idnits 2.17.00 (12 Aug 2021) /tmp/idnits50559/draft-thubert-roll-forwarding-frags-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 -- The document date (March 29, 2012) is 3705 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) == Unused Reference: 'I-D.ietf-roll-rpl' is defined on line 656, but no explicit reference was found in the text == Unused Reference: 'I-D.mathis-frag-harmful' is defined on line 663, but no explicit reference was found in the text ** Obsolete normative reference: RFC 2988 (Obsoleted by RFC 6298) == Outdated reference: draft-ietf-roll-rpl has been published as RFC 6550 -- Obsolete informational reference (is this intentional?): RFC 5405 (Obsoleted by RFC 8085) Summary: 1 error (**), 0 flaws (~~), 4 warnings (==), 2 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 ROLL P. Thubert, Ed. 3 Internet-Draft J. Hui 4 Intended status: Standards Track Cisco 5 Expires: September 30, 2012 March 29, 2012 7 LLN Fragment Forwarding and Recovery 8 draft-thubert-roll-forwarding-frags-00 10 Abstract 12 In order to be routed, a fragmented packet must be reassembled at 13 every hop of a multihop link where lower layer fragmentation occurs. 14 Considering that the IPv6 minimum MTU is 1280 bytes and that an an 15 802.15.4 frame can have a payload limited to 74 bytes in the worst 16 case, a packet might end up fragmented into as many as 18 fragments 17 at the 6LoWPAN shim layer. If a single one of those fragments is 18 lost in transmission, all fragments must be resent, further 19 contributing to the congestion that might have caused the initial 20 packet loss. This draft introduces a simple protocol to forward and 21 recover individual fragments that might be lost over multiple hops 22 between 6LoWPAN endpoints. 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 http://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 September 30, 2012. 41 Copyright Notice 43 Copyright (c) 2012 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 (http://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 . . . . . . . . . . . . . . . . . . . . . . . . . 3 59 2. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 4 60 3. Rationale . . . . . . . . . . . . . . . . . . . . . . . . . . 4 61 4. Requirements . . . . . . . . . . . . . . . . . . . . . . . . . 6 62 5. Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 63 6. New Dispatch types and headers . . . . . . . . . . . . . . . . 7 64 6.1. Recoverable Fragment Dispatch type and Header . . . . . . 8 65 6.2. Fragment Acknowledgement Dispatch type and Header . . . . 8 66 7. Fragments Recovery . . . . . . . . . . . . . . . . . . . . . . 10 67 8. Forwarding Fragments . . . . . . . . . . . . . . . . . . . . . 12 68 8.1. Upon the first fragment . . . . . . . . . . . . . . . . . 12 69 8.2. Upon the next fragments . . . . . . . . . . . . . . . . . 13 70 8.3. Upon the fragment acknowledgements . . . . . . . . . . . . 14 71 9. Security Considerations . . . . . . . . . . . . . . . . . . . 14 72 10. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 14 73 11. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 14 74 12. References . . . . . . . . . . . . . . . . . . . . . . . . . . 15 75 12.1. Normative References . . . . . . . . . . . . . . . . . . . 15 76 12.2. Informative References . . . . . . . . . . . . . . . . . . 15 77 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 15 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 802.15.4 frame can 84 carry 74 bytes or more in all cases, fragmentation is usually not 85 required. However, and though this happens only occasionally, a 86 number of mission critical applications do require the capability to 87 transfer larger chunks of data, for instance to support a firmware 88 upgrades of the LLN nodes or an extraction of logs from LLN nodes. 89 In the former case, the large chunk of data is transferred to the LLN 90 node, whereas in the latter, the large chunk flows away from the LLN 91 node. In both cases, the size can be on the order of 10K bytes or 92 more and an end-to-end reliable transport is required. 