to match the (next MAC address, 564 swapped_datagram_tag) found in an RFRAG Acknowledgment to the tuple 565 (previous MAC address, datagram_tag) used when forwarding a Fragment 566 Acknowledgment (RFRAG-ACK) back to the sender endpoint. 568 7.2. Upon the next fragments 570 Upon a next fragment (i.e. with a non-zero sequence), the router 571 looks up a LSP indexed by the tuple (MAC address, datagram_tag) found 572 in the fragment. If it is found, the router forwards the fragment 573 using the associated VRB as prescribed by 574 [I-D.watteyne-6lo-minimal-fragment]. 576 if the VRB for the tuple is not found, the router builds an RFRAG-ACK 577 to abort the transmission of the packet. The resulting message has 578 the following information: 580 o The source and destination MAC addresses are swapped from those 581 found in the fragment 583 o The datagram_tag set to the datagram_tag found in the fragment 585 o A null bitmap is used to signal the abort condition 587 At this point the router is all set and can send the RFRAG-ACK back 588 to the previous router. The RFRAG-ACK should normally be forwarded 589 all the way to the source using the reverse LSP state in the VRBs in 590 the intermediate routers as described in the next section. 592 7.3. Upon the RFRAG Acknowledgments 594 Upon an RFRAG-ACK, the router looks up a Reverse LSP indexed by the 595 tuple (MAC address, datagram_tag), which are respectively the source 596 MAC address of the received frame and the received datagram_tag. If 597 it is found, the router forwards the fragment using the associated 598 VRB as prescribed by [I-D.watteyne-6lo-minimal-fragment], but using 599 the Reverse LSP so that the RFRAG-ACK flows back to the sender 600 endpoint. 602 If the Reverse LSP is not found, the router MUST silently drop the 603 RFRAG-ACK message. 605 Either way, if the RFRAG-ACK indicates either an error (NULL bitmap) 606 or that the fragment was entirely received (FULL bitmap), arms a 607 short timer, and upon timeout, the VRB and all associate state are 608 destroyed. During that time, fragments of that datagram may still be 609 received, e.g. if the RFRAG-ACK was lost on the way back and the 610 source retried the last fragment. In that case, the router sends an 611 abort RFRAG-ACK along the Reverse LSP to complete the clean up. 613 8. Security Considerations 615 The process of recovering fragments does not appear to create any 616 opening for new threat compared to "Transmission of IPv6 Packets over 617 IEEE 802.15.4 Networks" [RFC4944]. 619 9. IANA Considerations 621 Need extensions for formats defined in "Transmission of IPv6 Packets 622 over IEEE 802.15.4 Networks" [RFC4944]. 624 10. Acknowledgments 626 The author wishes to thank Thomas Watteyne and Michael Richardson for 627 in-depth reviews and comments. Also many thanks to Jonathan Hui, Jay 628 Werb, Christos Polyzois, Soumitri Kolavennu, Pat Kinney, Margaret 629 Wasserman, Richard Kelsey, Carsten Bormann and Harry Courtice for 630 their various contributions. 632 11. References 634 11.1. Normative References 636 [I-D.watteyne-6lo-minimal-fragment] 637 Watteyne, T., Bormann, C., and P. Thubert, "LLN Minimal 638 Fragment Forwarding", draft-watteyne-6lo-minimal- 639 fragment-01 (work in progress), March 2018. 641 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 642 Requirement Levels", BCP 14, RFC 2119, 643 DOI 10.17487/RFC2119, March 1997, 644 . 646 [RFC4944] Montenegro, G., Kushalnagar, N., Hui, J., and D. Culler, 647 "Transmission of IPv6 Packets over IEEE 802.15.4 648 Networks", RFC 4944, DOI 10.17487/RFC4944, September 2007, 649 . 651 [RFC6282] Hui, J., Ed. and P. Thubert, "Compression Format for IPv6 652 Datagrams over IEEE 802.15.4-Based Networks", RFC 6282, 653 DOI 10.17487/RFC6282, September 2011, 654 . 656 [RFC6554] Hui, J., Vasseur, JP., Culler, D., and V. Manral, "An IPv6 657 Routing Header for Source Routes with the Routing Protocol 658 for Low-Power and Lossy Networks (RPL)", RFC 6554, 659 DOI 10.17487/RFC6554, March 2012, 660 . 