idnits 2.17.1 draft-thubert-6lo-forwarding-fragments-08.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 : ---------------------------------------------------------------------------- -- The draft header indicates that this document updates RFC4944, but the abstract doesn't seem to mention this, which it should. 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 16, 2018) is 2292 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) -- Possible downref: Non-RFC (?) normative reference: ref. 'IEEE.802.15.4' == Outdated reference: A later version (-30) exists of draft-ietf-6tisch-architecture-13 Summary: 0 errors (**), 0 flaws (~~), 2 warnings (==), 4 comments (--).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) J. Hui 5 Intended status: Standards Track Nest Labs 6 Expires: July 20, 2018 January 16, 2018 8 LLN Fragment Forwarding and Recovery 9 draft-thubert-6lo-forwarding-fragments-08 11 Abstract 13 Considering that an LLN frame can have a MAC payload below 100 bytes, 14 an IPv6 packet might be fragmented into more than 10 fragments at the 15 6LoWPAN layer. In a 6LoWPAN mesh-under network, the fragments can be 16 forwarded individually across the mesh, whereas a route-over mesh 17 network, a fragmented 6LoWPAN packet must be reassembled at every 18 hop, which causes latency and congestion. This draft introduces a 19 simple protocol to forward individual fragments across a route-over 20 mesh network, and, regardless of the type of mesh, recover the loss 21 of individual fragments across the mesh and protect the network 22 against bloat with a minimal flow control. 24 Status of This Memo 26 This Internet-Draft is submitted in full conformance with the 27 provisions of BCP 78 and BCP 79. 29 Internet-Drafts are working documents of the Internet Engineering 30 Task Force (IETF). Note that other groups may also distribute 31 working documents as Internet-Drafts. The list of current Internet- 32 Drafts is at https://datatracker.ietf.org/drafts/current/. 34 Internet-Drafts are draft documents valid for a maximum of six months 35 and may be updated, replaced, or obsoleted by other documents at any 36 time. It is inappropriate to use Internet-Drafts as reference 37 material or to cite them other than as "work in progress." 39 This Internet-Draft will expire on July 20, 2018. 41 Copyright Notice 43 Copyright (c) 2018 IETF Trust and the persons identified as the 44 document authors. All rights reserved. 46 This document is subject to BCP 78 and the IETF Trust's Legal 47 Provisions Relating to IETF Documents 48 (https://trustee.ietf.org/license-info) in effect on the date of 49 publication of this document. Please review these documents 50 carefully, as they describe your rights and restrictions with respect 51 to this document. Code Components extracted from this document must 52 include Simplified BSD License text as described in Section 4.e of 53 the Trust Legal Provisions and are provided without warranty as 54 described in the Simplified BSD License. 56 Table of Contents 58 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2 59 2. Updating RFC 4944 . . . . . . . . . . . . . . . . . . . . . . 3 60 3. Terminology and Referenced Work . . . . . . . . . . . . . . . 4 61 4. New Dispatch types and headers . . . . . . . . . . . . . . . 5 62 4.1. Recoverable Fragment Dispatch type and Header . . . . . . 5 63 4.2. RFRAG Acknowledgment Dispatch type and Header . . . . . . 7 64 5. Fragments Recovery . . . . . . . . . . . . . . . . . . . . . 9 65 6. Forwarding Fragments . . . . . . . . . . . . . . . . . . . . 10 66 6.1. Upon the first fragment . . . . . . . . . . . . . . . . . 11 67 6.2. Upon the next fragments . . . . . . . . . . . . . . . . . 12 68 6.3. Upon the RFRAG Acknowledgments . . . . . . . . . . . . . 12 69 7. Security Considerations . . . . . . . . . . . . . . . . . . . 13 70 8. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 13 71 9. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 13 72 10. References . . . . . . . . . . . . . . . . . . . . . . . . . 13 73 10.1. Normative References . . . . . . . . . . . . . . . . . . 13 74 10.2. Informative References . . . . . . . . . . . . . . . . . 14 75 Appendix A. Rationale . . . . . . . . . . . . . . . . . . . . . 15 76 Appendix B. Requirements . . . . . . . . . . . . . . . . . . . . 17 77 Appendix C. Considerations On Flow Control . . . . . . . . . . . 18 78 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 19 80 1. Introduction 82 In most Low Power and Lossy Network (LLN) applications, the bulk of 83 the traffic consists of small chunks of data (in the order few bytes 84 to a few tens of bytes) at a time. Given that an IEEE Std. 802.15.4 85 [IEEE.802.15.4] frame can carry 74 bytes or more in all cases, 86 fragmentation is usually not required. However, and though this 87 happens only occasionally, a number of mission critical applications 88 do require the capability to transfer larger chunks of data, for 89 instance to support a firmware upgrades of the LLN nodes or an 90 extraction of logs from LLN nodes. In the former case, the large 91 chunk of data is transferred to the LLN node, whereas in the latter, 92 the large chunk flows away from the LLN node. In both cases, the 93 size can be on the order of 10Kbytes or more and an end-to-end 94 reliable transport is required. 