94 Mechanisms such as TCP or application-layer segmentation will be used 95 to support end-to-end reliable transport. One option to support bulk 96 data transfer over a frame-size-constrained LLN is to set the Maximum 97 Segment Size to fit within the link maximum frame size. Doing so, 98 however, can add significant header overhead to each 802.15.4 frame. 99 This causes the end-to-end transport to be intimately aware of the 100 delivery properties of the underlaying LLN, which is a layer 101 violation. 103 An alternative mechanism combines the use of 6LoWPAN fragmentation in 104 addition to transport or application-layer segmentation. Increasing 105 the Maximum Segment Size reduces header overhead by the end-to-end 106 transport protocol. It also encourages the transport protocol to 107 reduce the number of outstanding datagrams, ideally to a single 108 datagram, thus reducing the need to support out-of-order delivery 109 common to LLNs. 111 [RFC4944] defines a datagram fragmentation mechanism for LLNs. 112 However, because [RFC4944] does not define a mechanism for recovering 113 fragments that are lost, datagram forwarding fails if even one 114 fragment is not delivered properly to the next IP hop. End-to-end 115 transport mechanisms will require retransmission of all fragments, 116 wasting resources in an already resource-constrained network. 118 Past experience with fragmentation has shown that missassociated or 119 lost fragments can lead to poor network behavior and, eventually, 120 trouble at application layer. The reader is encouraged to read 121 [RFC4963] and follow the references for more information. That 122 experience led to the definition of the Path MTU discovery [RFC1191] 123 protocol that limits fragmentation over the Internet. 125 For one-hop communications, a number of media propose a local 126 acknowledgement mechanism that is enough to protect the fragments. 128 In a multihop environment, an end-to-end fragment recovery mechanism 129 might be a good complement to a hop-by-hop MAC level recovery. This 130 draft introduces a simple protocol to recover individual fragments 131 between 6LoWPAN endpoints. Specifically in the case of UDP, valuable 132 additional information can be found in UDP Usage Guidelines for 133 Application Designers [RFC5405]. 135 2. Terminology 137 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 138 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 139 document are to be interpreted as described in [RFC2119]. 141 Readers are expected to be familiar with all the terms and concepts 142 that are discussed in "IPv6 over Low-Power Wireless Personal Area 143 Networks (6LoWPANs): Overview, Assumptions, Problem Statement, and 144 Goals" [RFC4919] and "Transmission of IPv6 Packets over IEEE 802.15.4 145 Networks" [RFC4944]. 147 ERP 149 Error Recovery Procedure. 151 6LoWPAN endpoints 153 The LLN nodes in charge of generating or expanding a 6LoWPAN 154 header from/to a full IPv6 packet. The 6LoWPAN endpoints are the 155 points where fragmentation and reassembly take place. 157 3. Rationale 159 There are a number of uses for large packets in Wireless Sensor 160 Networks. Such usages may not be the most typical or represent the 161 largest amount of traffic over the LLN; however, the associated 162 functionality can be critical enough to justify extra care for 163 ensuring effective transport of large packets across the LLN. 165 The list of those usages includes: 167 Towards the LLN node: 169 Packages of Commands: A number of commands or a full 170 configuration can by packaged as a single message to ensure 171 consistency and enable atomic execution or complete roll back. 172 Until such commands are fully received and interpreted, the 173 intended operation will not take effect. 175 Firmware update: For example, a new version of the LLN node 176 software is downloaded from a system manager over unicast or 177 multicast services. Such a reflashing operation typically 178 involves updating a large number of similar LLN nodes over a 179 relatively short period of time. 181 From the LLN node: 183 Waveform captures: A number of consecutive samples are measured 184 at a high rate for a short time and then transferred from a 185 sensor to a gateway or an edge server as a single large report. 