idnits 2.17.1 draft-ietf-babel-rfc6126bis-06.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 : ---------------------------------------------------------------------------- == There are 1 instance of lines with multicast IPv4 addresses in the document. If these are generic example addresses, they should be changed to use the 233.252.0.x range defined in RFC 5771 -- The draft header indicates that this document obsoletes RFC7557, 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 -- The document date (October 23, 2018) is 2012 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) ** Obsolete normative reference: RFC 5226 (Obsoleted by RFC 8126) Summary: 1 error (**), 0 flaws (~~), 2 warnings (==), 2 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network Working Group J. Chroboczek 3 Internet-Draft IRIF, University of Paris-Diderot 4 Obsoletes: 6126,7557 (if approved) D. Schinazi 5 Intended status: Standards Track Apple Inc. 6 Expires: April 26, 2019 October 23, 2018 8 The Babel Routing Protocol 9 draft-ietf-babel-rfc6126bis-06 11 Abstract 13 Babel is a loop-avoiding distance-vector routing protocol that is 14 robust and efficient both in ordinary wired networks and in wireless 15 mesh networks. This document describes the Babel routing protocol, 16 and obsoletes RFCs 6126 and 7557 18 Status of This Memo 20 This Internet-Draft is submitted in full conformance with the 21 provisions of BCP 78 and BCP 79. 23 Internet-Drafts are working documents of the Internet Engineering 24 Task Force (IETF). Note that other groups may also distribute 25 working documents as Internet-Drafts. The list of current Internet- 26 Drafts is at https://datatracker.ietf.org/drafts/current/. 28 Internet-Drafts are draft documents valid for a maximum of six months 29 and may be updated, replaced, or obsoleted by other documents at any 30 time. It is inappropriate to use Internet-Drafts as reference 31 material or to cite them other than as "work in progress." 33 This Internet-Draft will expire on April 26, 2019. 35 Copyright Notice 37 Copyright (c) 2018 IETF Trust and the persons identified as the 38 document authors. All rights reserved. 40 This document is subject to BCP 78 and the IETF Trust's Legal 41 Provisions Relating to IETF Documents 42 (https://trustee.ietf.org/license-info) in effect on the date of 43 publication of this document. Please review these documents 44 carefully, as they describe your rights and restrictions with respect 45 to this document. Code Components extracted from this document must 46 include Simplified BSD License text as described in Section 4.e of 47 the Trust Legal Provisions and are provided without warranty as 48 described in the Simplified BSD License. 50 Table of Contents 52 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 53 1.1. Features . . . . . . . . . . . . . . . . . . . . . . . . 3 54 1.2. Limitations . . . . . . . . . . . . . . . . . . . . . . . 4 55 1.3. Specification of Requirements . . . . . . . . . . . . . . 4 56 2. Conceptual Description of the Protocol . . . . . . . . . . . 5 57 2.1. Costs, Metrics and Neighbourship . . . . . . . . . . . . 5 58 2.2. The Bellman-Ford Algorithm . . . . . . . . . . . . . . . 5 59 2.3. Transient Loops in Bellman-Ford . . . . . . . . . . . . . 6 60 2.4. Feasibility Conditions . . . . . . . . . . . . . . . . . 7 61 2.5. Solving Starvation: Sequencing Routes . . . . . . . . . . 8 62 2.6. Requests . . . . . . . . . . . . . . . . . . . . . . . . 10 63 2.7. Multiple Routers . . . . . . . . . . . . . . . . . . . . 10 64 2.8. Overlapping Prefixes . . . . . . . . . . . . . . . . . . 11 65 3. Protocol Operation . . . . . . . . . . . . . . . . . . . . . 12 66 3.1. Message Transmission and Reception . . . . . . . . . . . 12 67 3.2. Data Structures . . . . . . . . . . . . . . . . . . . . . 12 68 3.3. Acknowledgments and acknowledgment requests . . . . . . . 16 69 3.4. Neighbour Acquisition . . . . . . . . . . . . . . . . . . 17 70 3.5. Routing Table Maintenance . . . . . . . . . . . . . . . . 20 71 3.6. Route Selection . . . . . . . . . . . . . . . . . . . . . 24 72 3.7. Sending Updates . . . . . . . . . . . . . . . . . . . . . 25 73 3.8. Explicit Requests . . . . . . . . . . . . . . . . . . . . 27 74 4. Protocol Encoding . . . . . . . . . . . . . . . . . . . . . . 31 75 4.1. Data Types . . . . . . . . . . . . . . . . . . . . . . . 32 76 4.2. Packet Format . . . . . . . . . . . . . . . . . . . . . . 33 77 4.3. TLV Format . . . . . . . . . . . . . . . . . . . . . . . 34 78 4.4. Sub-TLV Format . . . . . . . . . . . . . . . . . . . . . 34 79 4.5. Parser state . . . . . . . . . . . . . . . . . . . . . . 35 80 4.6. Details of Specific TLVs . . . . . . . . . . . . . . . . 36 81 4.7. Details of specific sub-TLVs . . . . . . . . . . . . . . 46 82 5. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 47 83 6. Security Considerations . . . . . . . . . . . . . . . . . . . 48 84 7. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 49 85 8. References . . . . . . . . . . . . . . . . . . . . . . . . . 49 86 8.1. Normative References . . . . . . . . . . . . . . . . . . 49 87 8.2. Informative References . . . . . . . . . . . . . . . . . 50 88 Appendix A. Cost and Metric Computation . . . . . . . . . . . . 50 89 A.1. Maintaining Hello History . . . . . . . . . . . . . . . . 51 90 A.2. Cost Computation . . . . . . . . . . . . . . . . . . . . 52 91 A.3. Metric Computation . . . . . . . . . . . . . . . . . . . 53 92 Appendix B. Constants . . . . . . . . . . . . . . . . . . . . . 54 93 Appendix C. Considerations for protocol extensions . . . . . . . 55 94 Appendix D. Stub Implementations . . . . . . . . . . . . . . . . 56 95 Appendix E. Software Availability . . . . . . . . . . . . . . . 57 96 Appendix F. Changes from previous versions . . . . . . . . . . . 57 97 F.1. Changes since RFC 6126 . . . . . . . . . . . . . . . . . 57 98 F.2. Changes since draft-ietf-babel-rfc6126bis-00 . . . . . . 58 99 F.3. Changes since draft-ietf-babel-rfc6126bis-01 . . . . . . 58 100 F.4. Changes since draft-ietf-babel-rfc6126bis-02 . . . . . . 58 101 F.5. Changes since draft-ietf-babel-rfc6126bis-03 . . . . . . 59 102 F.6. Changes since draft-ietf-babel-rfc6126bis-03 . . . . . . 59 103 F.7. Changes since draft-ietf-babel-rfc6126bis-04 . . . . . . 59 104 F.8. Changes since draft-ietf-babel-rfc6126bis-05 . . . . . . 60 105 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 60 107 1. Introduction 109 Babel is a loop-avoiding distance-vector routing protocol that is 110 designed to be robust and efficient both in networks using prefix- 111 based routing and in networks using flat routing ("mesh networks"), 112 and both in relatively stable wired networks and in highly dynamic 113 wireless networks. 115 1.1. Features 117 The main property that makes Babel suitable for unstable networks is 118 that, unlike naive distance-vector routing protocols [RIP], it 119 strongly limits the frequency and duration of routing pathologies 120 such as routing loops and black-holes during reconvergence. Even 121 after a mobility event is detected, a Babel network usually remains 122 loop-free. Babel then quickly reconverges to a configuration that 123 preserves the loop-freedom and connectedness of the network, but is 124 not necessarily optimal; in many cases, this operation requires no 125 packet exchanges at all. Babel then slowly converges, in a time on 126 the scale of minutes, to an optimal configuration. This is achieved 127 by using sequenced routes, a technique pioneered by Destination- 128 Sequenced Distance-Vector routing [DSDV]. 130 More precisely, Babel has the following properties: 132 o when every prefix is originated by at most one router, Babel never 133 suffers from routing loops; 135 o when a single prefix is originated by multiple routers, Babel may 136 occasionally create a transient routing loop for this particular 137 prefix; this loop disappears in a time proportional to its 138 diameter, and never again (up to an arbitrary garbage-collection 139 (GC) time) will the routers involved participate in a routing loop 140 for the same prefix; 142 o assuming bounded packet loss rates, any routing black-holes that 143 may appear after a mobility event are corrected in a time at most 144 proportional to the network's diameter. 146 Babel has provisions for link quality estimation and for fairly 147 arbitrary metrics. When configured suitably, Babel can implement 148 shortest-path routing, or it may use a metric based, for example, on 149 measured packet loss. 151 Babel nodes will successfully establish an association even when they 152 are configured with different parameters. For example, a mobile node 153 that is low on battery may choose to use larger time constants (hello 154 and update intervals, etc.) than a node that has access to wall 155 power. Conversely, a node that detects high levels of mobility may 156 choose to use smaller time constants. The ability to build such 157 heterogeneous networks makes Babel particularly adapted to the 158 unmanaged and wireless environment. 160 Finally, Babel is a hybrid routing protocol, in the sense that it can 161 carry routes for multiple network-layer protocols (IPv4 and IPv6), 162 whichever protocol the Babel packets are themselves being carried 163 over. 165 1.2. Limitations 167 Babel has two limitations that make it unsuitable for use in some 168 environments. First, Babel relies on periodic routing table updates 169 rather than using a reliable transport; hence, in large, stable 170 networks it generates more traffic than protocols that only send 171 updates when the network topology changes. In such networks, 172 protocols such as OSPF [OSPF], IS-IS [IS-IS], or the Enhanced 173 Interior Gateway Routing Protocol (EIGRP) [EIGRP] might be more 174 suitable. 176 Second, unless the optional algorithm described in Section 3.5.5 is 177 implemented, Babel does impose a hold time when a prefix is 178 retracted. While this hold time does not apply to the exact prefix 179 being retracted, and hence does not prevent fast reconvergence should 180 it become available again, it does apply to any shorter prefix that 181 covers it. This may make those implementations of Babel that do not 182 implement the optional algorithm described in Section 3.5.5 183 unsuitable for use in networks that implement automatic prefix 184 aggregation. 186 1.3. Specification of Requirements 188 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 189 "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and 190 "OPTIONAL" in this document are to be interpreted as described in BCP 191 14 [RFC2119] [RFC8174] when, and only when, they appear in all 192 capitals, as shown here. 194 2. Conceptual Description of the Protocol 196 Babel is a loop-avoiding distance vector protocol: it is based on the 197 Bellman-Ford protocol, just like the venerable RIP [RIP], but 198 includes a number of refinements that either prevent loop formation 199 altogether, or ensure that a loop disappears in a timely manner and 200 doesn't form again. 202 Conceptually, Bellman-Ford is executed in parallel for every source 203 of routing information (destination of data traffic). In the 204 following discussion, we fix a source S; the reader will recall that 205 the same algorithm is executed for all sources. 207 2.1. Costs, Metrics and Neighbourship 209 For every pair of neighbouring nodes A and B, Babel computes an 210 abstract value known as the cost of the link from A to B., written 211 C(A, B). Given a route between any two (not necessarily 212 neighbouring) nodes, the metric of the route is the sum of the costs 213 of all the edges along the route. The goal of the routing algorithm 214 is to compute, for every source S, the tree of routes of lowest 215 metric to S. 217 Costs and metrics need not be integers. In general, they can be 218 values in any algebra that satisfies two fairly general conditions 219 (Section 3.5.2). 221 A Babel node periodically sends Hello messages to all of its 222 neighbours; it also periodically sends an IHU ("I Heard You") message 223 to every neighbour from which it has recently heard a Hello. From 224 the information derived from Hello and IHU messages received from its 225 neighbour B, a node A computes the cost C(A, B) of the link from A to 226 B. 228 2.2. The Bellman-Ford Algorithm 230 Every node A maintains two pieces of data: its estimated distance to 231 S, written D(A), and its next-hop router to S, written NH(A). 232 Initially, D(S) = 0, D(A) is infinite, and NH(A) is undefined. 234 Periodically, every node B sends to all of its neighbours a route 235 update, a message containing D(B). When a neighbour A of B receives 236 the route update, it checks whether B is its selected next hop; if 237 that is the case, then NH(A) is set to B, and D(A) is set to C(A, B) 238 + D(B). If that is not the case, then A compares C(A, B) + D(B) to 239 its current value of D(A). If that value is smaller, meaning that 240 the received update advertises a route that is better than the 241 currently selected route, then NH(A) is set to B, and D(A) is set to 242 C(A, B) + D(B). 244 A number of refinements to this algorithm are possible, and are used 245 by Babel. In particular, convergence speed may be increased by 246 sending unscheduled "triggered updates" whenever a major change in 247 the topology is detected, in addition to the regular, scheduled 248 updates. Additionally, a node may maintain a number of alternate 249 routes, which are being advertised by neighbours other than its 250 selected neighbour, and which can be used immediately if the selected 251 route were to fail. 253 2.3. Transient Loops in Bellman-Ford 255 It is well known that a naive application of Bellman-Ford to 256 distributed routing can cause transient loops after a topology 257 change. Consider for example the following topology: 259 B 260 1 /| 261 1 / | 262 S --- A |1 263 \ | 264 1 \| 265 C 267 After convergence, D(B) = D(C) = 2, with NH(B) = NH(C) = A. 269 Suppose now that the link between S and A fails: 271 B 272 1 /| 273 / | 274 S A |1 275 \ | 276 1 \| 277 C 279 When it detects the failure of the link, A switches its next hop to B 280 (which is still advertising a route to S with metric 2), and 281 advertises a metric equal to 3, and then advertises a new route with 282 metric 3. This process of nodes changing selected neighbours and 283 increasing their metric continues until the advertised metric reaches 284 "infinity", a value larger than all the metrics that the routing 285 protocol is able to carry. 287 2.4. Feasibility Conditions 289 Bellman-Ford is a very robust algorithm: its convergence properties 290 are preserved when routers delay route acquisition or when they 291 discard some updates. Babel routers discard received route 292 announcements unless they can prove that accepting them cannot 293 possibly cause a routing loop. 295 More formally, we define a condition over route announcements, known 296 as the "feasibility condition", that guarantees the absence of 297 routing loops whenever all routers ignore route updates that do not 298 satisfy the feasibility condition. In effect, this makes Bellman- 299 Ford into a family of routing algorithms, parameterised by the 300 feasibility condition. 