96 "Transmission of IPv6 Packets over IEEE 802.15.4 Networks" [RFC4944] 97 defines the original 6LoWPAN datagram fragmentation mechanism for 98 LLNs. One critical issue with this original design is that routing 99 an IPv6 [RFC8200] packet across a route-over mesh requires to 100 reassemble the full packet at each hop, which may cause latency along 101 a path and an overall buffer bloat in the network. Those undesirable 102 effects can be alleviated by a hop-by-hop fragment forwarding 103 technique such as the one proposed in this specification, and 104 arguably this could be achieved without the need to define a new 105 protocol. However, adding that capability alone to the local 106 implementation of the original 6LoWPAN fragmentation would not 107 address the bulk of the issues raised against it, and may create new 108 issues like uncontrolled state in the network. 110 Another issue against RFC 4944 [RFC4944] is that it does not define a 111 mechanism to first discover the loss of a fragment along a multi-hop 112 path (e.g. having exhausted the link-layer retries at some hop on the 113 way), and then to recover that loss. With RFC 4944, the forwarding 114 of a whole datagram fails when one fragment is not delivered properly 115 to the destination 6LoWPAN endpoint. End-to-end transport or 116 application-level mechanisms may require a full retransmission of the 117 datagram, wasting resources in an already constrained network. 119 In that situation, the source 6LoWPAN endpoint will not be aware that 120 a loss occurred and will continue sending all fragments for a 121 datagram that is already doomed. The original support is missing 122 signaling to abort a multi-fragment transmission at any time and from 123 either end, and, if the capability to forward fragments is 124 implemented, clean up the related state in the network. It is also 125 lacking flow control capabilities to avoid participating to a 126 congestion that may in turn cause the loss of a fragment and trigger 127 the retransmission of the full datagram. 129 This specification proposes a method to forward fragments across a 130 multi-hop route-over mesh, and to recover individual fragments 131 between LLN endpoints. The method is designed to limit congestion 132 loss in the network and addresses the requirements that are detailed 133 in Appendix B. 135 2. Updating RFC 4944 137 This specification updates the fragmentation mechanism that is 138 specified in "Transmission of IPv6 Packets over IEEE 802.15.4 139 Networks" [RFC4944] for use in route-over LLNs by providing a model 140 where fragments can be forwarded end-to-end across a 6LoWPAN LLN, and 141 where fragments that are lost on the way can be recovered 142 individually. New dispatch types are defined in Section 4. 144 3. Terminology and Referenced Work 146 Past experience with fragmentation has shown that miss-associated or 147 lost fragments can lead to poor network behavior and, occasionally, 148 trouble at application layer. The reader is encouraged to read "IPv4 149 Reassembly Errors at High Data Rates" [RFC4963] and follow the 150 references for more information. 152 That experience led to the definition of "Path MTU discovery" 153 [RFC8201] (PMTUD) protocol that limits fragmentation over the 154 Internet. 156 Specifically in the case of UDP, valuable additional information can 157 be found in "UDP Usage Guidelines for Application Designers" 158 [RFC8085]. 160 Readers are expected to be familiar with all the terms and concepts 161 that are discussed in "IPv6 over Low-Power Wireless Personal Area 162 Networks (6LoWPANs): Overview, Assumptions, Problem Statement, and 163 Goals" [RFC4919] and "Transmission of IPv6 Packets over IEEE 802.15.4 164 Networks" [RFC4944]. 166 "The Benefits of Using Explicit Congestion Notification (ECN)" 167 [RFC8087] provides useful information on the potential benefits and 168 pitfalls of using ECN. 170 Quoting the "Multiprotocol Label Switching (MPLS) Architecture" 171 [RFC3031]: with MPLS, "packets are "labeled" before they are 172 forwarded. At subsequent hops, there is no further analysis of the 173 packet's network layer header. Rather, the label is used as an index 174 into a table which specifies the next hop, and a new label". The 175 MPLS technique is leveraged in the present specification to forward 176 fragments that actually do not have a network layer header, since the 177 fragmentation occurs below IP. 179 This specification uses the following terms: 181 6LoWPAN endpoints 183 The LLN nodes in charge of generating or expanding a 6LoWPAN 184 header from/to a full IPv6 packet. The 6LoWPAN endpoints are the 185 points where fragmentation and reassembly take place. 187 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 188 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 189 document are to be interpreted as described in [RFC2119]. 