187 Data logs: LLN nodes may generate large logs of sampled data for 188 later extraction. LLN nodes may also generate system logs to 189 assist in diagnosing problems on the node or network. 191 Large data packets: Rich data types might require more than one 192 fragment. 194 Uncontrolled firmware download or waveform upload can easily result 195 in a massive increase of the traffic and saturate the network. 197 When a fragment is lost in transmission, all fragments are resent, 198 further contributing to the congestion that caused the initial loss, 199 and potentially leading to congestion collapse. 201 This saturation may lead to excessive radio interference, or random 202 early discard (leaky bucket) in relaying nodes. Additional queuing 203 and memory congestion may result while waiting for a low power next 204 hop to emerge from its sleeping state. 206 To demonstrate the severity of the problem, consider a fairly 207 reliable 802.15.4 frame delivery rate of 99.9% over a single 802.15.4 208 hop. The expected delivery rate of a 5-fragment datagram would be 209 about 99.5% over a single 802.15.4 hop. However, the expected 210 delivery rate would drop to 95.1% over 10 hops, a reasonable network 211 diameter for LLN applications. The expected delivery rate for a 212 1280-byte datagram is 98.4% over a single hop and 85.2% over 10 hops. 214 Considering that [RFC4944] defines an MTU is 1280 bytes and that in 215 most incarnations (but 802.15.4G) a 802.15.4 frame can limit the MAC 216 payload to as few as 74 bytes, a packet might be fragmented into at 217 least 18 fragments at the 6LoWPAN shim layer. Taking into account 218 the worst-case header overhead for 6LoWPAN Fragmentation and Mesh 219 Addressing headers will increase the number of required fragments to 220 around 32. This level of fragmentation is much higher than that 221 traditionally experienced over the Internet with IPv4 fragments. At 222 the same time, the use of radios increases the probability of 223 transmission loss and Mesh-Under techniques compound that risk over 224 multiple hops. 226 4. Requirements 228 This paper proposes a method to recover individual fragments between 229 LLN endpoints. The method is designed to fit the following 230 requirements of a LLN (with or without a Mesh-Under routing 231 protocol): 233 Number of fragments 235 The recovery mechanism must support highly fragmented packets, 236 with a maximum of 32 fragments per packet. 238 Minimum acknowledgement overhead 240 Because the radio is half duplex, and because of silent time spent 241 in the various medium access mechanisms, an acknowledgment 242 consumes roughly as many resources as data fragment. 244 The recovery mechanism should be able to acknowledge multiple 245 fragments in a single message and not require an acknowledgement 246 at all if fragments are already protected at a lower layer. 248 Controlled latency 250 The recovery mechanism must succeed or give up within the time 251 boundary imposed by the recovery process of the Upper Layer 252 Protocols. 254 Support for out-of-order fragment delivery 256 A Mesh-Under load balancing mechanism such as the ISA100 Data Link 257 Layer can introduce out-of-sequence packets. 259 The recovery mechanism must account for packets that appear lost 260 but are actually only delayed over a different path. 262 Optional congestion control 264 The aggregation of multiple concurrent flows may lead to the 265 saturation of the radio network and congestion collapse. 267 The recovery mechanism should provide means for controlling the 268 number of fragments in transit over the LLN. 270 5. Overview 272 Considering that a multi-hop LLN can be a very sensitive environment 273 due to the limited queuing capabilities of a large population of its 274 nodes, this draft recommends a simple and conservative approach to 275 congestion control, based on TCP congestion avoidance. 