302 Many different feasibility conditions are possible. For example, BGP 303 can be modelled as being a distance-vector protocol with a (rather 304 drastic) feasibility condition: a routing update is only accepted 305 when the receiving node's AS number is not included in the update's 306 AS-Path attribute (note that BGP's feasibility condition does not 307 ensure the absence of transient "micro-loops" during reconvergence). 309 Another simple feasibility condition, used in the Destination- 310 Sequenced Distance-Vector (DSDV) routing protocol [DSDV] and in the 311 Ad hoc On-Demand Distance Vector (AODV) protocol, stems from the 312 following observation: a routing loop can only arise after a router 313 has switched to a route with a larger metric than the route that it 314 had previously selected. Hence, one could decide that a route is 315 feasible only when its metric at the local node would be no larger 316 than the metric of the currently selected route, i.e., an 317 announcement carrying a metric D(B) is accepted by A when C(A, B) + 318 D(B) <= D(A). If all routers obey this constraint, then the metric 319 at every router is nonincreasing, and the following invariant is 320 always preserved: if A has selected B as its successor, then D(B) < 321 D(A), which implies that the forwarding graph is loop-free. 323 Babel uses a slightly more refined feasibility condition, derived 324 from EIGRP [DUAL]. Given a router A, define the feasibility distance 325 of A, written FD(A), as the smallest metric that A has ever 326 advertised for S to any of its neighbours. An update sent by a 327 neighbour B of A is feasible when the metric D(B) advertised by B is 328 strictly smaller than A's feasibility distance, i.e., when D(B) < 329 FD(A). 331 It is easy to see that this latter condition is no more restrictive 332 than DSDV-feasibility. Suppose that node A obeys DSDV-feasibility; 333 then D(A) is nonincreasing, hence at all times D(A) <= FD(A). 334 Suppose now that A receives a DSDV-feasible update that advertises a 335 metric D(B). Since the update is DSDV-feasible, C(A, B) + D(B) <= 336 D(A), hence D(B) < D(A), and since D(A) <= FD(A), D(B) < FD(A). 338 To see that it is strictly less restrictive, consider the following 339 diagram, where A has selected the route through B, and D(A) = FD(A) = 340 2. Since D(C) = 1 < FD(A), the alternate route through C is feasible 341 for A, although its metric C(A, C) + D(C) = 5 is larger than that of 342 the currently selected route: 344 B 345 1 / \ 1 346 / \ 347 S A 348 \ / 349 1 \ / 4 350 C 352 To show that this feasibility condition still guarantees loop- 353 freedom, recall that at the time when A accepts an update from B, the 354 metric D(B) announced by B is no smaller than FD(B); since it is 355 smaller than FD(A), at that point in time FD(B) < FD(A). Since this 356 property is preserved when A sends updates, it remains true at all 357 times, which ensures that the forwarding graph has no loops. 359 2.5. Solving Starvation: Sequencing Routes 361 Obviously, the feasibility conditions defined above cause starvation 362 when a router runs out of feasible routes. Consider the following 363 diagram, where both A and B have selected the direct route to S: 365 A 366 1 /| D(A) = 1 367 / | FD(A) = 1 368 S |1 369 \ | D(B) = 2 370 2 \| FD(B) = 2 371 B 373 Suppose now that the link between A and S breaks: 375 A 376 | 377 | FD(A) = 1 378 S |1 379 \ | D(B) = 2 380 2 \| FD(B) = 2 381 B 383 The only route available from A to S, the one that goes through B, is 384 not feasible: A suffers from spurious starvation. At that point, the 385 whole subtree suffering from starvation must be reset, which is 386 essentially what EIGRP does when it performs a global synchronisation 387 of all the routers in the sarving subtree (the "active" phase of 388 EIGRP). 390 Babel reacts to starvation in a less drastic manner, by using 391 sequenced routes, a technique introduced by DSDV and adopted by AODV. 392 In addition to a metric, every route carries a sequence number, a 393 nondecreasing integer that is propagated unchanged through the 394 network and is only ever incremented by the source; a pair (s, m), 395 where s is a sequence number and m a metric, is called a distance. 397 A received update is feasible when either it is more recent than the 398 feasibility distance maintained by the receiving node, or it is 399 equally recent and the metric is strictly smaller. More formally, if 400 FD(A) = (s, m), then an update carrying the distance (s', m') is 401 feasible when either s' > s, or s = s' and m' < m. 403 Assuming the sequence number of S is 137, the diagram above becomes: 405 A 406 | 407 | FD(A) = (137, 1) 408 S |1 409 \ | D(B) = (137, 2) 410 2 \| FD(B) = (137, 2) 411 B 413 After S increases its sequence number, and the new sequence number is 414 propagated to B, we have: 416 A 417 | 418 | FD(A) = (137, 1) 419 S |1 420 \ | D(B) = (138, 2) 421 2 \| FD(B) = (138, 2) 422 B 424 at which point the route through B becomes feasible again. 426 Note that while sequence numbers are used for determining 427 feasibility, they are not necessarily used in route selection: a node 428 will normally ignore the sequence number when selecting the best 429 route to a given destination (Section 3.6). 431 2.6. Requests 433 In DSDV, the sequence number of a source is increased periodically. 434 A route becomes feasible again after the source increases its 435 sequence number, and the new sequence number is propagated through 436 the network, which may, in general, require a significant amount of 437 time. 439 Babel takes a different approach. When a node detects that it is 440 suffering from a potentially spurious starvation, it sends an 441 explicit request to the source for a new sequence number. This 442 request is forwarded hop by hop to the source, with no regard to the 443 feasibility condition. Upon receiving the request, the source 444 increases its sequence number and broadcasts an update, which is 445 forwarded to the requesting node. 447 Note that after a change in network topology not all such requests 448 will, in general, reach the source, as some will be sent over links 449 that are now broken. However, if the network is still connected, 450 then at least one among the nodes suffering from spurious starvation 451 has an (unfeasible) route to the source; hence, in the absence of 452 packet loss, at least one such request will reach the source. 453 (Resending requests a small number of times compensates for packet 454 loss.) 456 Since requests are forwarded with no regard to the feasibility 457 condition, they may, in general, be caught in a forwarding loop; this 458 is avoided by having nodes perform duplicate detection for the 459 requests that they forward. 461 2.7. Multiple Routers 463 The above discussion assumes that every prefix is originated by a 464 single router. In real networks, however, it is often necessary to 465 have a single prefix originated by multiple routers: for example, the 466 default route will be originated by all of the edge routers of a 467 routing domain. 469 Since synchronising sequence numbers between distinct routers is 470 problematic, Babel treats routes for the same prefix as distinct 471 entities when they are originated by different routers: every route 472 announcement carries the router-id of its originating router, and 473 feasibility distances are not maintained per prefix, but per source, 474 where a source is a pair of a router-id and a prefix. In effect, 475 Babel guarantees loop-freedom for the forwarding graph to every 476 source; since the union of multiple acyclic graphs is not in general 477 acyclic, Babel does not in general guarantee loop-freedom when a 478 prefix is originated by multiple routers, but any loops will be 479 broken in a time at most proportional to the diameter of the loop -- 480 as soon as an update has "gone around" the routing loop. 482 Consider for example the following topology, where A has selected the 483 default route through S, and B has selected the one through S': 485 1 1 1 486 ::/0 -- S --- A --- B --- S' -- ::/0 488 Suppose that both default routes fail at the same time; then nothing 489 prevents A from switching to B, and B simultaneously switching to A. 490 However, as soon as A has successfully advertised the new route to B, 491 the route through A will become unfeasible for B. Conversely, as 492 soon as B will have advertised the route through A, the route through 493 B will become unfeasible for A. 495 In effect, the routing loop disappears at the latest when routing 496 information has gone around the loop. Since this process can be 497 delayed by lost packets, Babel makes certain efforts to ensure that 498 updates are sent reliably after a router-id change (Section 3.7.2). 500 Additionally, after the routers have advertised the two routes, both 501 sources will be in their source tables, which will prevent them from 502 ever again participating in a routing loop involving routes from S 503 and S' (up to the source GC time, which, available memory permitting, 504 can be set to arbitrarily large values). 506 2.8. Overlapping Prefixes 508 In the above discussion, we have assumed that all prefixes are 509 disjoint, as is the case in flat ("mesh") routing. In practice, 510 however, prefixes may overlap: for example, the default route 511 overlaps with all of the routes present in the network. 513 After a route fails, it is not correct in general to switch to a 514 route that subsumes the failed route. Consider for example the 515 following configuration: 517 1 1 518 ::/0 -- A --- B --- C 520 Suppose that node C fails. If B forwards packets destined to C by 521 following the default route, a routing loop will form, and persist 522 until A learns of B's retraction of the direct route to C. B avoids 523 this pitfall by installing an "unreachable" route after a route is 524 retracted; this route is maintained until it can be guaranteed that 525 the former route has been retracted by all of B's neighbours 526 (Section 3.5.5). 528 3. Protocol Operation 530 Every Babel speaker is assigned a router-id, which is an arbitrary 531 string of 8 octets that is assumed unique across the routing domain. 532 For example, routers-ids could be assigned randomly, or they could 533 derived from a link-layer address. (The protocol encoding is 534 slightly more compact when router-ids are assigned in the same manner 535 as the IPv6 layer assigns host IDs.) 537 3.1. Message Transmission and Reception 539 Babel protocol packets are sent in the body of a UDP datagram (as 540 described in Section 4 below). Each Babel packet consists of zero or 541 more TLVs. Most TLVs may contain sub-TLVs. 543 The source address of a Babel packet is always a unicast address, 544 link-local in the case of IPv6. Babel packets may be sent to a well- 545 known (link-local) multicast address or to a (link-local) unicast 546 address. In normal operation, a Babel speaker sends both multicast 547 and unicast packets to its neighbours. 549 With the exception of Hello TLVs and acknowledgments, all Babel TLVs 550 can be sent to either unicast or multicast addresses, and their 551 semantics does not depend on whether the destination is a unicast or 552 a multicast address. Hence, a Babel speaker does not need to 553 determine the destination address of a packet that it receives in 554 order to interpret it. 556 A moderate amount of jitter may be applied to packets sent by a Babel 557 speaker: outgoing TLVs are buffered and SHOULD be sent with a small 558 random delay. This is done for two purposes: it avoids 559 synchronisation of multiple Babel speakers across a network [JITTER], 560 and it allows for the aggregation of multiple TLVs into a single 561 packet. 563 The exact delay and amount of jitter applied to a packet depends on 564 whether it contains any urgent TLVs. Acknowledgment TLVs MUST be 565 sent before the deadline specified in the corresponding request. The 566 particular class of updates specified in Section 3.7.2 MUST be sent 567 in a timely manner. The particular class of request and update TLVs 568 specified in Section 3.8.2 SHOULD be sent in a timely manner. 570 3.2. Data Structures 572 In this section, we give a description of the data structures that 573 every Babel speaker maintains. This description is conceptual: a 574 Babel speaker may use different data structures as long as the 575 resulting protocol is the same as the one described in this document. 577 For example, rather than maintaining a single table containing both 578 selected and unselected (fallback) routes, as described in 579 Section 3.2.6 belong, an actual implementation would probably use two 580 tables, one with selected routes and one with fallback routes. 582 3.2.1. Sequence number arithmetic 584 Sequence numbers (seqnos) appear in a number of Babel data 585 structures, and they are interpreted as integers modulo 2^16. For 586 the purposes of this document, arithmetic on sequence numbers is 587 defined as follows. 589 Given a seqno s and an integer n, the sum of s and n is defined by 591 s + n (modulo 2^16) = (s + n) MOD 2^16 593 or, equivalently, 595 s + n (modulo 2^16) = (s + n) AND 65535 597 where MOD is the modulo operation yielding a non-negative integer and 598 AND is the bitwise conjunction operation. 600 Given two sequence numbers s and s', the relation s is less than s' 601 (s < s') is defined by 603 s < s' (modulo 2^16) when 0 < ((s' - s) MOD 2^16) < 32768 605 or equivalently 607 s < s' (modulo 2^16) when s /= s' and ((s' - s) AND 32768) = 0. 609 3.2.2. Node Sequence Number 611 A node's sequence number is a 16-bit integer that is included in 612 route updates sent for routes originated by this node. 614 A node increments its sequence number (modulo 2^16) whenever it 615 receives a request for a new sequence number (Section 3.8.1.2). A 616 node SHOULD NOT increment its sequence number (seqno) spontaneously, 617 since increasing seqnos makes it less likely that other nodes will 618 have feasible alternate routes when their selected routes fail. 620 3.2.3. The Interface Table 622 The interface table contains the list of interfaces on which the node 623 speaks the Babel protocol. Every interface table entry contains the 624 interface's outgoing Multicast Hello seqno, a 16-bit integer that is 625 sent with each Multicast Hello TLV on this interface and is 626 incremented (modulo 2^16) whenever a Multicast Hello is sent. (Note 627 that an interface's Multicast Hello seqno is unrelated to the node's 628 seqno.) 630 There are two timers associated with each interface table entry -- 631 the multicast hello timer, which governs the sending of scheduled 632 Multicast Hello and IHU packets, and the update timer, which governs 633 the sending of periodic route updates. 