191 4. New Dispatch types and headers 193 This specification enables the 6LoWPAN fragmentation sublayer to 194 provide an MTU up to 2048 bytes to the upper layer, which can be the 195 6LoWPAN Header Compression sublayer that is defined in the 196 "Compression Format for IPv6 Datagrams" [RFC6282] specification. In 197 order to achieve this, this specification enables the fragmentation 198 and the reliable transmission of fragments over a multihop 6LoWPAN 199 mesh network. 201 This specification provides a technique that is derived from MPLS in 202 order to forward individual fragments across a 6LoWPAN route-over 203 mesh. The datagram_tag is used as a label; it is locally unique to 204 the node that is the source MAC address of the fragment, so together 205 the MAC address and the label can identify the fragment globally. A 206 node may build the datagram_tag in its own locally-significant way, 207 as long as the selected tag stays unique to the particular datagram 208 for the lifetime of that datagram. It results that the label does 209 not need to be globally unique but also that it must be swapped at 210 each hop as the source MAC address changes. 212 This specification extends RFC 4944 [RFC4944] with 4 new Dispatch 213 types, for Recoverable Fragment (RFRAG) headers with or without 214 Acknowledgment Request (RFRAG vs. RFRAG-ARQ), and for the RFRAG 215 Acknowledgment back, with or without ECN Echo (RFRAG-ACK vs. RFRAG- 216 ECHO). 218 (to be confirmed by IANA) The new 6LoWPAN Dispatch types use the 219 Value Bit Pattern of 11 1010xx from page 0 [RFC8025], as follows: 221 Pattern Header Type 222 +------------+------------------------------------------+ 223 | 11 101000 | RFRAG - Recoverable Fragment | 224 | 11 101001 | RFRAG-ARQ - RFRAG with Ack Request | 225 | 11 101010 | RFRAG-ACK - RFRAG Acknowledgment | 226 | 11 101011 | RFRAG-ECHO - RFRAG Ack with ECN Echo | 227 +------------+------------------------------------------+ 229 Figure 1: Additional Dispatch Value Bit Patterns 231 4.1. Recoverable Fragment Dispatch type and Header 233 In this specification, the size and offset of the fragments are 234 expressed on the compressed packet form as opposed to the 235 uncompressed - native - packet form. 237 The first fragment is recognized by a sequence of 0; it carries its 238 fragment_size and the datagram_size of the compressed packet, whereas 239 the other fragments carry their fragment_size and fragment_offset. 240 The last fragment for a datagram is recognized when its 241 fragment_offset and its fragment_size add up to the datagram_size. 243 Recoverable Fragments are sequenced and a bitmap is used in the RFRAG 244 Acknowledgment to indicate the received fragments by setting the 245 individual bits that correspond to their sequence. 247 1 2 3 248 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 1 249 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 250 |1 1 1 0 1 0 0 X|E|fragment_size| datagram_tag | 251 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 252 |sequence | fragment_offset | 253 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 254 X set == Ack Requested 256 Figure 2: RFRAG Dispatch type and Header 258 X: 1 bit; Ack Requested: when set, the sender requires an RFRAG 259 Acknowledgment from the receiver. 261 E: 1 bit; Explicit Congestion Notification; the "E" flag is reset by 262 the source of the fragment and set by intermediate routers to 263 signal that this fragment experienced congestion along its path. 265 Fragment_size: 7 bit unsigned integer; the size of this fragment in 266 a unit that depends on the MAC layer technology. For IEEE Std. 267 802.15.4, the unit is octet, and the maximum fragment size, which 268 is constrained by the maximum frame size of 128 octet minus the 269 overheads of the MAC and Fragment Headers, is not limited by this 270 encoding. 272 Sequence: 5 bit unsigned integer; the sequence number of the 273 fragment. Fragments are sequence numbered [0..N] where N is in 274 [0..31]. As long as the overheads enable a fragment size of 64 275 octets or more, this enables to fragment a packet of 2047 octets. 277 Fragment_offset: 11 bit unsigned integer; 279 * When set to a non-0 value, the semantics of the Fragment_offset 280 depends on the value of the Sequence. 282 + When the Sequence is not 0, this field indicates the offset 283 of the fragment in the compressed form. The fragment should 284 be forwarded based on an existing abel Switched Path (LSP) 285 as described in Section 6.2, or silently dropped if none is 286 found. 288 + When the Sequence is 0, denoting the first fragment of a 289 datagram, this field is overloaded to indicate the 290 total_size of the compressed packet, to help the receiver 291 allocate an adapted buffer for the reception and reassembly 292 operations. This format limits the maximum MTU on a 6LoWPAN 293 link to 2047 bytes, but 1280 bytes is the recommended value 294 to avoid issues with IPV6 Path MTU Discovery [RFC8201]. The 295 fragment should be routed based on the destination IPv6 296 address, and an LSP state should be installed as described 297 in Section 6.1. 