277 From the standpoint of a source 6LoWPAN endpoint, an outstanding 278 fragment is a fragment that was sent but for which no explicit 279 acknowledgment was received yet. This means that the fragment might 280 be on the way, received but not yet acknowledged, or the 281 acknowledgment might be on the way back. It is also possible that 282 either the fragment or the acknowledgment was lost on the way. 284 Because a meshed LLN might deliver frames out of order, it is 285 virtually impossible to differentiate these situations. In other 286 words, from the sender standpoint, all outstanding fragments might 287 still be in the network and contribute to its congestion. There is 288 an assumption, though, that after a certain amount of time, a frame 289 is either received or lost, so it is not causing congestion anymore. 290 This amount of time can be estimated based on the round trip delay 291 between the 6LoWPAN endpoints. The method detailed in [RFC2988] is 292 recommended for that computation. 294 6. New Dispatch types and headers 296 This specification extends "Transmission of IPv6 Packets over IEEE 297 802.15.4 Networks" [RFC4944] with 4 new dispatch types, for 298 Recoverable Fragments (RFRAG) headers with or without Acknowledgment 299 Request, and for the Acknowledgment back. 301 Pattern Header Type 302 +------------+-----------------------------------------------+ 303 | 11 101000 | RFRAG - Recoverable Fragment | 304 | 11 101001 | RFRAG-AR - RFRAG with Ack Request | 305 | 11 10101x | RFRAG-ACK - RFRAG Acknowledgment | 306 +------------+-----------------------------------------------+ 308 Figure 1: Additional Dispatch Value Bit Patterns 310 In the following sections, the semantics of "datagram_tag," 311 "datagram_offset" and "datagram_size" and the reassembly process are 312 changed from [RFC4944] Section 5.3. "Fragmentation Type and Header." 313 The size and offset are expressed on the compressed packet as opposed 314 to the uncompressed form. 316 6.1. Recoverable Fragment Dispatch type and Header 318 1 2 3 319 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 1 320 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 321 |1 1 1 0 1 0 0 X|datagram_offset| datagram_tag | 322 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 323 |Sequence | datagram_size | 324 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 325 X set == Ack Requested 327 Figure 2: Recoverable Fragment Dispatch type and Header 329 X bit 331 When set, the sender requires an Acknowledgment from the receiver 333 Sequence 335 The sequence number of the fragment. Fragments are numbered 336 [0..N] where N is in [0..31]. 338 6.2. Fragment Acknowledgement Dispatch type and Header 340 The specification also defines an acknowledgement bitmap that is used 341 to carry selective acknowlegements for the received fragments. A 342 given offset in the bitmap maps one to one with a given sequence 343 number. 345 The bitmap is compressed as a variable length field formed by control 346 bits and acknowledgement bits. The leftmost bits of the compressed 347 form are control bits up to the first 0. The rest is ack bits 348 encoded right to left: 350 Pattern Size Ack 351 +--------------------------------------+----------+----------+ 352 | 0XXXXXXX | 1 octet | 1 -> 7 | 353 | 10XXXXXX XXXXXXXX | 2 octets | 1 -> 14 | 354 | 110XXXXX XXXXXXXX XXXXXXXX | 3 octets | 1 -> 21 | 355 | 1110XXXX XXXXXXXX XXXXXXXX XXXXXXXX | 4 octets | 1 -> 28 | 356 +------------+-----------------------------------------------+ 358 Figure 3: Compressed acknowledgement bitmap encoding 360 The highest sequence number to be acknowledged determines the pattern 361 to be used. The format can be extended for more fragments in the 362 future but this specification only requires the support of up to 4 363 octets encoding, which enables to acknowledge up to 28 fragments. 