635 3.2.4. The Neighbour Table 637 The neighbour table contains the list of all neighbouring interfaces 638 from which a Babel packet has been recently received. The neighbour 639 table is indexed by pairs of the form (interface, address), and every 640 neighbour table entry contains the following data: 642 o the local node's interface over which this neighbour is reachable; 644 o the address of the neighbouring interface; 646 o a history of recently received Multicast Hello packets from this 647 neighbour; this can, for example, be a sequence of n bits, for 648 some small value n, indicating which of the n hellos most recently 649 sent by this neighbour have been received by the local node; 651 o a history of recently received Unicast Hello packets from this 652 neighbour; 654 o the "transmission cost" value from the last IHU packet received 655 from this neighbour, or FFFF hexadecimal (infinity) if the IHU 656 hold timer for this neighbour has expired; 658 o the neighbour's expected incoming Multicast Hello sequence number, 659 an integer modulo 2^16. 661 o the neighbour's expected incoming Unicast Hello sequence number, 662 an integer modulo 2^16. 664 o the neighbour's outgoing Unicast Hello sequence number, an integer 665 modulo 2^16 that is sent with each Unicast Hello TLV to this 666 neighbour and is incremented (modulo 2^16) whenever a Unicast 667 Hello is sent. (Note that a neighbour's outgoing Unicast Hello 668 seqno is distinct from the interface's outgoing Multicast Hello 669 seqno.) 671 There are three timers associated with each neighbour entry -- the 672 multicast hello timer, which is initialised from the interval value 673 carried by scheduled Multicast Hello TLVs, the unicast hello timer, 674 which is initialised from the interval value carried by scheduled 675 Unicast Hello TLVs, and the IHU timer, which is initialised to a 676 small multiple of the interval carried in IHU TLVs. 678 Note that the neighbour table is indexed by IP addresses, not by 679 router-ids: neighbourship is a relationship between interfaces, not 680 between nodes. Therefore, two nodes with multiple interfaces can 681 participate in multiple neighbourship relationships, a situation that 682 can notably arise when wireless nodes with multiple radios are 683 involved. 685 3.2.5. The Source Table 687 The source table is used to record feasibility distances. It is 688 indexed by triples of the form (prefix, plen, router-id), and every 689 source table entry contains the following data: 691 o the prefix (prefix, plen), where plen is the prefix length, that 692 this entry applies to; 694 o the router-id of a router originating this prefix; 696 o a pair (seqno, metric), this source's feasibility distance. 698 There is one timer associated with each entry in the source table -- 699 the source garbage-collection timer. It is initialised to a time on 700 the order of minutes and reset as specified in Section 3.7.3. 702 3.2.6. The Route Table 704 The route table contains the routes known to this node. It is 705 indexed by triples of the form (prefix, plen, neighbour), and every 706 route table entry contains the following data: 708 o the source (prefix, plen, router-id) for which this route is 709 advertised; 711 o the neighbour that advertised this route; 713 o the metric with which this route was advertised by the neighbour, 714 or FFFF hexadecimal (infinity) for a recently retracted route; 716 o the sequence number with which this route was advertised; 718 o the next-hop address of this route; 719 o a boolean flag indicating whether this route is selected, i.e., 720 whether it is currently being used for forwarding and is being 721 advertised. 723 There is one timer associated with each route table entry -- the 724 route expiry timer. It is initialised and reset as specified in 725 Section 3.5.4. 727 Note that there are two distinct (seqno, metric) pairs associated to 728 each route: the route's distance, which is stored in the route table, 729 and the feasibility distance, stored in the source table and shared 730 between all routes with the same source. 732 3.2.7. The Table of Pending Seqno Requests 734 The table of pending seqno requests contains a list of seqno requests 735 that the local node has sent (either because they have been 736 originated locally, or because they were forwarded) and to which no 737 reply has been received yet. This table is indexed by triples of the 738 form (prefix, plen, router-id), and every entry in this table 739 contains the following data: 741 o the prefix, router-id, and seqno being requested; 743 o the neighbour, if any, on behalf of which we are forwarding this 744 request; 746 o a small integer indicating the number of times that this request 747 will be resent if it remains unsatisfied. 749 There is one timer associated with each pending seqno request; it 750 governs both the resending of requests and their expiry. 752 3.3. Acknowledgments and acknowledgment requests 754 A Babel speaker may request that a neighbour receiving a given packet 755 reply with an explicit acknowledgment within a given time. While the 756 use of acknowledgment requests is optional, every Babel speaker MUST 757 be able to reply to such a request. 759 An acknowledgment MUST be sent to a unicast destination. On the 760 other hand, acknowledgment requests may be sent to either unicast or 761 multicast destinations, in which case they request an acknowledgment 762 from all of the receiving nodes. 764 When to request acknowledgments is a matter of local policy; the 765 simplest strategy is to never request acknowledgments and to rely on 766 periodic updates to ensure that any reachable routes are eventually 767 propagated throughout the routing domain. In order to improve 768 convergence speed and reduce the amount of control traffic, 769 acknowledgment requests MAY be used in order to reliably send urgent 770 updates (Section 3.7.2) and retractions (Section 3.5.5), especially 771 when the number of neighbours on a given interface is small. Since 772 Babel is designed to deal gracefully with packet loss on unreliable 773 media, sending all packets with acknowledgment requests is not 774 necessary, and NOT RECOMMENDED, as the acknowledgments cause 775 additional traffic and may force additional Address Resolution 776 Protocol (ARP) or Neighbour Discovery (ND) exchanges. 778 3.4. Neighbour Acquisition 780 Neighbour acquisition is the process by which a Babel node discovers 781 the set of neighbours heard over each of its interfaces and 782 ascertains bidirectional reachability. On unreliable media, 783 neighbour acquisition additionally provides some statistics that may 784 be useful for link quality computation. 786 Before it can exchange routing information with a neighbour, a Babel 787 node MUST create an entry for that neighbour in the neighbour table. 788 When to do that is implementation-specific; suitable strategies 789 include creating an entry when any Babel packet is received, or 790 creating an entry when a Hello TLV is parsed. Similarly, in order to 791 conserve system resources, an implementation SHOULD discard an entry 792 when it has been unused for long enough; suitable strategies include 793 dropping the neighbour after a timeout, and dropping a neighbour when 794 the associated Hello histories become empty (see Appendix A.2). 796 3.4.1. Reverse Reachability Detection 798 Every Babel node sends Hello TLVs to its neighbours to indicate that 799 it is alive, at regular or irregular intervals. Each Hello TLV 800 carries an increasing (modulo 2^16) sequence number and an upper 801 bound on the time interval until the next Hello of the same type (see 802 below). If the time interval is set to 0, then the Hello TLV does 803 not establish a new promise: the deadline carried by the previous 804 Hello of the same type still applies to the next Hello (if the most 805 recent scheduled Hello of the right kind was received at time t0 and 806 carried interval i, then the previous promise of sending another 807 Hello before time t0 + i still holds). We say that a Hello is 808 "scheduled" if it carries a non-zero interval, and "unscheduled" 809 otherwise. 811 There are two kinds of Hellos: Multicast Hellos, which use a per- 812 interface Hello counter (the Multicast Hello seqno), and Unicast 813 Hellos, which use a per-neighbour counter (the Multicast Hello 814 seqno). A Multicast Hello with a given seqno MUST be sent to all 815 neighbours on a given interface, either by sending it to a multicast 816 address or by sending it to one unicast address per neighbour (hence, 817 the term "Multicast Hello" is a slight misnomer). A Unicast Hello 818 carrying a given seqno should normally be sent to just one neighbour 819 (over unicast), since the sequence numbers of different neighbours 820 are not in general synchronised. 822 Multicast Hellos sent over multicast can be used for neighbour 823 discovery; hence, a node SHOULD send periodic (scheduled) Multicast 824 Hellos unless neighbour discovery is performed by means outside of 825 the Babel protocol. A node MAY send Unicast Hellos or unscheduled 826 Hellos of either kind for any reason, such as reducing the amount of 827 multicast traffic or improving reliability on link technologies with 828 poor support for link-layer multicast. 830 A node MAY send a scheduled Hello ahead of time. A node MAY change 831 its scheduled Hello interval. The Hello interval MAY be decreased at 832 any time; it MAY be increased immediately before sending a Hello TLV, 833 but SHOULD NOT be increased at other times. (Equivalently, a node 834 SHOULD send a scheduled Hello immediately after increasing its Hello 835 interval.) 837 How to deal with received Hello TLVs and what statistics to maintain 838 are considered local implementation matters; typically, a node will 839 maintain some sort of history of recently received Hellos. An 840 example of a suitable algorithm is described in Appendix A.1. 842 After receiving a Hello, or determining that it has missed one, the 843 node recomputes the association's cost (Section 3.4.3) and runs the 844 route selection procedure (Section 3.6). 846 3.4.2. Bidirectional Reachability Detection 848 In order to establish bidirectional reachability, every node sends 849 periodic IHU ("I Heard You") TLVs to each of its neighbours. Since 850 IHUs carry an explicit interval value, they MAY be sent less often 851 than Hellos in order to reduce the amount of routing traffic in dense 852 networks; in particular, they SHOULD be sent less often than Hellos 853 over links with little packet loss. While IHUs are conceptually 854 unicast, they MAY be sent to a multicast address in order to avoid an 855 ARP or Neighbour Discovery exchange and to aggregate multiple IHUs 856 into a single packet. 858 In addition to the periodic IHUs, a node MAY, at any time, send an 859 unscheduled IHU packet. It MAY also, at any time, decrease its IHU 860 interval, and it MAY increase its IHU interval immediately before 861 sending an IHU, but SHOULD NOT increase it at any other time. 863 (Equivalently, a node SHOULD send an extra IHU immediately after 864 increasing its Hello interval.) 866 Every IHU TLV contains two pieces of data: the link's rxcost 867 (reception cost) from the sender's perspective, used by the neighbour 868 for computing link costs (Section 3.4.3), and the interval between 869 periodic IHU packets. A node receiving an IHU sets the value of the 870 txcost (transmission cost) maintained in the neighbour table to the 871 value contained in the IHU, and resets the IHU timer associated to 872 this neighbour to a small multiple of the interval value received in 873 the IHU. When a neighbour's IHU timer expires, the neighbour's 874 txcost is set to infinity. 876 After updating a neighbour's txcost, the receiving node recomputes 877 the neighbour's cost (Section 3.4.3) and runs the route selection 878 procedure (Section 3.6). 880 3.4.3. Cost Computation 882 A neighbourship association's link cost is computed from the values 883 maintained in the neighbour table: the statistics kept in the 884 neighbour table about the reception of Hellos, and the txcost 885 computed from received IHU packets. 887 For every neighbour, a Babel node computes a value known as this 888 neighbour's rxcost. This value is usually derived from the Hello 889 history, which may be combined with other data, such as statistics 890 maintained by the link layer. The rxcost is sent to a neighbour in 891 each IHU. 893 Since nodes do not necessarily send periodic Unicast Hellos but do 894 usually send periodic Multicast Hellos (Section 3.4.1), a node SHOULD 895 use an algorithm that yields a finite rxcost when only Multicast 896 Hellos are received, unless interoperability with nodes that only 897 send Multicast Hellos is not required. 899 How the txcost and rxcost are combined in order to compute a link's 900 cost is a matter of local policy; as far as Babel's correctness is 901 concerned, only the following conditions MUST be satisfied: 903 o the cost is strictly positive; 905 o if no Hello TLVs of either kind were received recently, then the 906 cost is infinite; 908 o if the txcost is infinite, then the cost is infinite. 910 Note that while this document does not constrain cost computation any 911 further, not all cost computation strategies will give good results. 912 See Appendix A.2 for examples of strategies for computing a link's 913 cost that are known to work well in practice. 915 3.5. Routing Table Maintenance 917 Conceptually, a Babel update is a quintuple (prefix, plen, router-id, 918 seqno, metric), where (prefix, plen) is the prefix for which a route 919 is being advertised, router-id is the router-id of the router 920 originating this update, seqno is a nondecreasing (modulo 2^16) 921 integer that carries the originating router seqno, and metric is the 922 announced metric. 924 Before being accepted, an update is checked against the feasibility 925 condition (Section 3.5.1), which ensures that the route does not 926 create a routing loop. If the feasibility condition is not 927 satisfied, the update is either ignored or prevents the route from 928 being selected, as described in Section 3.5.4. If the feasibility 929 condition is satisfied, then the update cannot possibly cause a 930 routing loop. 932 3.5.1. The Feasibility Condition 934 The feasibility condition is applied to all received updates. The 935 feasibility condition compares the metric in the received update with 936 the metrics of the updates previously sent by the receiving node; 937 updates that fail the feasibility condition, and therefore have 938 metrics large enough to cause a routing loop, are either ignored or 939 prevent the resulting route from being selected. 