299 * When set to 0, this field indicates an abort condition and all 300 state regarding the datagram should be cleaned up once the 301 processing of the fragment is complete; the processing of the 302 fragment depends on whether there is an LSP already established 303 for this datagram, and the next hop is still reachable: 305 + if an LSP already exists and is not broken, the fragment is 306 to be forwarded along that LSP as described in Section 6.2, 307 but regardless of the value of the Sequence field; 309 + else, if the Sequence is 0, then the fragment is to be 310 routed as described in Section 6.1 but no state is conserved 311 afterwards. 313 If the fragment cannot be forwarded or routed, then it is 314 silently dropped. 316 4.2. RFRAG Acknowledgment Dispatch type and Header 318 This specification also defines a 4-octet RFRAG Acknowledgment bitmap 319 that is used by the reassembling end point to confirm selectively the 320 reception of individual fragments. A given offset in the bitmap maps 321 one to one with a given sequence number. 323 The offset of the bit in the bitmap indicates which fragment is 324 acknowledged as follows: 326 1 2 3 327 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 1 328 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 329 | RFRAG Acknowledgment Bitmap | 330 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 331 ^ ^ 332 | | bitmap indicating whether: 333 | +--- Fragment with sequence 10 was received 334 +----------------------- Fragment with sequence 00 was received 336 Figure 3: RFRAG Acknowledgment bitmap encoding 338 Figure 4 shows an example Acknowledgment bitmap which indicates that 339 all fragments from sequence 0 to 20 were received, except for 340 fragments 1, 2 and 16 that were either lost or are still in the 341 network over a slower path. 343 1 2 3 344 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 1 345 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 346 |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| 347 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 349 Figure 4: Expanding 3 octets encoding 351 The RFRAG Acknowledgment Bitmap is included in a RFRAG Acknowledgment 352 header, as follows: 354 1 2 3 355 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 1 356 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 357 |1 1 1 0 1 0 1 Y| datagram_tag | 358 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 359 | RFRAG Acknowledgment Bitmap (32 bits) | 360 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 362 Figure 5: RFRAG Acknowledgment Dispatch type and Header 364 Y: 1 bit; Explicit Congestion Notification Echo 366 When set, the sender indicates that at least one of the 367 acknowledged fragments was received with an Explicit Congestion 368 Notification, indicating that the path followed by the fragments 369 is subject to congestion. 371 RFRAG Acknowledgment Bitmap 372 An RFRAG Acknowledgment Bitmap, whereby setting the bit at offset 373 x indicates that fragment x was received, as shown in Figure 3. 374 All 0's is a NULL bitmap that indicates that the fragmentation 375 process is aborted. All 1's is a FULL bitmap that indicates that 376 the fragmentation process is complete, all fragments were received 377 at the reassembly end point. 379 5. Fragments Recovery 381 The Recoverable Fragment headers RFRAG and RFRAG-ARQ are used to 382 transport a fragment and optionally request an RFRAG Acknowledgment 383 that will confirm the good reception of a one or more fragments. An 384 RFRAG Acknowledgment can optionally carry an ECN indication; it is 385 carried as a standalone header in a message that is sent back to the 386 6LoWPAN endpoint that was the source of the fragments, as known by 387 its MAC address. The process ensures that at every hop, the source 388 MAC address and the datagram_tag in the received fragment are enough 389 information to send the RFRAG Acknowledgment back towards the source 390 6LoWPAN endpoint by reversing the MPLS operation. 392 The 6LoWPAN endpoint that fragments the packets at 6LoWPAN level (the 393 sender) also controls when the reassembling end point sends the RFRAG 394 Acknowledgments by setting the Ack Requested flag in the RFRAG 395 packets. It may set the Ack Requested flag on any fragment to 396 perform congestion control by limiting the number of outstanding 397 fragments, which are the fragments that have been sent but for which 398 reception or loss was not positively confirmed by the reassembling 399 endpoint. When the sender of the fragment knows that an underlying 400 link-layer mechanism protects the Fragments, it may refrain from 401 using the RFRAG Acknowledgment mechanism, and never set the Ack 402 Requested bit. When it receives a fragment with the ACK Request flag 403 set, the 6LoWPAN endpoint that reassembles the packets at 6LoWPAN 404 level (the receiver) sends back an RFRAG Acknowledgment to confirm 405 reception of all the fragments it has received so far. 407 The sender transfers a controlled number of fragments and MAY flag 408 the last fragment of a series with an RFRAG Acknowledgment Request. 