365 A 32 bits uncompressed bitmap is obtained by prepending zeroes to the 366 XXX in the pattern above. For instance: 368 0 1 2 3 4 5 6 7 369 +-+-+-+-+-+-+-+-+ 370 |0|1|1|0|1|1|1|1| is expanded as: 371 +-+-+-+-+-+-+-+-+ 372 1 2 3 373 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 1 374 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 375 |0|0|0|0|0|0|0|0|0|0|0|0|0|0|0|0|0|0|0|0|0|0|0|0|0|1|1|0|1|1|1|1| 376 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 378 Figure 4: Expanding 1 octet encoding 380 and 381 1 2 382 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 383 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 384 |1|1|0|1|1|1|1|0|1|1|1|1|1|1|1|1|1|1|1|1|1|0|0|1| is expanded as: 385 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 386 1 2 3 387 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 1 388 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 389 |0|0|0|0|0|0|0|0|0|0|0|1|1|1|1|0|1|1|1|1|1|1|1|1|1|1|1|1|1|0|0|1| 390 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 392 Figure 5: Expanding 3 octets encoding 394 whereas the 4 octets encoding is expanded by simply setting the first 395 3 bits to 0. The 32 bits uncompressed bitmap is written and read as 396 follows: 398 1 2 3 399 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 1 400 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 401 | Acknowledgment Bitmap | 402 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 403 ^ ^ 404 bitmap indicating whether: | | 405 Fragment with sequence 10 was received --+ | 406 Fragment with sequence 00 was received ----------------------+ 408 Figure 6: Expanded bitmap encoding 410 So in the example in Figure 5 it appears that all fragments from 411 sequence 0 to 20 were received but for sequence 1, 2 and 16 that were 412 either lost or are still in the network over a slower path. 414 The compressed form of the acknowledgement bitmap is carried in a 415 Fragment Acknowledgement as follows: 417 1 2 3 418 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 1 419 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 420 |1 1 1 0 1 0 1 Y| datagram_tag | 421 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 422 | Compressed Acknowledgment Bitmap (8 to 32 bits) 423 +-+-+-+-+-+-+-+-+-+ .... 425 Figure 7: Fragment Acknowledgement Dispatch type and Header 427 Y bit 429 Reserved for Explicit Congestion Notification (ECN) signalling 431 Compressed Acknowledgement Bitmap 433 An encoded form of an acknowledgement bitmap. 435 7. Fragments Recovery 437 The Recoverable Fragments header RFRAG and RFRAG-AR deprecate the 438 original fragment headers from [RFC4944] and replace them in the 439 fragmented packets. The Fragment Acknowledgement RFRAG-ACK is 440 introduced as a standalone header in message that is sent back to the 441 fragment source endpoint as known by its MAC address. This assumes 442 that the source MAC address in the fragment (is any) and datagram_tag 443 are enough information to send the Fragment Acknowledgement back to 444 the source fragmentation endpoint. 446 The 6LoWPAN endpoint that fragments the packets at 6LoWPAN level (the 447 sender) controls the Fragment Acknowledgements. If may do that at 448 any fragment to implement its own policy or perform congestion 449 control which is out of scope for this document. When the sender of 450 the fragment knows that an underlying mechanism protects the 451 Fragments already it MAY refrain from using the Acknowledgement 452 mechanism, and never set the Ack Requested bit. The 6LoWPAN endpoint 453 that recomposes the packets at 6LoWPAN level (the receiver) MUST 454 acknowledge the fragments it has received when asked to, and MAY 455 slightly defer that acknowledgement. 457 The sender transfers a controlled number of fragments and MAY flag 458 the last fragment of a series with an acknowledgment request. The 459 received MUST acknowledge a fragment with the acknowledgment request 460 bit set. If any fragment immediately preceding an acknowledgment 461 request is still missing, the receiver MAY intentionally delay its 462 acknowledgment to allow in-transit fragments to arrive. delaying the 463 acknowledgement might defeat the round trip delay computation so it 464 should be configurable and not enabled by default. 466 The receiver interacts with the sender using an Acknowledgment 467 message with a bitmap that indicates which fragments were actually 468 received. The bitmap is a 32bit SWORD, which accommodates up to 32 469 fragments and is sufficient for the 6LoWPAN MTU. For all n in 470 [0..31], bit n is set to 1 in the bitmap to indicate that fragment 471 with sequence n was received, otherwise the bit is set to 0. All 472 zeroes is a NULL bitmap that indicates that the fragmentation process 473 was cancelled by the receiver for that datagram. 