941 A feasibility distance is a pair (seqno, metric), where seqno is an 942 integer modulo 2^16 and metric is a positive integer. Feasibility 943 distances are compared lexicographically, with the first component 944 inverted: we say that a distance (seqno, metric) is strictly better 945 than a distance (seqno', metric'), written 947 (seqno, metric) < (seqno', metric') 949 when 951 seqno > seqno' or (seqno = seqno' and metric < metric') 953 where sequence numbers are compared modulo 2^16. 955 Given a source (prefix, plen, router-id), a node's feasibility 956 distance for this source is the minimum, according to the ordering 957 defined above, of the distances of all the finite updates ever sent 958 by this particular node for the prefix (prefix, plen) and the given 959 router-id. Feasibility distances are maintained in the source table, 960 the exact procedure is given in Section 3.7.3. 962 A received update is feasible when either it is a retraction (its 963 metric is FFFF hexadecimal), or the advertised distance is strictly 964 better, in the sense defined above, than the feasibility distance for 965 the corresponding source. More precisely, a route advertisement 966 carrying the quintuple (prefix, plen, router-id, seqno, metric) is 967 feasible if one of the following conditions holds: 969 o metric is infinite; or 971 o no entry exists in the source table indexed by (prefix, plen, 972 router-id); or 974 o an entry (prefix, plen, router-id, seqno', metric') exists in the 975 source table, and either 977 * seqno' < seqno or 979 * seqno = seqno' and metric < metric'. 981 Note that the feasibility condition considers the metric advertised 982 by the neighbour, not the route's metric; hence, a fluctuation in a 983 neighbour's cost cannot render a selected route unfeasible. Note 984 further that retractions (updates with infinite metric) are always 985 feasible, since they cannot possibly cause a routing loop. 987 3.5.2. Metric Computation 989 A route's metric is computed from the metric advertised by the 990 neighbour and the neighbour's link cost. Just like cost computation, 991 metric computation is considered a local policy matter; as far as 992 Babel is concerned, the function M(c, m) used for computing a metric 993 from a locally computed link cost and the metric advertised by a 994 neighbour MUST only satisfy the following conditions: 996 o if c is infinite, then M(c, m) is infinite; 998 o M is strictly monotonic: M(c, m) > m. 1000 Additionally, the metric SHOULD satisfy the following condition: 1002 o M is left-distributive: if m <= m', then M(c, m) <= M(c, m'). 1004 Note that while strict monotonicity is essential to the integrity of 1005 the network (persistent routing loops may arise if it is not 1006 satisfied), left distributivity is not: if it is not satisfied, Babel 1007 will still converge to a loop-free configuration, but might not reach 1008 a global optimum (in fact, a global optimum may not even exist). 1010 As with cost computation, not all strategies for computing route 1011 metrics will give good results. In particular, some metrics are more 1012 likely than others to lead to routing instabilities (route flapping). 1013 In Appendix A.3, we give a number of examples of strictly monotonic, 1014 left-distributive routing metrics that are known to work well in 1015 practice. 1017 3.5.3. Encoding of Updates 1019 In a large network, the bulk of Babel traffic consists of route 1020 updates; hence, some care has been given to encoding them 1021 efficiently. An Update TLV itself only contains the prefix, seqno, 1022 and metric, while the next hop is derived either from the network- 1023 layer source address of the packet or from an explicit Next Hop TLV 1024 in the same packet. The router-id is derived from a separate Router- 1025 Id TLV in the same packet, which optimises the case when multiple 1026 updates are sent with the same router-id. 1028 Additionally, a prefix of the advertised prefix can be omitted in an 1029 Update TLV, in which case it is copied from a previous Update TLV in 1030 the same packet -- this is known as address compression 1031 (Section 4.6.9). 1033 Finally, as a special optimisation for the case when a router-id 1034 coincides with the interface-id part of an IPv6 address, the router- 1035 id can optionally be derived from the low-order bits of the 1036 advertised prefix. 1038 The encoding of updates is described in detail in Section 4.6. 1040 3.5.4. Route Acquisition 1042 When a Babel node receives an update (prefix, plen, router-id, seqno, 1043 metric) from a neighbour neigh with a link cost value equal to cost, 1044 it checks whether it already has a route table entry indexed by 1045 (prefix, plen, neigh). 1047 If no such entry exists: 1049 o if the update is unfeasible, it MAY be ignored; 1051 o if the metric is infinite (the update is a retraction of a route 1052 we do not know about), the update is ignored; 1054 o otherwise, a new entry is created in the route table, indexed by 1055 (prefix, plen, neigh), with source equal to (prefix, plen, router- 1056 id), seqno equal to seqno and an advertised metric equal to the 1057 metric carried by the update. 1059 If such an entry exists: 1061 o if the entry is currently selected, the update is unfeasible, and 1062 the router-id of the update is equal to the router-id of the 1063 entry, then the update MAY be ignored; 1065 o otherwise, the entry's sequence number, advertised metric, metric, 1066 and router-id are updated and, if the advertised metric is not 1067 infinite, the route's expiry timer is reset to a small multiple of 1068 the Interval value included in the update. If the update is 1069 unfeasible, then the (now unfeasible) entry MUST be immediately 1070 unselected. If the update caused the router-id of the entry to 1071 change, an update (possibly a retraction) MUST be sent in a timely 1072 manner (see Section 3.7.2). 1074 Note that the route table may contain unfeasible routes, either 1075 because they were created by an unfeasible update or due to a metric 1076 fluctuation. Such routes are never selected, since they are not 1077 known to be loop-free; should all the feasible routes become 1078 unusable, however, the unfeasible routes can be made feasible and 1079 therefore possible to select by sending requests along them (see 1080 Section 3.8.2). 1082 When a route's expiry timer triggers, the behaviour depends on 1083 whether the route's metric is finite. If the metric is finite, it is 1084 set to infinity and the expiry timer is reset. If the metric is 1085 already infinite, the route is flushed from the route table. 1087 After the route table is updated, the route selection procedure 1088 (Section 3.6) is run. 1090 3.5.5. Hold Time 1092 When a prefix P is retracted, because all routes are unfeasible or 1093 have an infinite metric (whether due to the expiry timer or to other 1094 reasons), and a shorter prefix P' that covers P is reachable, P' 1095 cannot in general be used for routing packets destined to P without 1096 running the risk of creating a routing loop (Section 2.8). 1098 To avoid this issue, whenever a prefix P is retracted, a route table 1099 entry with infinite metric is maintained as described in 1100 Section 3.5.4 above. As long as this entry is maintained, packets 1101 destined to an address within P MUST NOT be forwarded by following a 1102 route for a shorter prefix. This entry is removed as soon as a 1103 finite-metric update for prefix P is received and the resulting route 1104 selected. If no such update is forthcoming, the infinite metric 1105 entry SHOULD be maintained at least until it is guaranteed that no 1106 neighbour has selected the current node as next-hop for prefix P. 1107 This can be achieved by either: 1109 o waiting until the route's expiry timer has expired 1110 (Section 3.5.4); 1112 o sending a retraction with an acknowledgment request (Section 3.3) 1113 to every reachable neighbour that has not explicitly retracted 1114 prefix P and waiting for all acknowledgments. 1116 The former option is simpler and ensures that at that point, any 1117 routes for prefix P pointing at the current node have expired. 1118 However, since the expiry time can be as high as a few minutes, doing 1119 that prevents automatic aggregation by creating spurious black-holes 1120 for aggregated routes. The latter option is RECOMMENDED as it 1121 dramatically reduces the time for which a prefix is unreachable in 1122 the presence of aggregated routes. 1124 3.6. Route Selection 1126 Route selection is the process by which a single route for a given 1127 prefix is selected to be used for forwarding packets and to be re- 1128 advertised to a node's neighbours. 1130 Babel is designed to allow flexible route selection policies. As far 1131 as the protocol's correctness is concerned, the route selection 1132 policy MUST only satisfy the following properties: 1134 o a route with infinite metric (a retracted route) is never 1135 selected; 1137 o an unfeasible route is never selected. 1139 Note, however, that Babel does not naturally guarantee the stability 1140 of routing, and configuring conflicting route selection policies on 1141 different routers may lead to persistent route oscillation. 1143 Route selection is a difficult problem, since a good route selection 1144 policy needs to take into account multiple mutually contradictory 1145 criteria; in roughly decreasing order of importance, these are: 1147 o routes with a small metric should be preferred to routes with a 1148 large metric; 1150 o switching router-ids should be avoided; 1152 o routes through stable neighbours should be preferred to routes 1153 through unstable ones; 1155 o stable routes should be preferred to unstable ones; 1157 o switching next hops should be avoided. 1159 A simple but useful strategy is to choose the feasible route with the 1160 smallest metric, with a small amount of hysteresis applied to avoid 1161 switching router-ids too often. 1163 After the route selection procedure is run, triggered updates 1164 (Section 3.7.2) and requests (Section 3.8.2) are sent. 1166 3.7. Sending Updates 1168 A Babel speaker advertises to its neighbours its set of selected 1169 routes. Normally, this is done by sending one or more multicast 1170 packets containing Update TLVs on all of its connected interfaces; 1171 however, on link technologies where multicast is significantly more 1172 expensive than unicast, a node MAY choose to send multiple copies of 1173 updates in unicast packets, especially when the number of neighbours 1174 is small. 1176 Additionally, in order to ensure that any black-holes are reliably 1177 cleared in a timely manner, a Babel node sends retractions (updates 1178 with an infinite metric) for any recently retracted prefixes. 1180 If an update is for a route injected into the Babel domain by the 1181 local node (e.g., it carries the address of a local interface, the 1182 prefix of a directly attached network, or a prefix redistributed from 1183 a different routing protocol), the router-id is set to the local 1184 node's router-id, the metric is set to some arbitrary finite value 1185 (typically 0), and the seqno is set to the local router's sequence 1186 number. 1188 If an update is for a route learned from another Babel speaker, the 1189 router-id and sequence number are copied from the route table entry, 1190 and the metric is computed as specified in Section 3.5.2. 1192 3.7.1. Periodic Updates 1194 Every Babel speaker periodically advertises all of its selected 1195 routes on all of its interfaces, including any recently retracted 1196 routes. Since Babel doesn't suffer from routing loops (there is no 1197 "counting to infinity") and relies heavily on triggered updates 1198 (Section 3.7.2), this full dump only needs to happen infrequently. 1200 3.7.2. Triggered Updates 1202 In addition to periodic routing updates, a Babel speaker sends 1203 unscheduled, or triggered, updates in order to inform its neighbours 1204 of a significant change in the network topology. 1206 A change of router-id for the selected route to a given prefix may be 1207 indicative of a routing loop in formation; hence, a node MUST send a 1208 triggered update in a timely manner whenever it changes the selected 1209 router-id for a given destination. Additionally, it SHOULD make a 1210 reasonable attempt at ensuring that all reachable neighbours receive 1211 this update. 1213 There are two strategies for ensuring that. If the number of 1214 neighbours is small, then it is reasonable to send the update 1215 together with an acknowledgment request; the update is resent until 1216 all neighbours have acknowledged the packet, up to some number of 1217 times. If the number of neighbours is large, however, requesting 1218 acknowledgments from all of them might cause a non-negligible amount 1219 of network traffic; in that case, it may be preferable to simply 1220 repeat the update some reasonable number of times (say, 5 for 1221 wireless and 2 for wired links). 1223 A route retraction is somewhat less worrying: if the route retraction 1224 doesn't reach all neighbours, a black-hole might be created, which, 1225 unlike a routing loop, does not endanger the integrity of the 1226 network. When a route is retracted, a node SHOULD send a triggered 1227 update and SHOULD make a reasonable attempt at ensuring that all 1228 neighbours receive this retraction. 1230 Finally, a node MAY send a triggered update when the metric for a 1231 given prefix changes in a significant manner, due to a received 1232 update, because a link's cost has changed, or because a different 1233 next hop has been selected. A node SHOULD NOT send triggered updates 1234 for other reasons, such as when there is a minor fluctuation in a 1235 route's metric, when the selected next hop changes, or to propagate a 1236 new sequence number (except to satisfy a request, as specified in 1237 Section 3.8). 1239 3.7.3. Maintaining Feasibility Distances 1241 Before sending an update (prefix, plen, router-id, seqno, metric) 1242 with finite metric (i.e., not a route retraction), a Babel node 1243 updates the feasibility distance maintained in the source table. 1244 This is done as follows. 1246 If no entry indexed by (prefix, plen, router-id) exists in the source 1247 table, then one is created with value (prefix, plen, router-id, 1248 seqno, metric). 1250 If an entry (prefix, plen, router-id, seqno', metric') exists, then 1251 it is updated as follows: 1253 o if seqno > seqno', then seqno' := seqno, metric' := metric; 1255 o if seqno = seqno' and metric' > metric, then metric' := metric; 1257 o otherwise, nothing needs to be done. 1259 The garbage-collection timer for the entry is then reset. Note that 1260 the feasibility distance is not updated and the garbage-collection 1261 timer is not reset when a retraction (an update with infinite metric) 1262 is sent. 1264 When the garbage-collection timer expires, the entry is removed from 1265 the source table. 