409 The received MUST acknowledge a fragment with the acknowledgment 410 request bit set. If any fragment immediately preceding an 411 acknowledgment request is still missing, the receiver MAY 412 intentionally delay its acknowledgment to allow in-transit fragments 413 to arrive. Delaying the acknowledgment might defeat the round trip 414 delay computation so it should be configurable and not enabled by 415 default. 417 The receiver MAY issue unsolicited acknowledgments. An unsolicited 418 acknowledgment signals to the sender endpoint that it can resume 419 sending if it had reached its maximum number of outstanding 420 fragments. Another use is to inform that the reassembling endpoint 421 has cancelled the process of an individual datagram. Note that 422 acknowledgments might consume precious resources so the use of 423 unsolicited acknowledgments should be configurable and not enabled by 424 default. 426 An observation is that streamlining forwarding of fragments generally 427 reduces the latency over the LLN mesh, providing room for retries 428 within existing upper-layer reliability mechanisms. The sender 429 protects the transmission over the LLN mesh with a retry timer that 430 is computed according to the method detailed in [RFC6298]. It is 431 expected that the upper layer retries obey the recommendations in 432 "UDP Usage Guidelines" [RFC8085], in which case a single round of 433 fragment recovery should fit within the upper layer recovery timers. 435 Fragments are sent in a round robin fashion: the sender sends all the 436 fragments for a first time before it retries any lost fragment; lost 437 fragments are retried in sequence, oldest first. This mechanism 438 enables the receiver to acknowledge fragments that were delayed in 439 the network before they are actually retried. 441 When the sender decides that a packet should be dropped and the 442 fragmentation process canceled, it sends a pseudo fragment with the 443 fragment_offset, sequence and fragment_size all set to 0, and no 444 data. Upon reception of this message, the receiver should clean up 445 all resources for the packet associated to the datagram_tag. If an 446 acknowledgment is requested, the receiver responds with a NULL 447 bitmap. 449 The receiver might need to cancel the process of a fragmented packet 450 for internal reasons, for instance if it is out of reassembly 451 buffers, or considers that this packet is already fully reassembled 452 and passed to the upper layer. In that case, the receiver SHOULD 453 indicate so to the sender with a NULL bitmap. Upon an acknowledgment 454 with a NULL bitmap, the sender MUST abort the current fragmented 455 transmission of the datagram. 457 6. Forwarding Fragments 459 It is assumed that the first Fragment is large enough to carry the 460 IPv6 header and make routing decisions. If that is not so, then this 461 specification MUST NOT be used. 463 This specification enables intermediate routers to forward fragments 464 with no intermediate reconstruction of the entire packet. The first 465 fragment carries the IP header and it is routed all the way from the 466 fragmenting end point to the reassembling end point. Upon the first 467 fragment, the routers along the path install a label-switched path 468 (LSP), and the following fragments are label-switched along that 469 path. As a consequence, alternate routes not possible for individual 470 fragments. The datagram_tag is used to carry the label, that is 471 swapped at each hop. All fragments follow the same path and 472 fragments are delivered in the order at which they are sent. 474 6.1. Upon the first fragment 476 In Route-Over mode, the source and destination MAC addressed in a 477 frame change at each hop. The label that is formed and placed in the 478 datagram_tag is associated to the source MAC and only valid (and 479 unique) for that source MAC. Say the first fragment has: 481 o Source IPv6 address = IP_A (maybe hops away) 483 o Destination IPv6 address = IP_B (maybe hops away) 485 o Source MAC = MAC_previous 487 o Datagram_tag= DT_previous 489 The intermediate router that forwards individual fragments performs 490 the following action: 492 1. a route lookup to get the Next hop IPv6 towards IP_B, which 493 resolves as IP_next. 495 2. a MAC address resolution to get the MAC address associated to 496 IP_next, which resolves as MAC_next 498 Since it is a first fragment of a packet from that source MAC address 499 MAC_previous for that tag DT_previous, the router: 501 1. cleans up any leftover resource associated to the tuple 502 (MAC_previous, DT_previous) 504 2. allocates a new label for that flow, DT_next, from a Least 505 Recently Used pool or some similar procedure. 