475 The receiver MAY issue unsolicited acknowledgments. An unsolicited 476 acknowledgment enables the sender endpoint to resume sending if it 477 had reached its maximum number of outstanding fragments or indicate 478 that the receiver has cancelled the process of an individual 479 datagram. Note that acknowledgments might consume precious resources 480 so the use of unsolicited acknowledgments should be configurable and 481 not enabled by default. 483 The sender arms a retry timer to cover the fragment that carries the 484 Acknowledgment request. Upon time out, the sender assumes that all 485 the fragments on the way are received or lost. The process must have 486 completed within an acceptable time that is within the boundaries of 487 upper layer retries. The method detailed in [RFC2988] is recommended 488 for the computation of the retry timer. It is expected that the 489 upper layer retries obey the same or friendly rules in which case a 490 single round of fragment recovery should fit within the upper layer 491 recovery timers. 493 Fragments are sent in a round robin fashion: the sender sends all the 494 fragments for a first time before it retries any lost fragment; lost 495 fragments are retried in sequence, oldest first. This mechanism 496 enables the receiver to acknowledge fragments that were delayed in 497 the network before they are actually retried. 499 When the sender decides that a packet should be dropped and the 500 fragmentation process canceled, it sends a pseudo fragment with the 501 datagram_offset, sequence and datagram_size all set to zero, and no 502 data. Upon reception of this message, the receiver should clean up 503 all resources for the packet associated to the datagram_tag. If an 504 acknowledgement is requested, the receiver responds with a NULL 505 bitmap. 507 The receiver might need to cancel the process of a fragmented packet 508 for internal reasons, for instance if it is out of recomposition 509 buffers, or considers that this packet is already fully recomposed 510 and passed to the upper layer. In that case, the receiver SHOULD 511 indicate so to the sender with a NULL bitmap. Upon an 512 acknowledgement with a NULL bitmap, the sender MUST drop the 513 datagram. 515 8. Forwarding Fragments 517 This specification enables intermediate routers to forward fragments 518 with no intermediate reconstruction of the entire packet. Upon the 519 first fragment, the routers lay an label along the path that is 520 followed by that fragment (that is IP routed), and all further 521 fragments are label switched along that path. As a consequence, 522 alternate routes not possible for individual fragments. The datagram 523 tag is used to carry the label, that is swapped at each hop. 525 8.1. Upon the first fragment 527 In route over the L2 source changes at each hop. The label that is 528 formed adnd placed in the datagram tag is associated to the source 529 MAC and only valid (and unique) for that source MAC. Say the first 530 fragment has: 532 Source IPv6 address = IP_A (maybe hops away) 534 Destination IPv6 address = IP_B (maybe hops away) 536 Source MAC = MAC_prv (prv as previous) 538 Datagram_tag= DT_prv 540 The intermediate router that forwards individual fragments does the 541 following: 543 a route lookup to get Next hop IPv6 towards IP_B, which resolves 544 as IP_nxt (nxt as next) 546 a ND resolution to get the MAC address associated to IP_nxt, which 547 resolves as MAC_nxt 549 Since it is a first fragment of a packet from that source MAC address 550 MAC_prv for that tag DT_prv, the router: 552 cleans up any leftover resource associated to the tupple (MAC_prv, 553 DT_prv) 555 allocates a new label for that flow, DT_nxt, from a Least Recently 556 Used pool or some siumilar procedure. 558 allocates a Label swap structure indexed by (MAC_prv, DT_prv) that 559 contains (MAC_nxt, DT_nxt) 561 allocates a Label swap structure indexed by (MAC_nxt, DT_nxt) that 562 contains (MAC_prv, DT_prv) 564 swaps the MAC info to from self to MAC_nxt 566 Swaps the datagram_tag to DT_nxt 568 At this point the router is all set and can forward the packet to 569 nxt. 571 8.2. Upon the next fragments 573 Upon next fragments (that are not first fragment), the router expects 574 to have already Label swap structure indexed by (MAC_prv, DT_prv). 