idnits 2.17.1 draft-thubert-roll-forwarding-frags-01.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- No issues found here. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (February 25, 2013) is 4050 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.mathis-frag-harmful' is defined on line 644, but no explicit reference was found in the text == Unused Reference: 'RFC6550' is defined on line 681, but no explicit reference was found in the text ** Obsolete normative reference: RFC 2988 (Obsoleted by RFC 6298) -- Obsolete informational reference (is this intentional?): RFC 2309 (Obsoleted by RFC 7567) -- Obsolete informational reference (is this intentional?): RFC 2581 (Obsoleted by RFC 5681) -- Obsolete informational reference (is this intentional?): RFC 5405 (Obsoleted by RFC 8085) Summary: 1 error (**), 0 flaws (~~), 3 warnings (==), 4 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: August 29, 2013 February 25, 2013 7 LLN Fragment Forwarding and Recovery 8 draft-thubert-roll-forwarding-frags-01 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 August 29, 2013. 41 Copyright Notice 43 Copyright (c) 2013 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 . . . . . . . . . . . . . . . . 8 64 6.1. Recoverable Fragment Dispatch type and Header . . . . . . 8 65 6.2. Fragment Acknowledgement Dispatch type and Header . . . . 9 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 . . . . . . . . . . . . 13 71 9. Security Considerations . . . . . . . . . . . . . . . . . . . 14 72 10. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 14 73 11. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 14 74 12. References . . . . . . . . . . . . . . . . . . . . . . . . . . 14 75 12.1. Normative References . . . . . . . . . . . . . . . . . . . 14 76 12.2. Informative References . . . . . . . . . . . . . . . . . . 15 77 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 16 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 Congestion on the forward path is assumed in case of packet loss, and 278 packet loss is assumed upon time out. 280 Congestion on the forward path can also be indicated by an Explicit 281 Congestion Notification (ECN) mechanism. Though whether and how ECN 282 [RFC3168] is carried out over the LoWPAN is out of scope, this draft 283 provides a way for the destination endpoint to echo an ECN indication 284 back to the source endpoint in an acknowledgment message as 285 represented in Figure 5 in Section 6.2. 287 From the standpoint of a source 6LoWPAN endpoint, an outstanding 288 fragment is a fragment that was sent but for which no explicit 289 acknowledgment was received yet. This means that the fragment might 290 be on the way, received but not yet acknowledged, or the 291 acknowledgment might be on the way back. It is also possible that 292 either the fragment or the acknowledgment was lost on the way. 294 Because a meshed LLN might deliver frames out of order, it is 295 virtually impossible to differentiate these situations. In other 296 words, from the sender standpoint, all outstanding fragments might 297 still be in the network and contribute to its congestion. There is 298 an assumption, though, that after a certain amount of time, a frame 299 is either received or lost, so it is not causing congestion anymore. 300 This amount of time can be estimated based on the round trip delay 301 between the 6LoWPAN endpoints. The method detailed in [RFC2988] is 302 recommended for that computation. 304 The reader is encouraged to read through "Congestion Control 305 Principles" [RFC2914]. Additionally [RFC2309] and [RFC2581] provide 306 deeper information on why this mechanism is needed and how TCP 307 handles Congestion Control. Basically, the goal here is to manage 308 the amount of fragments present in the network; this is achieved by 309 to reducing the number of outstanding fragments over a congested path 310 by throttling the sources. 312 Section 7 describes how the sender decides how many fragments are 313 (re)sent before an acknowledgment is required, and how the sender 314 adapts that number to the network conditions. 316 6. New Dispatch types and headers 318 This specification extends "Transmission of IPv6 Packets over IEEE 319 802.15.4 Networks" [RFC4944] with 4 new dispatch types, for 320 Recoverable Fragments (RFRAG) headers with or without Acknowledgment 321 Request, and for the Acknowledgment back, with or without ECN Echo. 323 Pattern Header Type 324 +------------+-----------------------------------------------+ 325 | 11 101000 | RFRAG - Recoverable Fragment | 326 | 11 101001 | RFRAG-AR - RFRAG with Ack Request | 327 | 11 101010 | RFRAG-ACK - RFRAG Acknowledgment | 328 | 11 101011 | RFRAG-AEC - RFRAG Ack with ECN Echo | 329 +------------+-----------------------------------------------+ 331 Figure 1: Additional Dispatch Value Bit Patterns 333 In the following sections, the semantics of "datagram_tag," 334 "datagram_offset" and "datagram_size" and the reassembly process are 335 changed from [RFC4944] Section 5.3. "Fragmentation Type and Header." 