507 3. allocates an abstract label-swap entry indexed by (MAC_previous, 508 DT_previous) that contains (MAC_next, DT_next) 510 4. allocates a reflective abstract label-swap structure indexed by 511 (MAC_next, DT_next) that contains (MAC_previous, DT_previous); 512 this enables the reverse MPLS switching operation that is used to 513 route the RFRAG-ACK. 515 5. change the source MAC address from MAC_prev to MAC_self 516 6. change the destination MAC address to from MAC_self to MAC_next 518 7. Swaps the datagram_tag to DT_next 520 At this point the router is all set and can forward the fragment to 521 next. 523 6.2. Upon the next fragments 525 Upon next fragments (that are not first fragment), the router expects 526 to have already installed a label-swap structure indexed by 527 (MAC_previous, DT_previous). The router: 529 1. looks up the label-swap entry for (MAC_previous, DT_previous), 530 which resolves as (MAC_next, DT_next) 532 2. swaps the MAC info to from self to MAC_next; 534 3. Swaps the datagram_tag to DT_next 536 if the label-swap entry for (MAC_previous, DT_previous) is not found, 537 the router builds an RFRAG-ACK to indicate the error. The resulting 538 message has the following information: 540 o MAC info set to from self to MAC_previous as found in the fragment 542 o The datagram_tag set to DT_previous 544 o Null bitmap to indicate the error 546 At this point the router is all set and can send the RFRAG-ACK back 547 ot the previous router. 549 6.3. Upon the RFRAG Acknowledgments 551 Upon an RFRAG Acknowledgment, the router expects to already have 552 label-swap structure indexed by (MAC_next, DT_next), which are 553 respectively the source MAC address of the received frame and the 554 received datagram_tag. DT_next should have been computed by this 555 router and this router should have assigned it to this particular 556 datagram. The router: 558 1. looks up the label-swap entry for (MAC_next, DT_next), which 559 resolves as (MAC_previous, DT_previous) 561 2. swaps the MAC info to from self to MAC_previous; 563 3. Swaps the datagram_tag to DT_previous 564 At this point the router is all set and can forward the RFRAG-ACK to 565 previous. 567 If the label-swap entry for (MAC_next, DT_next) is not found, it MUST 568 silently drop the packet. 570 If the RFRAG-ACK indicates either an error (NULL bitmap) or that the 571 fragment was entirely received (FULL bitmap), the router schedules 572 the label-swap entries for recycling. If the RFRAG-ACK is lost on 573 the way back, the source may retry the last fragment, which will 574 result as an error RFRAG-ACK from the first router on the way that 575 has already cleaned up. 577 7. Security Considerations 579 The process of recovering fragments does not appear to create any 580 opening for new threat compared to "Transmission of IPv6 Packets over 581 IEEE 802.15.4 Networks" [RFC4944]. 583 8. IANA Considerations 585 Need extensions for formats defined in "Transmission of IPv6 Packets 586 over IEEE 802.15.4 Networks" [RFC4944]. 588 9. Acknowledgments 590 The author wishes to thank Thomas Watteyne and Michael Richardson for 591 in-depth reviews and comments. Also many thanks to Jay Werb, 592 Christos Polyzois, Soumitri Kolavennu, Pat Kinney, Margaret 593 Wasserman, Richard Kelsey, Carsten Bormann and Harry Courtice for 594 their various contributions. 596 10. References 598 10.1. Normative References 600 [IEEE.802.15.4] 601 IEEE, "IEEE Standard for Low-Rate Wireless Networks", 602 IEEE Standard 802.15.4, DOI 10.1109/IEEESTD.2016.7460875, 603 . 605 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 606 Requirement Levels", BCP 14, RFC 2119, 607 DOI 10.17487/RFC2119, March 1997, 608 . 610 [RFC4944] Montenegro, G., Kushalnagar, N., Hui, J., and D. Culler, 611 "Transmission of IPv6 Packets over IEEE 802.15.4 612 Networks", RFC 4944, DOI 10.17487/RFC4944, September 2007, 613 . 615 [RFC6282] Hui, J., Ed. and P. Thubert, "Compression Format for IPv6 616 Datagrams over IEEE 802.15.4-Based Networks", RFC 6282, 617 DOI 10.17487/RFC6282, September 2011, 618 . 620 [RFC6298] Paxson, V., Allman, M., Chu, J., and M. Sargent, 621 "Computing TCP's Retransmission Timer", RFC 6298, 622 DOI 10.17487/RFC6298, June 2011, 623 . 625 [RFC8025] Thubert, P., Ed. and R. Cragie, "IPv6 over Low-Power 626 Wireless Personal Area Network (6LoWPAN) Paging Dispatch", 627 RFC 8025, DOI 10.17487/RFC8025, November 2016, 628 . 630 10.2. Informative References 632 [I-D.ietf-6tisch-architecture] 633 Thubert, P., "An Architecture for IPv6 over the TSCH mode 634 of IEEE 802.15.4", draft-ietf-6tisch-architecture-13 (work 635 in progress), November 2017. 637 [RFC2914] Floyd, S., "Congestion Control Principles", BCP 41, 638 RFC 2914, DOI 10.17487/RFC2914, September 2000, 639 . 641 [RFC3031] Rosen, E., Viswanathan, A., and R. Callon, "Multiprotocol 642 Label Switching Architecture", RFC 3031, 643 DOI 10.17487/RFC3031, January 2001, 644 . 646 [RFC3168] Ramakrishnan, K., Floyd, S., and D. Black, "The Addition 647 of Explicit Congestion Notification (ECN) to IP", 648 RFC 3168, DOI 10.17487/RFC3168, September 2001, 649 . 651 [RFC4919] Kushalnagar, N., Montenegro, G., and C. Schumacher, "IPv6 652 over Low-Power Wireless Personal Area Networks (6LoWPANs): 653 Overview, Assumptions, Problem Statement, and Goals", 654 RFC 4919, DOI 10.17487/RFC4919, August 2007, 655 . 657 [RFC4963] Heffner, J., Mathis, M., and B. Chandler, "IPv4 Reassembly 658 Errors at High Data Rates", RFC 4963, 659 DOI 10.17487/RFC4963, July 2007, 660 . 662 [RFC5681] Allman, M., Paxson, V., and E. Blanton, "TCP Congestion 663 Control", RFC 5681, DOI 10.17487/RFC5681, September 2009, 664 . 