575 The router: 577 lookups up the Label swap entry for (MAC_prv, DT_prv), which 578 resolves as (MAC_nxt, DT_nxt) 580 swaps the MAC info to from self to MAC_nxt; 582 Swaps the datagram_tag to DT_nxt 584 At this point the router is all set and can forward the packet to 585 nxt. 587 if the Label swap entry for (MAC_src, DT_src) is not found, the 588 router builds an RFRAG-ACK to indicate the error. The acknowledgment 589 message has the following information: 591 MAC info set to from self to MAC_prv as found in the fragment 593 Swaps the datagram_tag set to DT_prv 595 Bitmap of all zeroes to indicate the error 597 At this point the router is all set and can send the RFRAG-ACK back 598 ot the previous router. 600 8.3. Upon the fragment acknowledgements 602 Upon fragment acknowledgements next fragments (that are not first 603 fragment), the router expects to have already Label swap structure 604 indexed by (MAC_nxt, DT_nxt). The router: 606 lookups up the Label swap entry for (MAC_nxt, DT_nxt), which 607 resolves as (MAC_prv, DT_prv) 609 swaps the MAC info to from self to MAC_prv; 611 Swaps the datagram_tag to DT_prv 613 At this point the router is all set and can forward the RFRAG-ACK to 614 prv. 616 if the Label swap entry for (MAC_nxt, DT_nxt) is not found, it simply 617 drops the packet. 619 if the RFRAG-ACK indicates either an error or that the fragment was 620 fully receive, the router schedules the Label swap entries for 621 recycling. If the RFRAG-ACK is lost on the way back, the source may 622 retry the last fragments, which will result as an error RFRAG-ACK 623 from the first router on the way that has already cleaned up. 625 9. Security Considerations 627 The process of recovering fragments does not appear to create any 628 opening for new threat compared to "Transmission of IPv6 Packets over 629 IEEE 802.15.4 Networks" [RFC4944]. 631 10. IANA Considerations 633 Need extensions for formats defined in "Transmission of IPv6 Packets 634 over IEEE 802.15.4 Networks" [RFC4944]. 636 11. Acknowledgments 638 The author wishes to thank Jay Werb, Christos Polyzois, Soumitri 639 Kolavennu and Harry Courtice for their contribution and review. 641 12. References 642 12.1. Normative References 644 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 645 Requirement Levels", BCP 14, RFC 2119, March 1997. 647 [RFC2988] Paxson, V. and M. Allman, "Computing TCP's Retransmission 648 Timer", RFC 2988, November 2000. 650 [RFC4944] Montenegro, G., Kushalnagar, N., Hui, J., and D. Culler, 651 "Transmission of IPv6 Packets over IEEE 802.15.4 652 Networks", RFC 4944, September 2007. 654 12.2. Informative References 656 [I-D.ietf-roll-rpl] 657 Brandt, A., Vasseur, J., Hui, J., Pister, K., Thubert, P., 658 Levis, P., Struik, R., Kelsey, R., Clausen, T., and T. 659 Winter, "RPL: IPv6 Routing Protocol for Low power and 660 Lossy Networks", draft-ietf-roll-rpl-19 (work in 661 progress), March 2011. 663 [I-D.mathis-frag-harmful] 664 Mathis, M., "Fragmentation Considered Very Harmful", 665 draft-mathis-frag-harmful-00 (work in progress), 666 July 2004. 668 [RFC1191] Mogul, J. and S. Deering, "Path MTU discovery", RFC 1191, 669 November 1990. 671 [RFC4919] Kushalnagar, N., Montenegro, G., and C. Schumacher, "IPv6 672 over Low-Power Wireless Personal Area Networks (6LoWPANs): 673 Overview, Assumptions, Problem Statement, and Goals", 674 RFC 4919, August 2007. 676 [RFC4963] Heffner, J., Mathis, M., and B. Chandler, "IPv4 Reassembly 677 Errors at High Data Rates", RFC 4963, July 2007. 679 [RFC5405] Eggert, L. and G. Fairhurst, "Unicast UDP Usage Guidelines 680 for Application Designers", BCP 145, RFC 5405, 681 November 2008. 683 Authors' Addresses 685 Pascal Thubert (editor) 686 Cisco Systems 687 Village d'Entreprises Green Side 688 400, Avenue de Roumanille 689 Batiment T3 690 Biot - Sophia Antipolis 06410 691 FRANCE 693 Phone: +33 4 97 23 26 34 694 Email: pthubert@cisco.com 696 Jonathan W. Hui 697 Cisco Systems 698 560 McCarthy Blvd. 699 MILPITAS, California 95035 700 USA 702 Email: johui@cisco.com