1267 3.7.4. Split Horizon 1269 When running over a transitive, symmetric link technology, e.g., a 1270 point-to-point link or a wired LAN technology such as Ethernet, a 1271 Babel node SHOULD use an optimisation known as split horizon. When 1272 split horizon is used on a given interface, a routing update for 1273 prefix P is not sent on the particular interface over which the 1274 selected route towards prefix P was learnt. 1276 Split horizon SHOULD NOT be applied to an interface unless the 1277 interface is known to be symmetric and transitive; in particular, 1278 split horizon is not applicable to decentralised wireless link 1279 technologies (e.g., IEEE 802.11 in ad hoc mode) when routing updates 1280 are sent over multicast. 1282 3.8. Explicit Requests 1284 In normal operation, a node's route table is populated by the regular 1285 and triggered updates sent by its neighbours. Under some 1286 circumstances, however, a node sends explicit requests in order to 1287 cause a resynchronisation with the source after a mobility event or 1288 to prevent a route from spuriously expiring. 1290 The Babel protocol provides two kinds of explicit requests: route 1291 requests, which simply request an update for a given prefix, and 1292 seqno requests, which request an update for a given prefix with a 1293 specific sequence number. The former are never forwarded; the latter 1294 are forwarded if they cannot be satisfied by the receiver. 1296 3.8.1. Handling Requests 1298 Upon receiving a request, a node either forwards the request or sends 1299 an update in reply to the request, as described in the following 1300 sections. If this causes an update to be sent, the update is either 1301 sent to a multicast address on the interface on which the request was 1302 received, or to the unicast address of the neighbour that sent the 1303 request. 1305 The exact behaviour is different for route requests and seqno 1306 requests. 1308 3.8.1.1. Route Requests 1310 When a node receives a route request for a given prefix, it checks 1311 its route table for a selected route to this exact prefix. If such a 1312 route exists, it MUST send an update (over unicast or over 1313 multicast); if such a route does not exist, it MUST send a retraction 1314 for that prefix. 1316 When a node receives a wildcard route request, it SHOULD send a full 1317 route table dump. Full route dumps MAY be rate-limited, especially 1318 if they are sent over multicast. 1320 3.8.1.2. Seqno Requests 1322 When a node receives a seqno request for a given router-id and 1323 sequence number, it checks whether its route table contains a 1324 selected entry for that prefix. If a selected route for the given 1325 prefix exists, it has finite metric, and either the router-ids are 1326 different or the router-ids are equal and the entry's sequence number 1327 is no smaller (modulo 2^16) than the requested sequence number, the 1328 node MUST send an update for the given prefix. If the router-ids 1329 match but the requested seqno is larger (modulo 2^16) than the route 1330 entry's, the node compares the router-id against its own router-id. 1331 If the router-id is its own, then it increases its sequence number by 1332 1 (modulo 2^16) and sends an update. A node MUST NOT increase its 1333 sequence number by more than 1 in response to a seqno request. 1335 Otherwise, if the requested router-id is not its own, the received 1336 request's hop count is 2 or more, and the node is advertising the 1337 prefix to its neighbours, the node selects a neighbour to forward the 1338 request to as follows: 1340 o if the node has one or more feasible routes toward the requested 1341 prefix with a next hop that is not the requesting node, then the 1342 node MUST forward the request to the next hop of one such route; 1344 o otherwise, if the node has one or more (not necessarily feasible) 1345 routes to the requested prefix with a next hop that is not the 1346 requesting node, then the node SHOULD forward the request to the 1347 next hop of one such route. 1349 In order to actually forward the request, the node decrements the hop 1350 count and sends the request in a unicast packet destined to the 1351 selected neighbour. 1353 A node SHOULD maintain a list of recently forwarded seqno requests 1354 and forward the reply (an update with a seqno sufficiently large to 1355 satisfy the request) in a timely manner. A node SHOULD compare every 1356 incoming seqno request against its list of recently forwarded seqno 1357 requests and avoid forwarding it if it is redundant (i.e., if it has 1358 recently sent a request with the same prefix, router-id and a seqno 1359 that is not smaller modulo 2^16). 1361 Since the request-forwarding mechanism does not necessarily obey the 1362 feasibility condition, it may get caught in routing loops; hence, 1363 requests carry a hop count to limit the time during which they remain 1364 in the network. However, since requests are only ever forwarded as 1365 unicast packets, the initial hop count need not be kept particularly 1366 low, and performing an expanding horizon search is not necessary. A 1367 single request MUST NOT be duplicated: it MUST NOT be forwarded to a 1368 multicast address, and it MUST NOT be forwarded to multiple 1369 neighbours. However, if a seqno request is resent by its originator, 1370 the subsequent copies MAY be forwarded to a different neighbour than 1371 the initial one. 1373 3.8.2. Sending Requests 1375 A Babel node MAY send a route or seqno request at any time, to a 1376 multicast or a unicast address; there is only one case when 1377 originating requests is required (Section 3.8.2.1). 1379 3.8.2.1. Avoiding Starvation 1381 When a route is retracted or expires, a Babel node usually switches 1382 to another feasible route for the same prefix. It may be the case, 1383 however, that no such routes are available. 1385 A node that has lost all feasible routes to a given destination but 1386 still has unexpired unfeasible routes to that destination MUST send a 1387 seqno request; if it doesn't have any such routes, it MAY still send 1388 a seqno request. The router-id of the request is set to the router- 1389 id of the route that it has just lost, and the requested seqno is the 1390 value contained in the source table plus 1. 1392 If the node has any (unfeasible) routes to the requested destination, 1393 then it MUST send the request to at least one of the next-hop 1394 neighbours that advertised these routes, and SHOULD send it to all of 1395 them; in any case, it MAY send the request to any other neighbours, 1396 whether they advertise a route to the requested destination or not. 1397 A simple implementation strategy is therefore to unconditionally 1398 multicast the request over all interfaces. 1400 Similar requests will be sent by other nodes that are affected by the 1401 route's loss. If the network is still connected, and assuming no 1402 packet loss, then at least one of these requests will be forwarded to 1403 the source, resulting in a route being advertised with a new sequence 1404 number. (Due to duplicate suppression, only a small number of such 1405 requests will actually reach the source.) 1407 In order to compensate for packet loss, a node SHOULD repeat such a 1408 request a small number of times if no route becomes feasible within a 1409 short time. In the presence of heavy packet loss, however, all such 1410 requests might be lost; in that case, the mechanism in the next 1411 section will eventually ensure that a new seqno is received. 1413 3.8.2.2. Dealing with Unfeasible Updates 1415 When a route's metric increases, a node might receive an unfeasible 1416 update for a route that it has currently selected. As specified in 1417 Section 3.5.1, the receiving node will either ignore the update or 1418 unselect the route. 1420 In order to keep routes from spuriously expiring because they have 1421 become unfeasible, a node SHOULD send a unicast seqno request when it 1422 receives an unfeasible update for a route that is currently selected. 1423 The requested sequence number is computed from the source table as in 1424 Section 3.8.2.1 above. 1426 Additionally, since metric computation does not necessarily coincide 1427 with the delay in propagating updates, a node might receive an 1428 unfeasible update from a currently unselected neighbour that is 1429 preferable to the currently selected route (e.g., because it has a 1430 much smaller metric); in that case, the node SHOULD send a unicast 1431 seqno request to the neighbour that advertised the preferable update. 1433 3.8.2.3. Preventing Routes from Expiring 1435 In normal operation, a route's expiry timer never triggers: since a 1436 route's hold time is computed from an explicit interval included in 1437 Update TLVs, a new update (possibly a retraction) should arrive in 1438 time to prevent a route from expiring. 1440 In the presence of packet loss, however, it may be the case that no 1441 update is successfully received for an extended period of time, 1442 causing a route to expire. In order to avoid such spurious expiry, 1443 shortly before a selected route expires, a Babel node SHOULD send a 1444 unicast route request to the neighbour that advertised this route; 1445 since nodes always send either updates or retractions in response to 1446 non-wildcard route requests (Section 3.8.1.1), this will usually 1447 result in the route being either refreshed or retracted. 1449 3.8.2.4. Acquiring New Neighbours 1451 In order to speed up convergence after a mobility event, a node MAY 1452 send a unicast wildcard request after acquiring a new neighbour. 1453 Additionally, a node MAY send a small number of multicast wildcard 1454 requests shortly after booting. Note however that doing that 1455 carelessly can cause serious congestion when a whole network is 1456 rebooted, especially on link layers with high per-packet overhead 1457 (e.g., IEEE 802.11). 1459 4. Protocol Encoding 1461 A Babel packet is sent as the body of a UDP datagram, with network- 1462 layer hop count set to 1, destined to a well-known multicast address 1463 or to a unicast address, over IPv4 or IPv6; in the case of IPv6, 1464 these addresses are link-local. Both the source and destination UDP 1465 port are set to a well-known port number. A Babel packet MUST be 1466 silently ignored unless its source address is either a link-local 1467 IPv6 address or an IPv4 address belonging to the local network, and 1468 its source port is the well-known Babel port. It MAY be silently 1469 ignored if its destination address is a global IPv6 address. 1471 In order to minimise the number of packets being sent while avoiding 1472 lower-layer fragmentation, a Babel node SHOULD attempt to maximise 1473 the size of the packets it sends, up to the outgoing interface's MTU 1474 adjusted for lower-layer headers (28 octets for UDP over IPv4, 48 1475 octets for UDP over IPv6). It MUST NOT send packets larger than the 1476 attached interface's MTU adjusted for lower-layer headers or 512 1477 octets, whichever is larger, but not exceeding 2^16 - 1 adjusted for 1478 lower-layer headers. Every Babel speaker MUST be able to receive 1479 packets that are as large as any attached interface's MTU adjusted 1480 for lower-layer headers or 512 octets, whichever is larger. Babel 1481 packets MUST NOT be sent in IPv6 Jumbograms. 1483 In order to avoid global synchronisation of a Babel network and to 1484 aggregate multiple TLVs into large packets, a Babel node SHOULD 1485 buffer every TLV and delay sending a packet by a small, randomly 1486 chosen delay [JITTER]. In order to allow accurate computation of 1487 packet loss rates, this delay MUST NOT be larger than half the 1488 advertised Hello interval. 1490 4.1. Data Types 1492 4.1.1. Interval 1494 Relative times are carried as 16-bit values specifying a number of 1495 centiseconds (hundredths of a second). This allows times up to 1496 roughly 11 minutes with a granularity of 10ms, which should cover all 1497 reasonable applications of Babel. 1499 4.1.2. Router-Id 1501 A router-id is an arbitrary 8-octet value. A router-id MUST NOT 1502 consist of either all zeroes or all ones. 1504 4.1.3. Address 1506 Since the bulk of the protocol is taken by addresses, multiple ways 1507 of encoding addresses are defined. Additionally, a common subnet 1508 prefix may be omitted when multiple addresses are sent in a single 1509 packet -- this is known as address compression (Section 4.6.9). 1511 Address encodings: 1513 o AE 0: wildcard address. The value is 0 octets long. 1515 o AE 1: IPv4 address. Compression is allowed. 4 octets or less. 1517 o AE 2: IPv6 address. Compression is allowed. 16 octets or less. 1519 o AE 3: link-local IPv6 address. Compression is not allowed. The 1520 value is 8 octets long, a prefix of fe80::/64 is implied. 1522 The address family associated to an address encoding is either IPv4 1523 or IPv6; it is undefined for AE 0, IPv4 for AE 1, and IPv6 for AEs 2 1524 and 3. 1526 4.1.4. Prefixes 1528 A network prefix is encoded just like a network address, but it is 1529 stored in the smallest number of octets that are enough to hold the 1530 significant bits (up to the prefix length). 1532 4.2. Packet Format 1534 A Babel packet consists of a 4-octet header, followed by a sequence 1535 of TLVs (the packet body), optionally followed by a second sequence 1536 of TLVs (the packet trailer). 1538 0 1 2 3 1539 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 1540 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1541 | Magic | Version | Body length | 1542 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1543 | Packet Body ... 1544 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+- 1545 | Packet Trailer... 1546 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+- 1548 Fields : 1550 Magic The arbitrary but carefully chosen value 42 (decimal); 1551 packets with a first octet different from 42 MUST be 1552 silently ignored. 1554 Version This document specifies version 2 of the Babel protocol. 1555 Packets with a second octet different from 2 MUST be 1556 silently ignored. 1558 Body length The length in octets of the body following the packet 1559 header (excluding the Magic, Version and Body length 1560 fields, and excluding the packet trailer). 1562 Packet Body The packet body; a sequence of TLVs. 1564 Packet Trailer The packet trailer; another sequence of TLVs. 1566 The packet body and the packet trailer are both sequences of TLVs. A 1567 TLV may appear in the packet trailer only if it is explicitly allowed 1568 to do so: the receiver MUST silently ignore any TLV that appears in 1569 the packet trailer unless it is explicitly specified to be allowed in 1570 the packet trailer. Among the TLVs defined in this document, only 1571 Pad1 and PadN are allowed in the packet trailer; since these TLVs are 1572 ignored in any case, an implementation MAY silently ignore the packet 1573 trailer unless it implements at least one extension that uses the 1574 packet trailer. 