666 [RFC7554] Watteyne, T., Ed., Palattella, M., and L. Grieco, "Using 667 IEEE 802.15.4e Time-Slotted Channel Hopping (TSCH) in the 668 Internet of Things (IoT): Problem Statement", RFC 7554, 669 DOI 10.17487/RFC7554, May 2015, 670 . 672 [RFC7567] Baker, F., Ed. and G. Fairhurst, Ed., "IETF 673 Recommendations Regarding Active Queue Management", 674 BCP 197, RFC 7567, DOI 10.17487/RFC7567, July 2015, 675 . 677 [RFC8085] Eggert, L., Fairhurst, G., and G. Shepherd, "UDP Usage 678 Guidelines", BCP 145, RFC 8085, DOI 10.17487/RFC8085, 679 March 2017, . 681 [RFC8087] Fairhurst, G. and M. Welzl, "The Benefits of Using 682 Explicit Congestion Notification (ECN)", RFC 8087, 683 DOI 10.17487/RFC8087, March 2017, 684 . 686 [RFC8200] Deering, S. and R. Hinden, "Internet Protocol, Version 6 687 (IPv6) Specification", STD 86, RFC 8200, 688 DOI 10.17487/RFC8200, July 2017, 689 . 691 [RFC8201] McCann, J., Deering, S., Mogul, J., and R. Hinden, Ed., 692 "Path MTU Discovery for IP version 6", STD 87, RFC 8201, 693 DOI 10.17487/RFC8201, July 2017, 694 . 696 Appendix A. Rationale 698 There are a number of uses for large packets in Wireless Sensor 699 Networks. Such usages may not be the most typical or represent the 700 largest amount of traffic over the LLN; however, the associated 701 functionality can be critical enough to justify extra care for 702 ensuring effective transport of large packets across the LLN. 704 The list of those usages includes: 706 Towards the LLN node: 708 Firmware update: For example, a new version of the LLN node 709 software is downloaded from a system manager over unicast or 710 multicast services. Such a reflashing operation typically 711 involves updating a large number of similar LLN nodes over a 712 relatively short period of time. 714 Packages of Commands: A number of commands or a full 715 configuration can be packaged as a single message to ensure 716 consistency and enable atomic execution or complete roll back. 717 Until such commands are fully received and interpreted, the 718 intended operation will not take effect. 720 From the LLN node: 722 Waveform captures: A number of consecutive samples are measured 723 at a high rate for a short time and then transferred from a 724 sensor to a gateway or an edge server as a single large report. 726 Data logs: LLN nodes may generate large logs of sampled data for 727 later extraction. LLN nodes may also generate system logs to 728 assist in diagnosing problems on the node or network. 730 Large data packets: Rich data types might require more than one 731 fragment. 733 Uncontrolled firmware download or waveform upload can easily result 734 in a massive increase of the traffic and saturate the network. 736 When a fragment is lost in transmission, the lack of recovery in the 737 original fragmentation system of RFC 4944 implies that all fragments 738 are resent, further contributing to the congestion that caused the 739 initial loss, and potentially leading to congestion collapse. 741 This saturation may lead to excessive radio interference, or random 742 early discard (leaky bucket) in relaying nodes. Additional queuing 743 and memory congestion may result while waiting for a low power next 744 hop to emerge from its sleeping state. 746 Considering that RFC 4944 defines an MTU is 1280 bytes and that in 747 most incarnations (but 802.15.4g) a IEEE Std. 802.15.4 frame can 748 limit the MAC payload to as few as 74 bytes, a packet might be 749 fragmented into at least 18 fragments at the 6LoWPAN shim layer. 750 Taking into account the worst-case header overhead for 6LoWPAN 751 Fragmentation and Mesh Addressing headers will increase the number of 752 required fragments to around 32. This level of fragmentation is much 753 higher than that traditionally experienced over the Internet with 754 IPv4 fragments. At the same time, the use of radios increases the 755 probability of transmission loss and Mesh-Under techniques compound 756 that risk over multiple hops. 758 Mechanisms such as TCP or application-layer segmentation could be 759 used to support end-to-end reliable transport. One option to support 760 bulk data transfer over a frame-size-constrained LLN is to set the 761 Maximum Segment Size to fit within the link maximum frame size. 762 Doing so, however, can add significant header overhead to each 763 802.15.4 frame. In addition, deploying such a mechanism requires 764 that the end-to-end transport is aware of the delivery properties of 765 the underlying LLN, which is a layer violation, and difficult to 766 achieve from the far end of the IPv6 network. 768 Appendix B. Requirements 770 For one-hop communications, a number of Low Power and Lossy Network 771 (LLN) link-layers propose a local acknowledgment mechanism that is 772 enough to detect and recover the loss of fragments. In a multihop 773 environment, an end-to-end fragment recovery mechanism might be a 774 good complement to a hop-by-hop MAC level recovery. This draft 775 introduces a simple protocol to recover individual fragments between 776 6LoWPAN endpoints that may be multiple hops away. The method 777 addresses the following requirements of a LLN: 779 Number of fragments 781 The recovery mechanism must support highly fragmented packets, 782 with a maximum of 32 fragments per packet. 