662 [RFC8025] Thubert, P., Ed. and R. Cragie, "IPv6 over Low-Power 663 Wireless Personal Area Network (6LoWPAN) Paging Dispatch", 664 RFC 8025, DOI 10.17487/RFC8025, November 2016, 665 . 667 [RFC8174] Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC 668 2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174, 669 May 2017, . 671 11.2. Informative References 673 [I-D.ietf-6tisch-architecture] 674 Thubert, P., "An Architecture for IPv6 over the TSCH mode 675 of IEEE 802.15.4", draft-ietf-6tisch-architecture-14 (work 676 in progress), April 2018. 678 [IEEE.802.15.4] 679 IEEE, "IEEE Standard for Low-Rate Wireless Networks", 680 IEEE Standard 802.15.4, DOI 10.1109/IEEE 681 P802.15.4-REVd/D01, 682 . 684 [RFC2914] Floyd, S., "Congestion Control Principles", BCP 41, 685 RFC 2914, DOI 10.17487/RFC2914, September 2000, 686 . 688 [RFC3031] Rosen, E., Viswanathan, A., and R. Callon, "Multiprotocol 689 Label Switching Architecture", RFC 3031, 690 DOI 10.17487/RFC3031, January 2001, 691 . 693 [RFC3168] Ramakrishnan, K., Floyd, S., and D. Black, "The Addition 694 of Explicit Congestion Notification (ECN) to IP", 695 RFC 3168, DOI 10.17487/RFC3168, September 2001, 696 . 698 [RFC4919] Kushalnagar, N., Montenegro, G., and C. Schumacher, "IPv6 699 over Low-Power Wireless Personal Area Networks (6LoWPANs): 700 Overview, Assumptions, Problem Statement, and Goals", 701 RFC 4919, DOI 10.17487/RFC4919, August 2007, 702 . 704 [RFC4963] Heffner, J., Mathis, M., and B. Chandler, "IPv4 Reassembly 705 Errors at High Data Rates", RFC 4963, 706 DOI 10.17487/RFC4963, July 2007, 707 . 709 [RFC5681] Allman, M., Paxson, V., and E. Blanton, "TCP Congestion 710 Control", RFC 5681, DOI 10.17487/RFC5681, September 2009, 711 . 713 [RFC6298] Paxson, V., Allman, M., Chu, J., and M. Sargent, 714 "Computing TCP's Retransmission Timer", RFC 6298, 715 DOI 10.17487/RFC6298, June 2011, 716 . 718 [RFC6606] Kim, E., Kaspar, D., Gomez, C., and C. Bormann, "Problem 719 Statement and Requirements for IPv6 over Low-Power 720 Wireless Personal Area Network (6LoWPAN) Routing", 721 RFC 6606, DOI 10.17487/RFC6606, May 2012, 722 . 724 [RFC7554] Watteyne, T., Ed., Palattella, M., and L. Grieco, "Using 725 IEEE 802.15.4e Time-Slotted Channel Hopping (TSCH) in the 726 Internet of Things (IoT): Problem Statement", RFC 7554, 727 DOI 10.17487/RFC7554, May 2015, 728 . 730 [RFC7567] Baker, F., Ed. and G. Fairhurst, Ed., "IETF 731 Recommendations Regarding Active Queue Management", 732 BCP 197, RFC 7567, DOI 10.17487/RFC7567, July 2015, 733 . 735 [RFC8085] Eggert, L., Fairhurst, G., and G. Shepherd, "UDP Usage 736 Guidelines", BCP 145, RFC 8085, DOI 10.17487/RFC8085, 737 March 2017, . 739 [RFC8087] Fairhurst, G. and M. Welzl, "The Benefits of Using 740 Explicit Congestion Notification (ECN)", RFC 8087, 741 DOI 10.17487/RFC8087, March 2017, 742 . 744 [RFC8200] Deering, S. and R. Hinden, "Internet Protocol, Version 6 745 (IPv6) Specification", STD 86, RFC 8200, 746 DOI 10.17487/RFC8200, July 2017, 747 . 749 [RFC8201] McCann, J., Deering, S., Mogul, J., and R. Hinden, Ed., 750 "Path MTU Discovery for IP version 6", STD 87, RFC 8201, 751 DOI 10.17487/RFC8201, July 2017, 752 . 754 Appendix A. Rationale 756 There are a number of uses for large packets in Wireless Sensor 757 Networks. Such usages may not be the most typical or represent the 758 largest amount of traffic over the LLN; however, the associated 759 functionality can be critical enough to justify extra care for 760 ensuring effective transport of large packets across the LLN. 762 The list of those usages includes: 764 Towards the LLN node: 766 Firmware update: For example, a new version of the LLN node 767 software is downloaded from a system manager over unicast or 768 multicast services. Such a reflashing operation typically 769 involves updating a large number of similar LLN nodes over a 770 relatively short period of time. 772 Packages of Commands: A number of commands or a full 773 configuration can be packaged as a single message to ensure 774 consistency and enable atomic execution or complete roll back. 775 Until such commands are fully received and interpreted, the 776 intended operation will not take effect. 