1576 4.3. TLV Format 1578 With the exception of Pad1, all TLVs have the following structure: 1580 0 1 2 3 1581 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 1582 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1583 | Type | Length | Payload... 1584 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+- 1586 Fields : 1588 Type The type of the TLV. 1590 Length The length of the body, exclusive of the Type and Length 1591 fields. If the body is longer than the expected length of 1592 a given type of TLV, any extra data MUST be silently 1593 ignored. 1595 Payload The TLV payload, which consists of a body and, for selected 1596 TLV types, an optional list of sub-TLVs. 1598 TLVs with an unknown type value MUST be silently ignored. 1600 4.4. Sub-TLV Format 1602 Every TLV carries an explicit length in its header; however, most 1603 TLVs are self-terminating, in the sense that it is possible to 1604 determine the length of the body without reference to the explicit 1605 Length field. If a TLV has a self-terminating format, then it MAY 1606 allow a sequence of sub-TLVs to follow the body. 1608 Sub-TLVs have the same structure as TLVs. With the exception of 1609 PAD1, all TLVs have the following structure: 1611 0 1 2 3 1612 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 1613 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1614 | Type | Length | Body... 1615 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+- 1617 Fields : 1619 Type The type of the sub-TLV. 1621 Length The length of the body, in octets, exclusive of the Type 1622 and Length fields. 1624 Body The sub-TLV body, the interpretation of which depends on 1625 both the type of the sub-TLV and the type of the TLV within 1626 which it is embedded. 1628 The most-significant bit of the sub-TLV, called the mandatory bit, 1629 indicates how to handle unknown sub-TLVs. If the mandatory bit is 1630 not set, then an unknown sub-TLV MUST be silently ignored, and the 1631 rest of the TLV processed normally. If the mandatory bit is set, 1632 then the whole enclosing TLV MUST be silently ignored (except for 1633 updating the parser state by a Router-Id, Next-Hop or Update TLV, see 1634 Section 4.6.7, Section 4.6.8, and Section 4.6.9). 1636 4.5. Parser state 1638 Babel uses a stateful parser: a TLV may refer to data from a previous 1639 TLV. The parser state consists of the following pieces of data: 1641 o for each address encoding that allows compression, the current 1642 default prefix; this is undefined at the start of the packet, and 1643 is updated by each Update TLV with the Prefix flag set 1644 (Section 4.6.9); 1646 o for each address family (IPv4 or IPv6), the current next-hop; this 1647 is the source address of the enclosing packet for the matching 1648 address family at the start of a packet, and is updated by each 1649 Next-Hop TLV (Section 4.6.8); 1651 o the current router-id; this is undefined at the start of the 1652 packet, and is updated by each Router-ID TLV (Section 4.6.7) and 1653 by each Update TLV with Router-Id flag set. 1655 Since the parser state is separate from the bulk of Babel's state, 1656 and since for correct parsing it must be identical across 1657 implementations, it is updated before checking for mandatory TLVs: 1658 parsing a TLV MUST update the parser state even if the TLV is 1659 otherwise ignored due to an unknown mandatory sub-TLV. 1661 None of the TLVs that modify the parser state are allowed in the 1662 packet trailer; hence, an implementation may choose to use a 1663 dedicated stateless parser to parse the packet trailer. 1665 4.6. Details of Specific TLVs 1667 4.6.1. Pad1 1669 0 1670 0 1 2 3 4 5 6 7 1671 +-+-+-+-+-+-+-+-+ 1672 | Type = 0 | 1673 +-+-+-+-+-+-+-+-+ 1675 Fields : 1677 Type Set to 0 to indicate a Pad1 TLV. 1679 This TLV is silently ignored on reception. It is allowed in the 1680 packet trailer. 1682 4.6.2. PadN 1684 0 1 2 3 1685 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 1686 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1687 | Type = 1 | Length | MBZ... 1688 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+- 1690 Fields : 1692 Type Set to 1 to indicate a PadN TLV. 1694 Length The length of the body, exclusive of the Type and Length 1695 fields. 1697 MBZ Set to 0 on transmission. 1699 This TLV is silently ignored on reception. It is allowed in the 1700 packet trailer. 1702 4.6.3. Acknowledgment Request 1704 0 1 2 3 1705 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 1706 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1707 | Type = 2 | Length | Reserved | 1708 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1709 | Nonce | Interval | 1710 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1711 This TLV requests that the receiver send an Acknowledgment TLV within 1712 the number of centiseconds specified by the Interval field. 1714 Fields : 1716 Type Set to 2 to indicate an Acknowledgment Request TLV. 1718 Length The length of the body, exclusive of the Type and Length 1719 fields. 1721 Reserved Sent as 0 and MUST be ignored on reception. 1723 Nonce An arbitrary value that will be echoed in the receiver's 1724 Acknowledgment TLV. 1726 Interval A time interval in centiseconds after which the sender will 1727 assume that this packet has been lost. This MUST NOT be 0. 1728 The receiver MUST send an Acknowledgment TLV before this 1729 time has elapsed (with a margin allowing for propagation 1730 time). 1732 This TLV is self-terminating, and allows sub-TLVs. 1734 4.6.4. Acknowledgment 1736 0 1 2 3 1737 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 1738 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1739 | Type = 3 | Length | Nonce | 1740 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1742 This TLV is sent by a node upon receiving an Acknowledgment Request. 1744 Fields : 1746 Type Set to 3 to indicate an Acknowledgment TLV. 1748 Length The length of the body, exclusive of the Type and Length 1749 fields. 1751 Nonce Set to the Nonce value of the Acknowledgment Request that 1752 prompted this Acknowledgment. 1754 Since nonce values are not globally unique, this TLV MUST be sent to 1755 a unicast address. 1757 This TLV is self-terminating, and allows sub-TLVs. 1759 4.6.5. Hello 1761 0 1 2 3 1762 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 1763 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1764 | Type = 4 | Length | Flags | 1765 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1766 | Seqno | Interval | 1767 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1769 This TLV is used for neighbour discovery and for determining a 1770 neighbour's reception cost. 1772 Fields : 1774 Type Set to 4 to indicate a Hello TLV. 1776 Length The length of the body, exclusive of the Type and Length 1777 fields. 1779 Flags The individual bits of this field specify special handling 1780 of this TLV (see below). 1782 Seqno If the Unicast flag is set, this is the value of the 1783 sending node's outgoing Unicast Hello seqno for this 1784 neighbour. Otherwise, it is the sending node's outgoing 1785 Multicast Hello seqno for this interface. 1787 Interval If non-zero, this is an upper bound, expressed in 1788 centiseconds, on the time after which the sending node will 1789 send a new scheduled Hello TLV with the same setting of the 1790 Unicast flag. If this is 0, then this Hello represents an 1791 unscheduled Hello, and doesn't carry any new information 1792 about times at which Hellos are sent. 1794 The Flags field is interpreted as follows: 1796 0 1 1797 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 1798 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1799 |U|X|X|X|X|X|X|X|X|X|X|X|X|X|X|X| 1800 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1802 o U (Unicast) flag (8000 hexadecimal): if set, then this Hello 1803 represents a Unicast Hello, otherwise it represents a Multicast 1804 Hello; 1806 o X: all other bits MUST be sent as 0 and silently ignored on 1807 reception. 1809 Every time a Hello is sent, the corresponding seqno counter MUST be 1810 incremented. Since there is a single seqno counter for all the 1811 Multicast Hellos sent by a given node over a given interface, if the 1812 Unicast flag is not set, this TLV MUST be sent to all neighbors on 1813 this link, which can be achieved by sending to a multicast 1814 destination, or by sending multiple packets to the unicast addresses 1815 of all reachable neighbours. Conversely, if the Unicast flag is set, 1816 this TLV MUST be sent to a single neighbour, which can achieved by 1817 sending to a unicast destination. In order to avoid large 1818 discontinuities in link quality, multiple Hello TLVs SHOULD NOT be 1819 sent in the same packet. 1821 This TLV is self-terminating, and allows sub-TLVs. 1823 4.6.6. IHU 1825 0 1 2 3 1826 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 1827 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1828 | Type = 5 | Length | AE | Reserved | 1829 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1830 | Rxcost | Interval | 1831 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1832 | Address... 1833 +-+-+-+-+-+-+-+-+-+-+-+- 1835 An IHU ("I Heard You") TLV is used for confirming bidirectional 1836 reachability and carrying a link's transmission cost. 1838 Fields : 1840 Type Set to 5 to indicate an IHU TLV. 1842 Length The length of the body, exclusive of the Type and Length 1843 fields. 1845 AE The encoding of the Address field. This should be 1 or 3 1846 in most cases. As an optimisation, it MAY be 0 if the TLV 1847 is sent to a unicast address, if the association is over a 1848 point-to-point link, or when bidirectional reachability is 1849 ascertained by means outside of the Babel protocol. 1851 Reserved Sent as 0 and MUST be ignored on reception. 1853 Rxcost The rxcost according to the sending node of the interface 1854 whose address is specified in the Address field. The value 1855 FFFF hexadecimal (infinity) indicates that this interface 1856 is unreachable. 1858 Interval An upper bound, expressed in centiseconds, on the time 1859 after which the sending node will send a new IHU; this MUST 1860 NOT be 0. The receiving node will use this value in order 1861 to compute a hold time for this symmetric association. 1863 Address The address of the destination node, in the format 1864 specified by the AE field. Address compression is not 1865 allowed. 1867 Conceptually, an IHU is destined to a single neighbour. However, IHU 1868 TLVs contain an explicit destination address, and MAY be sent to a 1869 multicast address, as this allows aggregation of IHUs destined to 1870 distinct neighbours into a single packet and avoids the need for an 1871 ARP or Neighbour Discovery exchange when a neighbour is not being 1872 used for data traffic. 1874 IHU TLVs with an unknown value in the AE field MUST be silently 1875 ignored. 1877 This TLV is self-terminating, and allows sub-TLVs. 1879 4.6.7. Router-Id 1881 0 1 2 3 1882 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 1883 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1884 | Type = 6 | Length | Reserved | 1885 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1886 | | 1887 + Router-Id + 1888 | | 1889 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1891 A Router-Id TLV establishes a router-id that is implied by subsequent 1892 Update TLVs. This TLV sets the router-id even if it is otherwise 1893 ignored due to an unknown mandatory sub-TLV. 1895 Fields : 1897 Type Set to 6 to indicate a Router-Id TLV. 1899 Length The length of the body, exclusive of the Type and Length 1900 fields. 1902 Reserved Sent as 0 and MUST be ignored on reception. 1904 Router-Id The router-id for routes advertised in subsequent Update 1905 TLVs. This MUST NOT consist of all zeroes or all ones. 1907 This TLV is self-terminating, and allows sub-TLVs. 1909 4.6.8. Next Hop 1911 0 1 2 3 1912 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 1913 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1914 | Type = 7 | Length | AE | Reserved | 1915 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1916 | Next hop... 1917 +-+-+-+-+-+-+-+-+-+-+-+- 1919 A Next Hop TLV establishes a next-hop address for a given address 1920 family (IPv4 or IPv6) that is implied in subsequent Update TLVs. 1921 This TLV sets up the next-hop for subsequent Update TLVs even if it 1922 is otherwise ignored due to an unknown mandatory sub-TLV. 1924 Fields : 1926 Type Set to 7 to indicate a Next Hop TLV. 1928 Length The length of the body, exclusive of the Type and Length 1929 fields. 1931 AE The encoding of the Address field. This SHOULD be 1 (IPv4) 1932 or 3 (link-local IPv6), and MUST NOT be 0. 1934 Reserved Sent as 0 and MUST be ignored on reception. 1936 Next hop The next-hop address advertised by subsequent Update TLVs, 1937 for this address family. 1939 When the address family matches the network-layer protocol that this 1940 packet is transported over, a Next Hop TLV is not needed: in the 1941 absence of a Next Hop TLV in a given address family, the next hop 1942 address is taken to be the source address of the packet. 1944 Next Hop TLVs with an unknown value for the AE field MUST be silently 1945 ignored. 1947 This TLV is self-terminating, and allows sub-TLVs. 1949 4.6.9. Update 1951 0 1 2 3 1952 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 1953 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1954 | Type = 8 | Length | AE | Flags | 1955 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1956 | Plen | Omitted | Interval | 1957 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1958 | Seqno | Metric | 1959 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1960 | Prefix... 1961 +-+-+-+-+-+-+-+-+-+-+-+- 1963 An Update TLV advertises or retracts a route. As an optimisation, it 1964 can optionally have the side effect of establishing a new implied 1965 router-id and a new default prefix. 1967 Fields : 1969 Type Set to 8 to indicate an Update TLV. 1971 Length The length of the body, exclusive of the Type and Length 1972 fields. 1974 AE The encoding of the Prefix field. 1976 Flags The individual bits of this field specify special handling 1977 of this TLV (see below). 1979 Plen The length of the advertised prefix. 1981 Omitted The number of octets that have been omitted at the 1982 beginning of the advertised prefix and that should be taken 1983 from a preceding Update TLV in the same address family with 1984 the Prefix flag set. 1986 Interval An upper bound, expressed in centiseconds, on the time 1987 after which the sending node will send a new update for 1988 this prefix. This MUST NOT be 0. The receiving node will 1989 use this value to compute a hold time for the route table 1990 entry. The value FFFF hexadecimal (infinity) expresses 1991 that this announcement will not be repeated unless a 1992 request is received (Section 3.8.2.3). 1994 Seqno The originator's sequence number for this update. 1996 Metric The sender's metric for this route. The value FFFF 1997 hexadecimal (infinity) means that this is a route 1998 retraction. 