784 Minimum acknowledgment overhead 786 Because the radio is half duplex, and because of silent time spent 787 in the various medium access mechanisms, an acknowledgment 788 consumes roughly as many resources as data fragment. 790 The new end-to-end fragment recovery mechanism should be able to 791 acknowledge multiple fragments in a single message and not require 792 an acknowledgment at all if fragments are already protected at a 793 lower layer. 795 Controlled latency 797 The recovery mechanism must succeed or give up within the time 798 boundary imposed by the recovery process of the Upper Layer 799 Protocols. 801 Optional congestion control 802 The aggregation of multiple concurrent flows may lead to the 803 saturation of the radio network and congestion collapse. 805 The recovery mechanism should provide means for controlling the 806 number of fragments in transit over the LLN. 808 Appendix C. Considerations On Flow Control 810 Considering that a multi-hop LLN can be a very sensitive environment 811 due to the limited queuing capabilities of a large population of its 812 nodes, this draft recommends a simple and conservative approach to 813 congestion control, based on TCP congestion avoidance. 815 Congestion on the forward path is assumed in case of packet loss, and 816 packet loss is assumed upon time out. The draft allows to control 817 the number of outstanding fragments, that have been transmitted but 818 for which an acknowledgment was not received yet. It must be noted 819 that the number of outstanding fragments should not exceed the number 820 of hops in the network, but the way to figure the number of hops is 821 out of scope for this document. 823 Congestion on the forward path can also be indicated by an Explicit 824 Congestion Notification (ECN) mechanism. Though whether and how ECN 825 [RFC3168] is carried out over the LoWPAN is out of scope, this draft 826 provides a way for the destination endpoint to echo an ECN indication 827 back to the source endpoint in an acknowledgment message as 828 represented in Figure 5 in Section 4.2. 830 It must be noted that congestion and collision are different topics. 831 In particular, when a mesh operates on a same channel over multiple 832 hops, then the forwarding of a fragment over a certain hop may 833 collide with the forwarding of a next fragment that is following over 834 a previous hop but in a same interference domain. This draft enables 835 an end-to-end flow control, but leaves it to the sender stack to pace 836 individual fragments within a transmit window, so that a given 837 fragment is sent only when the previous fragment has had a chance to 838 progress beyond the interference domain of this hop. In the case of 839 6TiSCH [I-D.ietf-6tisch-architecture], which operates over the 840 TimeSlotted Channel Hopping [RFC7554] (TSCH) mode of operation of 841 IEEE802.14.5, a fragment is forwarded over a different channel at a 842 different time and it makes full sense to transmit the next fragment 843 as soon as the previous fragment has had its chance to be forwarded 844 at the next hop. 846 From the standpoint of a source 6LoWPAN endpoint, an outstanding 847 fragment is a fragment that was sent but for which no explicit 848 acknowledgment was received yet. This means that the fragment might 849 be on the way, received but not yet acknowledged, or the 850 acknowledgment might be on the way back. It is also possible that 851 either the fragment or the acknowledgment was lost on the way. 853 From the sender standpoint, all outstanding fragments might still be 854 in the network and contribute to its congestion. There is an 855 assumption, though, that after a certain amount of time, a frame is 856 either received or lost, so it is not causing congestion anymore. 857 This amount of time can be estimated based on the round trip delay 858 between the 6LoWPAN endpoints. The method detailed in [RFC6298] is 859 recommended for that computation. 861 The reader is encouraged to read through "Congestion Control 862 Principles" [RFC2914]. Additionally [RFC7567] and [RFC5681] provide 863 deeper information on why this mechanism is needed and how TCP 864 handles Congestion Control. Basically, the goal here is to manage 865 the amount of fragments present in the network; this is achieved by 866 to reducing the number of outstanding fragments over a congested path 867 by throttling the sources. 869 Section 5 describes how the sender decides how many fragments are 870 (re)sent before an acknowledgment is required, and how the sender 871 adapts that number to the network conditions. 873 Authors' Addresses 875 Pascal Thubert (editor) 876 Cisco Systems, Inc 877 Building D 878 45 Allee des Ormes - BP1200 879 MOUGINS - Sophia Antipolis 06254 880 FRANCE 882 Phone: +33 497 23 26 34 883 Email: pthubert@cisco.com 885 Jonathan W. Hui 886 Nest Labs 887 3400 Hillview Ave 888 Palo Alto, California 94304 889 USA 891 Email: jonhui@nestlabs.com