336 The size and offset are expressed on the compressed packet per 337 [RFC6282] as opposed to the uncompressed - native packet - form. 339 6.1. Recoverable Fragment Dispatch type and Header 341 1 2 3 342 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 1 343 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 344 |1 1 1 0 1 0 0 X|datagram_offset| datagram_tag | 345 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 346 |Sequence | datagram_size | 347 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 348 X set == Ack Requested 350 Figure 2: Recoverable Fragment Dispatch type and Header 352 X bit 354 When set, the sender requires an Acknowledgment from the receiver 356 Sequence 357 The sequence number of the fragment. Fragments are numbered 358 [0..N] where N is in [0..31]. 360 6.2. Fragment Acknowledgement Dispatch type and Header 362 The specification also defines a 4-octet acknowledgement bitmap that 363 is used to carry selective acknowlegements for the received 364 fragments. A given offset in the bitmap maps one to one with a given 365 sequence number. 367 The offset of the bit in the bitmap indicates which fragment is 368 acknowledged as follows: 370 1 2 3 371 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 1 372 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 373 | Acknowledgment Bitmap | 374 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 375 ^ ^ 376 | | bitmap indicating whether: 377 | +--- Fragment with sequence 10 was received 378 +----------------------- Fragment with sequence 00 was received 380 Figure 3: Acknowledgement bitmap encoding 382 So in the example below Figure 4 it appears that all fragments from 383 sequence 0 to 20 were received but for sequence 1, 2 and 16 that were 384 either lost or are still in the network over a slower path. 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 |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| 390 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 392 Figure 4: Expanding 3 octets encoding 394 The acknowledgement bitmap is carried in a Fragment Acknowledgement 395 as follows: 397 1 2 3 398 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 1 399 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 400 |1 1 1 0 1 0 1 Y| datagram_tag | 401 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 402 | Acknowledgment Bitmap (32 bits) | 403 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 404 Figure 5: Fragment Acknowledgement Dispatch type and Header 406 Y bit 408 Explicit Congestion Notification (ECN) signalling 410 When set, the sender indicates that at least one of the 411 acknowledged fragments was received with an Explicit Congestion 412 Notification, indicating that the path followed by the fragments 413 is subject to congestion. 415 Acknowledgement Bitmap 417 An acknowledgement bitmap, whereby but at offset x indicates that 418 fragment x was received. 420 7. Fragments Recovery 422 The Recoverable Fragments header RFRAG and RFRAG-AR deprecate the 423 original fragment headers from [RFC4944] and replace them in the 424 fragmented packets. The Fragment Acknowledgement RFRAG-ACK is 425 introduced as a standalone header in message that is sent back to the 426 fragment source endpoint as known by its MAC address. This assumes 427 that the source MAC address in the fragment (is any) and datagram_tag 428 are enough information to send the Fragment Acknowledgement back to 429 the source fragmentation endpoint. 431 The 6LoWPAN endpoint that fragments the packets at 6LoWPAN level (the 432 sender) controls the Fragment Acknowledgements. If may do that at 433 any fragment to implement its own policy or perform congestion 434 control which is out of scope for this document. When the sender of 435 the fragment knows that an underlying mechanism protects the 436 Fragments already it MAY refrain from using the Acknowledgement 437 mechanism, and never set the Ack Requested bit. The 6LoWPAN endpoint 438 that recomposes the packets at 6LoWPAN level (the receiver) MUST 439 acknowledge the fragments it has received when asked to, and MAY 440 slightly defer that acknowledgement. 442 The sender transfers a controlled number of fragments and MAY flag 443 the last fragment of a series with an acknowledgment request. The 444 received MUST acknowledge a fragment with the acknowledgment request 445 bit set. If any fragment immediately preceding an acknowledgment 446 request is still missing, the receiver MAY intentionally delay its 447 acknowledgment to allow in-transit fragments to arrive. delaying the 448 acknowledgement might defeat the round trip delay computation so it 449 should be configurable and not enabled by default. 