2000 Prefix The prefix being advertised. This field's size is 2001 (Plen/8 - Omitted) rounded upwards. 2003 The Flags field is interpreted as follows: 2005 0 1 2 3 4 5 6 7 2006 +-+-+-+-+-+-+-+-+ 2007 |P|R|X|X|X|X|X|X| 2008 +-+-+-+-+-+-+-+-+ 2010 o P (Prefix) flag (80 hexadecimal): if set, then this Update 2011 establishes a new default prefix for subsequent Update TLVs with a 2012 matching address encoding within the same packet, even if this TLV 2013 is otherwise ignored due to an unknown mandatory sub-TLV; 2015 o R (Router-Id) flag (40 hexadecimal): if set, then this TLV 2016 establishes a new default router-id for this TLV and subsequent 2017 Update TLVs in the same packet, even if this TLV is otherwise 2018 ignored due to an unknown mandatory sub-TLV. This router-id is 2019 computed from the first address of the advertised prefix as 2020 follows: 2022 * if the length of the address is 8 octets or more, then the new 2023 router-id is taken from the 8 last octets of the address; 2025 * if the length of the address is smaller than 8 octets, then the 2026 new router-id consists of the required number of zero octets 2027 followed by the address, i.e., the address is stored on the 2028 right of the router-id. For example, for an IPv4 address, the 2029 router-id consists of 4 octets of zeroes followed by the IPv4 2030 address. 2032 o X: all other bits MUST be sent as 0 and silently ignored on 2033 reception. 2035 The prefix being advertised by an Update TLV is computed as follows: 2037 o the first Omitted octets of the prefix are taken from the previous 2038 Update TLV with the Prefix flag set and the same address encoding, 2039 even if it was ignored due to an unknown mandatory sub-TLV; 2041 o the next (Plen/8 - Omitted) rounded upwards octets are taken from 2042 the Prefix field; 2044 o the remaining octets are set to 0. If AE is 3 (link-local IPv6), 2045 Omitted MUST be 0) 2047 If the Metric field is finite, the router-id of the originating node 2048 for this announcement is taken from the prefix advertised by this 2049 Update if the Router-Id flag is set, computed as described above. 2050 Otherwise, it is taken either from the preceding Router-Id packet, or 2051 the preceding Update packet with the Router-Id flag set, whichever 2052 comes last, even if that TLV is otherwise ignored due to an unknown 2053 mandatory sub-TLV. 2055 The next-hop address for this update is taken from the last preceding 2056 Next Hop TLV with a matching address family (IPv4 or IPv6) in the 2057 same packet even if it was otherwise ignored due to an unknown 2058 mandatory sub-TLV; if no such TLV exists, it is taken from the 2059 network-layer source address of this packet. 2061 If the metric field is FFFF hexadecimal, this TLV specifies a 2062 retraction. In that case, the router-id, next-hop and seqno are not 2063 used. AE MAY then be 0, in which case this Update retracts all of 2064 the routes previously advertised by the sending interface. If the 2065 metric is finite, AE MUST NOT be 0. If the metric is infinite and AE 2066 is 0, Plen and Omitted MUST both be 0. 2068 Update TLVs with an unknown value in the AE field MUST be silently 2069 ignored. 2071 This TLV is self-terminating, and allows sub-TLVs. 2073 4.6.10. Route Request 2075 0 1 2 3 2076 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 2077 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 2078 | Type = 9 | Length | AE | Plen | 2079 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 2080 | Prefix... 2081 +-+-+-+-+-+-+-+-+-+-+-+- 2083 A Route Request TLV prompts the receiver to send an update for a 2084 given prefix, or a full route table dump. 2086 Fields : 2088 Type Set to 9 to indicate a Route Request TLV. 2090 Length The length of the body, exclusive of the Type and Length 2091 fields. 2093 AE The encoding of the Prefix field. The value 0 specifies 2094 that this is a request for a full route table dump (a 2095 wildcard request). 2097 Plen The length of the requested prefix. 2099 Prefix The prefix being requested. This field's size is Plen/8 2100 rounded upwards. 2102 A Request TLV prompts the receiver to send an update message 2103 (possibly a retraction) for the prefix specified by the AE, Plen, and 2104 Prefix fields, or a full dump of its route table if AE is 0 (in which 2105 case Plen MUST be 0 and Prefix is of length 0). 2107 This TLV is self-terminating, and allows sub-TLVs. 2109 4.6.11. Seqno Request 2111 0 1 2 3 2112 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 2113 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 2114 | Type = 10 | Length | AE | Plen | 2115 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 2116 | Seqno | Hop Count | Reserved | 2117 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 2118 | | 2119 + Router-Id + 2120 | | 2121 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 2122 | Prefix... 2123 +-+-+-+-+-+-+-+-+-+-+ 2125 A Seqno Request TLV prompts the receiver to send an Update for a 2126 given prefix with a given sequence number, or to forward the request 2127 further if it cannot be satisfied locally. 2129 Fields : 2131 Type Set to 10 to indicate a Seqno Request message. 2133 Length The length of the body, exclusive of the Type and Length 2134 fields. 2136 AE The encoding of the Prefix field. This MUST NOT be 0. 2138 Plen The length of the requested prefix. 2140 Seqno The sequence number that is being requested. 2142 Hop Count The maximum number of times that this TLV may be forwarded, 2143 plus 1. This MUST NOT be 0. 2145 Reserved Sent as 0 and MUST be ignored on reception. 2147 Router Id The Router-Id that is being requested. This MUST NOT 2148 consist of all zeroes or all ones. 2150 Prefix The prefix being requested. This field's size is Plen/8 2151 rounded upwards. 2153 A Seqno Request TLV prompts the receiving node to send a finite- 2154 metric Update for the prefix specified by the AE, Plen, and Prefix 2155 fields, with either a router-id different from what is specified by 2156 the Router-Id field, or a Seqno no less (modulo 2^16) than what is 2157 specified by the Seqno field. If this request cannot be satisfied 2158 locally, then it is forwarded according to the rules set out in 2159 Section 3.8.1.2. 2161 While a Seqno Request MAY be sent to a multicast address, it MUST NOT 2162 be forwarded to a multicast address and MUST NOT be forwarded to more 2163 than one neighbour. A request MUST NOT be forwarded if its Hop Count 2164 field is 1. 2166 This TLV is self-terminating, and allows sub-TLVs. 2168 4.7. Details of specific sub-TLVs 2170 4.7.1. Pad1 2172 0 1 2 3 4 5 6 7 2173 +-+-+-+-+-+-+-+-+ 2174 | Type = 0 | 2175 +-+-+-+-+-+-+-+-+ 2177 Fields : 2179 Type Set to 0 to indicate a Pad1 sub-TLV. 2181 This sub-TLV is silently ignored on reception. It is allowed within 2182 any TLV that allows sub-TLVs. 2184 4.7.2. PadN 2185 0 1 2 3 2186 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 2187 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 2188 | Type = 1 | Length | MBZ... 2189 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+- 2191 Fields : 2193 Type Set to 1 to indicate a PadN sub-TLV. 2195 Length The length of the body, in octets, exclusive of the Type 2196 and Length fields. 2198 MBZ Set to 0 on transmission. 2200 This sub-TLV is silently ignored on reception. It is allowed within 2201 any TLV that allows sub-TLVs. 2203 5. IANA Considerations 2205 IANA has registered the UDP port number 6696, called "babel", for use 2206 by the Babel protocol. 2208 IANA has registered the IPv6 multicast group ff02::1:6 and the IPv4 2209 multicast group 224.0.0.111 for use by the Babel protocol. 2211 IANA has created a registry called "Babel TLV Types". The values in 2212 this registry are not changed by this specification. 2214 IANA has created a registry called "Babel sub-TLV Types". Due to the 2215 addition of a Mandatory bit to the Babel protocol, the values in the 2216 "Babel sub-TLV Types" registry are amended as follows: 2218 +---------+-----------------------------------------+---------------+ 2219 | Type | Name | Reference | 2220 +---------+-----------------------------------------+---------------+ 2221 | 0 | Pad1 | this document | 2222 | | | | 2223 | 1 | PadN | this document | 2224 | | | | 2225 | 112-126 | Reserved for Experimental Use | this document | 2226 | | | | 2227 | 127 | Reserved for expansion of the type | this document | 2228 | | space | | 2229 | | | | 2230 | 240-254 | Reserved for Experimental Use | this document | 2231 | | | | 2232 | 255 | Reserved for expansion of the type | this document | 2233 | | space | | 2234 +---------+-----------------------------------------+---------------+ 2236 Existing assignments in the "Babel sub-TLV Types" registry in the 2237 range 2 to 111 are not changed by this specification. The values 224 2238 through 239, previously reserved for Experimental Use, are now 2239 unassigned. 2241 IANA has created a registry called "Babel Flags Values". IANA is 2242 instructed to rename this registry to "Babel Update Flags Values", 2243 with its contents unchanged. 2245 IANA is instructed to create a new registry called "Babel Hello Flags 2246 Values". The allocation policy for this registry is Specification 2247 Required [RFC5226]. The initial values in this registry are as 2248 follows: 2250 +------+------------+---------------+ 2251 | Bit | Name | Reference | 2252 +------+------------+---------------+ 2253 | 0 | Unicast | this document | 2254 | | | | 2255 | 1-15 | Unassigned | | 2256 +------+------------+---------------+ 2258 IANA is instructed to replace all references to RFCs 6126 and 7557 in 2259 all of the registries mentioned above by references to this document. 2261 6. Security Considerations 2263 As defined in this document, Babel is a completely insecure protocol. 2264 Any attacker can misdirect data traffic by advertising routes with a 2265 low metric or a high seqno. This issue can be solved either by a 2266 lower-layer security mechanism (e.g., link-layer security), or by 2267 deploying a suitable authentication mechanism within Babel itself. 2268 With the exception of Hello TLVs used for discovery, Babel control 2269 traffic can be carried over unicast, which makes it possible to 2270 protect Babel traffic with a protocol that can only protect unicast 2271 data, for example IPsec with IKEv2, or DTLS. 2273 The information that a Babel node announces to the whole routing 2274 domain is often sufficient to determine a mobile node's physical 2275 location with reasonable precision. The privacy issues that this 2276 causes can be mitigated somewhat by using randomly chosen router-ids 2277 and randomly chosen IP addresses, and changing them periodically. 2279 When carried over IPv6, Babel packets are ignored unless they are 2280 sent from a link-local IPv6 address; since routers don't forward 2281 link-local IPv6 packets, this provides protection against spoofed 2282 Babel packets being sent from the global Internet. No such natural 2283 protection exists when Babel packets are carried over IPv4. 2285 7. Acknowledgments 2287 A number of people have contributed text and ideas to this 2288 specification. The authors are particularly indebted to Matthieu 2289 Boutier, Gwendoline Chouasne, Margaret Cullen, Donald Eastlake and 2290 Toke Hoiland-Jorgensen. Earlier versions of this document greatly 2291 benefited from the input of Joel Halpern. The address compression 2292 technique was inspired by [PACKETBB]. 2294 8. References 2296 8.1. Normative References 2298 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 2299 Requirement Levels", BCP 14, RFC 2119, 2300 DOI 10.17487/RFC2119, March 1997. 2302 [RFC5226] Narten, T. and H. Alvestrand, "Guidelines for Writing an 2303 IANA Considerations Section in RFCs", BCP 26, RFC 5226, 2304 May 2008. 2306 [RFC8174] Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC 2307 2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174, 2308 May 2017. 2310 8.2. Informative References 2312 [DSDV] Perkins, C. and P. Bhagwat, "Highly Dynamic Destination- 2313 Sequenced Distance-Vector Routing (DSDV) for Mobile 2314 Computers", ACM SIGCOMM'94 Conference on Communications 2315 Architectures, Protocols and Applications 234-244, 1994. 2317 [DUAL] Garcia Luna Aceves, J., "Loop-Free Routing Using Diffusing 2318 Computations", IEEE/ACM Transactions on Networking 1:1, 2319 February 1993. 2321 [EIGRP] Albrightson, B., Garcia Luna Aceves, J., and J. Boyle, 2322 "EIGRP -- a Fast Routing Protocol Based on Distance 2323 Vectors", Proc. Interop 94, 1994. 2325 [ETX] De Couto, D., Aguayo, D., Bicket, J., and R. Morris, "A 2326 high-throughput path metric for multi-hop wireless 2327 networks", Proc. MobiCom 2003, 2003. 2329 [IS-IS] "Information technology -- Telecommunications and 2330 information exchange between systems -- Intermediate 2331 System to Intermediate System intra-domain routeing 2332 information exchange protocol for use in conjunction with 2333 the protocol for providing the connectionless-mode network 2334 service (ISO 8473)", ISO/IEC 10589:2002, 2002. 2336 [JITTER] Floyd, S. and V. Jacobson, "The synchronization of 2337 periodic routing messages", IEEE/ACM Transactions on 2338 Networking 2, 2, 122-136, April 1994. 2340 [OSPF] Moy, J., "OSPF Version 2", RFC 2328, April 1998. 2342 [PACKETBB] 2343 Clausen, T., Dearlove, C., Dean, J., and C. Adjih, 2344 "Generalized Mobile Ad Hoc Network (MANET) Packet/Message 2345 Format", RFC 5444, February 2009. 2347 [RIP] Malkin, G., "RIP Version 2", RFC 2453, November 1998. 2349 Appendix A. Cost and Metric Computation 2351 The strategy for computing link costs and route metrics is a local 2352 matter; Babel itself only requires that it comply with the conditions 2353 given in Section 3.4.3 and Section 3.5.2. Different nodes may use 2354 different strategies in a single network and may use different 2355 strategies on different interface types. This section describes the 2356 strategies used by the sample implementation of Babel. 2358 The sample implementation of Babel sends periodic Multicast Hellos, 2359 and never sends Unicast Hellos. It maintains statistics about the 2360 last 16 received Hello TLVs of each kind (Appendix A.1), computes 2361 costs by using the 2-out-of-3 strategy (Appendix A.2.1) on wired 2362 links, and ETX (Appendix A.2.2) on wireless links. It uses an 2363 additive algebra for metric computation (Appendix A.3.1). 2365 A.1. Maintaining Hello History 2367 For each neighbour, the sample implementation of Babel maintains two 2368 sets of Hello history, one for each kind of Hello, and an expected 2369 sequence number, one for Multicast and one for Unicast Hellos. Each 2370 Hello history is a vector of 16 bits, where a 1 value represents a 2371 received Hello, and a 0 value a missed Hello. For each kind of 2372 Hello, the expected sequence number, written ne, is the sequence 2373 number that is expected to be carried by the next received Hello from 2374 this neighbour. 