778 From the LLN node: 780 Waveform captures: A number of consecutive samples are measured 781 at a high rate for a short time and then transferred from a 782 sensor to a gateway or an edge server as a single large report. 784 Data logs: LLN nodes may generate large logs of sampled data for 785 later extraction. LLN nodes may also generate system logs to 786 assist in diagnosing problems on the node or network. 788 Large data packets: Rich data types might require more than one 789 fragment. 791 Uncontrolled firmware download or waveform upload can easily result 792 in a massive increase of the traffic and saturate the network. 794 When a fragment is lost in transmission, the lack of recovery in the 795 original fragmentation system of RFC 4944 implies that all fragments 796 are resent, further contributing to the congestion that caused the 797 initial loss, and potentially leading to congestion collapse. 799 This saturation may lead to excessive radio interference, or random 800 early discard (leaky bucket) in relaying nodes. Additional queuing 801 and memory congestion may result while waiting for a low power next 802 hop to emerge from its sleeping state. 804 Considering that RFC 4944 defines an MTU is 1280 bytes and that in 805 most incarnations (but 802.15.4g) a IEEE Std. 802.15.4 frame can 806 limit the MAC payload to as few as 74 bytes, a packet might be 807 fragmented into at least 18 fragments at the 6LoWPAN shim layer. 808 Taking into account the worst-case header overhead for 6LoWPAN 809 Fragmentation and Mesh Addressing headers will increase the number of 810 required fragments to around 32. This level of fragmentation is much 811 higher than that traditionally experienced over the Internet with 812 IPv4 fragments. At the same time, the use of radios increases the 813 probability of transmission loss and Mesh-Under techniques compound 814 that risk over multiple hops. 816 Mechanisms such as TCP or application-layer segmentation could be 817 used to support end-to-end reliable transport. One option to support 818 bulk data transfer over a frame-size-constrained LLN is to set the 819 Maximum Segment Size to fit within the link maximum frame size. 820 Doing so, however, can add significant header overhead to each 821 802.15.4 frame. In addition, deploying such a mechanism requires 822 that the end-to-end transport is aware of the delivery properties of 823 the underlying LLN, which is a layer violation, and difficult to 824 achieve from the far end of the IPv6 network. 826 Appendix B. Requirements 828 For one-hop communications, a number of Low Power and Lossy Network 829 (LLN) link-layers propose a local acknowledgment mechanism that is 830 enough to detect and recover the loss of fragments. In a multihop 831 environment, an end-to-end fragment recovery mechanism might be a 832 good complement to a hop-by-hop MAC level recovery. This draft 833 introduces a simple protocol to recover individual fragments between 834 6LoWPAN endpoints that may be multiple hops away. The method 835 addresses the following requirements of a LLN: 837 Number of fragments 839 The recovery mechanism must support highly fragmented packets, 840 with a maximum of 32 fragments per packet. 842 Minimum acknowledgment overhead 843 Because the radio is half duplex, and because of silent time spent 844 in the various medium access mechanisms, an acknowledgment 845 consumes roughly as many resources as data fragment. 847 The new end-to-end fragment recovery mechanism should be able to 848 acknowledge multiple fragments in a single message and not require 849 an acknowledgment at all if fragments are already protected at a 850 lower layer. 852 Controlled latency 854 The recovery mechanism must succeed or give up within the time 855 boundary imposed by the recovery process of the Upper Layer 856 Protocols. 858 Optional congestion control 860 The aggregation of multiple concurrent flows may lead to the 861 saturation of the radio network and congestion collapse. 