451 The receiver interacts with the sender using an Acknowledgment 452 message with a bitmap that indicates which fragments were actually 453 received. The bitmap is a 32bit SWORD, which accommodates up to 32 454 fragments and is sufficient for the 6LoWPAN MTU. For all n in 455 [0..31], bit n is set to 1 in the bitmap to indicate that fragment 456 with sequence n was received, otherwise the bit is set to 0. All 457 zeroes is a NULL bitmap that indicates that the fragmentation process 458 was cancelled by the receiver for that datagram. 460 The receiver MAY issue unsolicited acknowledgments. An unsolicited 461 acknowledgment enables the sender endpoint to resume sending if it 462 had reached its maximum number of outstanding fragments or indicate 463 that the receiver has cancelled the process of an individual 464 datagram. Note that acknowledgments might consume precious resources 465 so the use of unsolicited acknowledgments should be configurable and 466 not enabled by default. 468 The sender arms a retry timer to cover the fragment that carries the 469 Acknowledgment request. Upon time out, the sender assumes that all 470 the fragments on the way are received or lost. The process must have 471 completed within an acceptable time that is within the boundaries of 472 upper layer retries. The method detailed in [RFC2988] is recommended 473 for the computation of the retry timer. It is expected that the 474 upper layer retries obey the same or friendly rules in which case a 475 single round of fragment recovery should fit within the upper layer 476 recovery timers. 478 Fragments are sent in a round robin fashion: the sender sends all the 479 fragments for a first time before it retries any lost fragment; lost 480 fragments are retried in sequence, oldest first. This mechanism 481 enables the receiver to acknowledge fragments that were delayed in 482 the network before they are actually retried. 484 When the sender decides that a packet should be dropped and the 485 fragmentation process canceled, it sends a pseudo fragment with the 486 datagram_offset, sequence and datagram_size all set to zero, and no 487 data. Upon reception of this message, the receiver should clean up 488 all resources for the packet associated to the datagram_tag. If an 489 acknowledgement is requested, the receiver responds with a NULL 490 bitmap. 492 The receiver might need to cancel the process of a fragmented packet 493 for internal reasons, for instance if it is out of recomposition 494 buffers, or considers that this packet is already fully recomposed 495 and passed to the upper layer. In that case, the receiver SHOULD 496 indicate so to the sender with a NULL bitmap. Upon an 497 acknowledgement with a NULL bitmap, the sender MUST drop the 498 datagram. 500 8. Forwarding Fragments 502 This specification enables intermediate routers to forward fragments 503 with no intermediate reconstruction of the entire packet. Upon the 504 first fragment, the routers lay an label along the path that is 505 followed by that fragment (that is IP routed), and all further 506 fragments are label switched along that path. As a consequence, 507 alternate routes not possible for individual fragments. The datagram 508 tag is used to carry the label, that is swapped at each hop. 510 8.1. Upon the first fragment 512 In route over the L2 source changes at each hop. The label that is 513 formed adnd placed in the datagram tag is associated to the source 514 MAC and only valid (and unique) for that source MAC. Say the first 515 fragment has: 517 Source IPv6 address = IP_A (maybe hops away) 519 Destination IPv6 address = IP_B (maybe hops away) 521 Source MAC = MAC_prv (prv as previous) 523 Datagram_tag= DT_prv 525 The intermediate router that forwards individual fragments does the 526 following: 528 a route lookup to get Next hop IPv6 towards IP_B, which resolves 529 as IP_nxt (nxt as next) 531 a ND resolution to get the MAC address associated to IP_nxt, which 532 resolves as MAC_nxt 534 Since it is a first fragment of a packet from that source MAC address 535 MAC_prv for that tag DT_prv, the router: 537 cleans up any leftover resource associated to the tupple (MAC_prv, 538 DT_prv) 540 allocates a new label for that flow, DT_nxt, from a Least Recently 541 Used pool or some siumilar procedure. 543 allocates a Label swap structure indexed by (MAC_prv, DT_prv) that 544 contains (MAC_nxt, DT_nxt) 546 allocates a Label swap structure indexed by (MAC_nxt, DT_nxt) that 547 contains (MAC_prv, DT_prv) 548 swaps the MAC info to from self to MAC_nxt 550 Swaps the datagram_tag to DT_nxt 552 At this point the router is all set and can forward the packet to 553 nxt. 555 8.2. Upon the next fragments 557 Upon next fragments (that are not first fragment), the router expects 558 to have already Label swap structure indexed by (MAC_prv, DT_prv). 