863 The recovery mechanism should provide means for controlling the 864 number of fragments in transit over the LLN. 866 Appendix C. Considerations On Flow Control 868 Considering that a multi-hop LLN can be a very sensitive environment 869 due to the limited queuing capabilities of a large population of its 870 nodes, this draft recommends a simple and conservative approach to 871 congestion control, based on TCP congestion avoidance. 873 Congestion on the forward path is assumed in case of packet loss, and 874 packet loss is assumed upon time out. The draft allows to control 875 the number of outstanding fragments, that have been transmitted but 876 for which an acknowledgment was not received yet. It must be noted 877 that the number of outstanding fragments should not exceed the number 878 of hops in the network, but the way to figure the number of hops is 879 out of scope for this document. 881 Congestion on the forward path can also be indicated by an Explicit 882 Congestion Notification (ECN) mechanism. Though whether and how ECN 883 [RFC3168] is carried out over the LoWPAN is out of scope, this draft 884 provides a way for the destination endpoint to echo an ECN indication 885 back to the source endpoint in an acknowledgment message as 886 represented in Figure 5 in Section 5.2. 888 It must be noted that congestion and collision are different topics. 889 In particular, when a mesh operates on a same channel over multiple 890 hops, then the forwarding of a fragment over a certain hop may 891 collide with the forwarding of a next fragment that is following over 892 a previous hop but in a same interference domain. This draft enables 893 an end-to-end flow control, but leaves it to the sender stack to pace 894 individual fragments within a transmit window, so that a given 895 fragment is sent only when the previous fragment has had a chance to 896 progress beyond the interference domain of this hop. In the case of 897 6TiSCH [I-D.ietf-6tisch-architecture], which operates over the 898 TimeSlotted Channel Hopping [RFC7554] (TSCH) mode of operation of 899 IEEE802.14.5, a fragment is forwarded over a different channel at a 900 different time and it makes full sense to transmit the next fragment 901 as soon as the previous fragment has had its chance to be forwarded 902 at the next hop. 904 From the standpoint of a source 6LoWPAN endpoint, an outstanding 905 fragment is a fragment that was sent but for which no explicit 906 acknowledgment was received yet. This means that the fragment might 907 be on the way, received but not yet acknowledged, or the 908 acknowledgment might be on the way back. It is also possible that 909 either the fragment or the acknowledgment was lost on the way. 911 From the sender standpoint, all outstanding fragments might still be 912 in the network and contribute to its congestion. There is an 913 assumption, though, that after a certain amount of time, a frame is 914 either received or lost, so it is not causing congestion anymore. 915 This amount of time can be estimated based on the round trip delay 916 between the 6LoWPAN endpoints. The method detailed in [RFC6298] is 917 recommended for that computation. 919 The reader is encouraged to read through "Congestion Control 920 Principles" [RFC2914]. Additionally [RFC7567] and [RFC5681] provide 921 deeper information on why this mechanism is needed and how TCP 922 handles Congestion Control. Basically, the goal here is to manage 923 the amount of fragments present in the network; this is achieved by 924 to reducing the number of outstanding fragments over a congested path 925 by throttling the sources. 927 Section 6 describes how the sender decides how many fragments are 928 (re)sent before an acknowledgment is required, and how the sender 929 adapts that number to the network conditions. 931 Author's Address 932 Pascal Thubert (editor) 933 Cisco Systems, Inc 934 Building D 935 45 Allee des Ormes - BP1200 936 MOUGINS - Sophia Antipolis 06254 937 FRANCE 939 Phone: +33 497 23 26 34 940 Email: pthubert@cisco.com