559 The router: 561 lookups up the Label swap entry for (MAC_prv, DT_prv), which 562 resolves as (MAC_nxt, DT_nxt) 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 if the Label swap entry for (MAC_src, DT_src) is not found, the 572 router builds an RFRAG-ACK to indicate the error. The acknowledgment 573 message has the following information: 575 MAC info set to from self to MAC_prv as found in the fragment 577 Swaps the datagram_tag set to DT_prv 579 Bitmap of all zeroes to indicate the error 581 At this point the router is all set and can send the RFRAG-ACK back 582 ot the previous router. 584 8.3. Upon the fragment acknowledgements 586 Upon fragment acknowledgements next fragments (that are not first 587 fragment), the router expects to have already Label swap structure 588 indexed by (MAC_nxt, DT_nxt). The router: 590 lookups up the Label swap entry for (MAC_nxt, DT_nxt), which 591 resolves as (MAC_prv, DT_prv) 593 swaps the MAC info to from self to MAC_prv; 594 Swaps the datagram_tag to DT_prv 596 At this point the router is all set and can forward the RFRAG-ACK to 597 prv. 599 if the Label swap entry for (MAC_nxt, DT_nxt) is not found, it simply 600 drops the packet. 602 if the RFRAG-ACK indicates either an error or that the fragment was 603 fully receive, the router schedules the Label swap entries for 604 recycling. If the RFRAG-ACK is lost on the way back, the source may 605 retry the last fragments, which will result as an error RFRAG-ACK 606 from the first router on the way that has already cleaned up. 608 9. Security Considerations 610 The process of recovering fragments does not appear to create any 611 opening for new threat compared to "Transmission of IPv6 Packets over 612 IEEE 802.15.4 Networks" [RFC4944]. 614 10. IANA Considerations 616 Need extensions for formats defined in "Transmission of IPv6 Packets 617 over IEEE 802.15.4 Networks" [RFC4944]. 619 11. Acknowledgments 621 The author wishes to thank Jay Werb, Christos Polyzois, Soumitri 622 Kolavennu and Harry Courtice for their contribution and review. 624 12. References 626 12.1. Normative References 628 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 629 Requirement Levels", BCP 14, RFC 2119, March 1997. 631 [RFC2988] Paxson, V. and M. Allman, "Computing TCP's Retransmission 632 Timer", RFC 2988, November 2000. 634 [RFC4944] Montenegro, G., Kushalnagar, N., Hui, J., and D. Culler, 635 "Transmission of IPv6 Packets over IEEE 802.15.4 636 Networks", RFC 4944, September 2007. 638 [RFC6282] Hui, J. and P. Thubert, "Compression Format for IPv6 639 Datagrams over IEEE 802.15.4-Based Networks", RFC 6282, 640 September 2011. 642 12.2. Informative References 644 [I-D.mathis-frag-harmful] 645 Mathis, M., "Fragmentation Considered Very Harmful", 646 draft-mathis-frag-harmful-00 (work in progress), 647 July 2004. 649 [RFC1191] Mogul, J. and S. Deering, "Path MTU discovery", RFC 1191, 650 November 1990. 652 [RFC2309] Braden, B., Clark, D., Crowcroft, J., Davie, B., Deering, 653 S., Estrin, D., Floyd, S., Jacobson, V., Minshall, G., 654 Partridge, C., Peterson, L., Ramakrishnan, K., Shenker, 655 S., Wroclawski, J., and L. Zhang, "Recommendations on 656 Queue Management and Congestion Avoidance in the 657 Internet", RFC 2309, April 1998. 659 [RFC2581] Allman, M., Paxson, V., and W. Stevens, "TCP Congestion 660 Control", RFC 2581, April 1999. 662 [RFC2914] Floyd, S., "Congestion Control Principles", BCP 41, 663 RFC 2914, September 2000. 665 [RFC3168] Ramakrishnan, K., Floyd, S., and D. Black, "The Addition 666 of Explicit Congestion Notification (ECN) to IP", 667 RFC 3168, September 2001. 669 [RFC4919] Kushalnagar, N., Montenegro, G., and C. Schumacher, "IPv6 670 over Low-Power Wireless Personal Area Networks (6LoWPANs): 671 Overview, Assumptions, Problem Statement, and Goals", 672 RFC 4919, August 2007. 674 [RFC4963] Heffner, J., Mathis, M., and B. Chandler, "IPv4 Reassembly 675 Errors at High Data Rates", RFC 4963, July 2007. 677 [RFC5405] Eggert, L. and G. Fairhurst, "Unicast UDP Usage Guidelines 678 for Application Designers", BCP 145, RFC 5405, 679 November 2008. 681 [RFC6550] Winter, T., Thubert, P., Brandt, A., Hui, J., Kelsey, R., 682 Levis, P., Pister, K., Struik, R., Vasseur, JP., and R. 683 Alexander, "RPL: IPv6 Routing Protocol for Low-Power and 684 Lossy Networks", RFC 6550, March 2012. 686 Authors' Addresses 688 Pascal Thubert (editor) 689 Cisco Systems 690 Village d'Entreprises Green Side 691 400, Avenue de Roumanille 692 Batiment T3 693 Biot - Sophia Antipolis 06410 694 FRANCE 696 Phone: +33 4 97 23 26 34 697 Email: pthubert@cisco.com 699 Jonathan W. Hui 700 Cisco Systems 701 560 McCarthy Blvd. 702 MILPITAS, California 95035 703 USA 705 Email: johui@cisco.com