2376 Whenever it receives a Hello packet of a given kind from a neighbour, 2377 a node compares the received sequence number nr for that kind of 2378 Hello with its expected sequence number ne. Depending on the outcome 2379 of this comparison, one of the following actions is taken: 2381 o if the two differ by more than 16 (modulo 2^16), then the sending 2382 node has probably rebooted and lost its sequence number; the whole 2383 associated neighbour table entry is flushed and a new one is 2384 created; 2386 o otherwise, if the received nr is smaller (modulo 2^16) than the 2387 expected sequence number ne, then the sending node has increased 2388 its Hello interval without us noticing; the receiving node removes 2389 the last (ne - nr) entries from this neighbour's Hello history (we 2390 "undo history"); 2392 o otherwise, if nr is larger (modulo 2^16) than ne, then the sending 2393 node has decreased its Hello interval, and some Hellos were lost; 2394 the receiving node adds (nr - ne) 0 bits to the Hello history (we 2395 "fast-forward"). 2397 The receiving node then appends a 1 bit to the Hello history and sets 2398 ne to (nr + 1). If the Interval field of the received Hello is not 2399 zero, it resets the neighbour's hello timer to 1.5 times the 2400 advertised Interval (the extra margin allows for delay due to 2401 jitter). 2403 Whenever either Hello timer associated to a neighbour expires, the 2404 local node adds a 0 bit to this neighbour's Hello history, and 2405 increments the expected Hello number. If both Hello histories are 2406 empty (they contain 0 bits only), the neighbour entry is flushed; 2407 otherwise, the relevant hello timer is reset to the value advertised 2408 in the last Hello of that kind received from this neighbour (no extra 2409 margin is necessary in this case, since jitter was already taken into 2410 account when computing the timeout that has just expired). 2412 A.2. Cost Computation 2414 This section discusses how to compute costs based on Hello history. 2416 A.2.1. k-out-of-j 2418 K-out-of-j link sensing is suitable for wired links that are either 2419 up, in which case they only occasionally drop a packet, or down, in 2420 which case they drop all packets. 2422 The k-out-of-j strategy is parameterised by two small integers k and 2423 j, such that 0 < k <= j, and the nominal link cost, a constant K >= 2424 1. A node keeps a history of the last j hellos; if k or more of 2425 those have been correctly received, the link is assumed to be up, and 2426 the rxcost is set to K; otherwise, the link is assumed to be down, 2427 and the rxcost is set to infinity. 2429 Since Babel supports two kinds of Hellos, a Babel node performs k- 2430 out-of-j twice for each neighbour, once on the Unicast and once on 2431 the Multicast Hello history. If either of the instances of k-out- 2432 of-j indicates that the link is up, then the link is assumed to be 2433 up, and the rxcost is set to K; if both instances indicate that the 2434 link is down, then the link is assumed to be down, and the rxcost is 2435 set to infinity. In other words, the resulting rxcost is the minimum 2436 of the rxcosts yielded by the two instances of k-out-of-j link 2437 sensing. 2439 The cost of a link performing k-out-of-j link sensing is defined as 2440 follows: 2442 o cost = FFFF hexadecimal if rxcost = FFFF hexadecimal; 2444 o cost = txcost otherwise. 2446 A.2.2. ETX 2448 Unlike wired links, which are bimodal (either up or down), wireless 2449 links exhibit continuous variation of the link quality. Naive 2450 application of hop-count routing in networks that use wireless links 2451 for transit tends to select long, lossy links in preference to 2452 shorter, lossless links, which can dramatically reduce throughput. 2454 For that reason, a routing protocol designed to support wireless 2455 links must perform some form of link-quality estimation. 2457 ETX [ETX] is a simple link-quality estimation algorithm that is 2458 designed to work well with the IEEE 802.11 MAC. By default, the 2459 IEEE 802.11 MAC performs ARQ and rate adaptation on unicast frames, 2460 but not on multicast frames, which are sent at a fixed rate with no 2461 ARQ; therefore, measuring the loss rate of multicast frames yields a 2462 useful estimate of a link's quality. 2464 A node performing ETX link quality estimation uses a neighbour's 2465 Multicast Hello history to compute an estimate, written beta, of the 2466 probability that a Hello TLV is successfully received. Beta can be 2467 computed as the fraction of 1 bits within a small number (say, 6) of 2468 the most recent entries in the Multicast Hello history, or it can be 2469 an exponential average, or some combination of both approaches. 2471 Let alpha be MIN(1, 256/txcost), an estimate of the probability of 2472 successfully sending a Hello TLV. The cost is then computed by 2474 cost = 256/(alpha * beta) 2476 or, equivalently, 2478 cost = (MAX(txcost, 256) * rxcost) / 256. 2480 Since the IEEE 802.11 MAC performs ARQ on unicast frames, unicast 2481 frames do not provide a useful measure of link quality, and therefore 2482 ETX ignores the Unicast Hello history. Thus, a node performing ETX 2483 link-quality estimation will not route through neighbouring nodes 2484 unless they send periodic Multicast Hellos (possibly in addition to 2485 Unicast Hellos). 2487 A.3. Metric Computation 2489 As described in Section 3.5.2, the metric advertised by a neighbour 2490 is combined with the link cost to yield a metric. 2492 A.3.1. Additive Metrics 2494 The simplest approach for obtaining a monotonic, left-distributive 2495 metric is to define the metric of a route as the sum of the costs of 2496 the component links. More formally, if a neighbour advertises a 2497 route with metric m over a link with cost c, then the resulting route 2498 has metric M(c, m) = c + m. 2500 A multiplicative metric can be converted into an additive one by 2501 taking the logarithm (in some suitable base) of the link costs. 2503 A.3.2. External Sources of Willingness 2505 A node may want to vary its willingness to forward packets by taking 2506 into account information that is external to the Babel protocol, such 2507 as the monetary cost of a link, the node's battery status, CPU load, 2508 etc. This can be done by adding to every route's metric a value k 2509 that depends on the external data. For example, if a battery-powered 2510 node receives an update with metric m over a link with cost c, it 2511 might compute a metric M(c, m) = k + c + m, where k depends on the 2512 battery status. 2514 In order to preserve strict monotonicity (Section 3.5.2), the value k 2515 must be greater than -c. 2517 Appendix B. Constants 2519 The choice of time constants is a trade-off between fast detection of 2520 mobility events and protocol overhead. Two implementations of Babel 2521 with different time constants will interoperate, although the 2522 resulting convergence time will most likely be dictated by the slower 2523 of the two. 2525 Experience with the sample implementation of Babel indicates that the 2526 Hello interval is the most important time constant: a mobility event 2527 is detected within 1.5 to 3 Hello intervals. Due to Babel's reliance 2528 on triggered updates and explicit requests, the Update interval only 2529 has an effect on the time it takes for accurate metrics to be 2530 propagated after variations in link costs too small to trigger an 2531 unscheduled update or in the presence of packet loss. 2533 At the time of writing, the sample implementation of Babel uses the 2534 following default values: 2536 Multicast Hello Interval: 4 seconds. 2538 IHU Interval: the advertised IHU interval is always 3 times the 2539 Multicast Hello interval. IHUs are actually sent with each Hello 2540 on lossy links (as determined from the Hello history), but only 2541 with every third Multicast Hello on lossless links. 2543 Unicast Hello Interval: the sample implementation never sends 2544 scheduled Unicast Hellos; 2546 Update Interval: 4 times the Multicast Hello interval. 2548 IHU Hold Time: 3.5 times the advertised IHU interval. 2550 Route Expiry Time: 3.5 times the advertised update interval. 2552 Source GC time: 3 minutes. 2554 Request timeout: initially 2 seconds, doubled every time a request 2555 is resent, up to a maximum of three times. 2557 The amount of jitter applied to a packet depends on whether it 2558 contains any urgent TLVs or not (Section 3.1). Urgent triggered 2559 updates and urgent requests are delayed by no more than 200ms; 2560 acknowledgments, by no more than the associated deadline; and other 2561 TLVs by no more than one-half the Multicast Hello interval. 2563 Appendix C. Considerations for protocol extensions 2565 Babel is an extensible protocol, and this document defines a number 2566 of mechanisms that can be used to extend the protocol in a backwards 2567 compatible manner: 2569 o increasing the version number in the packet header; 2571 o defining new TLVs; 2573 o defining new sub-TLVs (with or without the mandatory bit set); 2575 o defining new AEs; 2577 o using the packet trailer. 2579 This appendix is intended to guide designers of protocol extensions 2580 in chosing a particular encoding. 2582 The version number in the Babel header should only be increased if 2583 the new version is not backwards compatible with the original 2584 protocol. 2586 In many cases, an extension could be implemented either by defining a 2587 new TLV, or by adding a new sub-TLV to an existing TLV. For example, 2588 an extension whose purpose is to attach additional data to route 2589 updates can be implemented either by creating a new "enriched" Update 2590 TLV, by adding a non-mandatory sub-TLV to the Update TLV, or by 2591 adding a mandatory sub-TLV. 2593 The various encodings are treated differently by implementations that 2594 do not understand the extension. In the case of a new TLV or of a 2595 sub-TLV with the mandatory bit set, the whole TLV is ignored by 2596 implementations that do not implement the extension, while in the 2597 case of a non-mandatory sub-TLV, the TLV is parsed and acted upon, 2598 and only the unknown sub-TLV is silently ignored. Therefore, a non- 2599 mandatory sub-TLV should be used by extensions that extend the Update 2600 in a compatible manner (the extension data may be silently ignored), 2601 while a mandatory sub-TLV or a new TLV must be used by extensions 2602 that make incompatible extensions to the meaning of the TLV (the 2603 whole TLV must be thrown away if the extension data is not 2604 understood). 2606 Experience shows that the need for additional data tends to crop up 2607 in the most unexpected places. Hence, it is recommended that 2608 extensions that define new TLVs should make them self-terminating, 2609 and allow attaching sub-TLVs to them. 2611 Adding a new AE is essentially equivalent to adding a new TLV: Update 2612 TLVs with an unknown AE are ignored, just like unknown TLVs. 2613 However, adding a new AE is more involved than adding a new TLV, 2614 since it creates a new set of compression state. Additionally, since 2615 the Next Hop TLV creates state specific to a given address family, as 2616 opposed to a given AE, a new AE for a previously defined address 2617 family must not be used in the Next Hop TLV if backwards 2618 compatibility is required. A similar issue arises with Update TLVs 2619 with unknown AEs establishing a new router-id (due to the Router-Id 2620 flag being set). Therefore, defining new AEs must be done with care 2621 if compatibility with unextended implementations is required. 2623 The packet trailer is intended to carry cryptographic signatures that 2624 only cover the packet body; storing the cryptographic signatures in 2625 the packet trailer avoids clearing the signature before computing a 2626 hash of the packet body, and makes it possible to check a 2627 cryptographic signature before running the full, stateful TLV parser. 2628 Thus, any TLV that is allowed to appear in the packet trailer must 2629 not need to be protected by cryptographic security protocols, it 2630 should be easy to parse, and should not require stateful parsing. 2632 Appendix D. Stub Implementations 2634 Babel is a fairly economic protocol. Updates take between 12 and 40 2635 octets per destination, depending on the address family and how 2636 successful compression is; in a double-stack flat network, an average 2637 of less than 24 octets per update is typical. The route table 2638 occupies about 35 octets per IPv6 entry. To put these values into 2639 perspective, a single full-size Ethernet frame can carry some 65 2640 route updates, and a megabyte of memory can contain a 20000-entry 2641 route table and the associated source table. 2643 Babel is also a reasonably simple protocol. The sample 2644 implementation consists of less than 12 000 lines of C code, and it 2645 compiles to less than 120 kB of text on a 32-bit CISC architecture; 2646 about half of this figure is due to protocol extensions and user- 2647 interface code. 2649 Nonetheless, in some very constrained environments, such as PDAs, 2650 microwave ovens, or abacuses, it may be desirable to have subset 2651 implementations of the protocol. 2653 There are many different definitions of a stub router, but for the 2654 needs of this section a stub implementation of Babel is one that 2655 announces one or more directly attached prefixes into a Babel network 2656 but doesn't reannounce any routes that it has learnt from its 2657 neighbours. It may either maintain a full routing table, or simply 2658 select a default gateway amongst any one of its neighbours that 2659 announces a default route. Since a stub implementation never 2660 forwards packets except from or to directly attached links, it cannot 2661 possibly participate in a routing loop, and hence it need not 2662 evaluate the feasibility condition or maintain a source table. 2664 No matter how primitive, a stub implementation MUST parse sub-TLVs 2665 attached to any TLVs that it understands and check the mandatory bit. 2666 It MUST answer acknowledgment requests and MUST participate in the 2667 Hello/IHU protocol. It MUST also be able to reply to seqno requests 2668 for routes that it announces and SHOULD be able to reply to route 2669 requests. 2671 Experience shows that an IPv6-only stub implementation of Babel can 2672 be written in less than 1000 lines of C code and compile to 13 kB of 2673 text on 32-bit CISC architecture. 2675 Appendix E. Software Availability 2677 The sample implementation of Babel is available from 2678 . 2680 Appendix F. Changes from previous versions 2682 F.1. Changes since RFC 6126 2684 o Changed UDP port number to 6696. 2686 o Consistently use router-id rather than id. 2688 o Clarified that the source garbage collection timer is reset after 2689 sending an update even if the entry was not modified. 2691 o In section "Seqno Requests", fixed an erroneous "route request". 2693 o In the description of the Seqno Request TLV, added the description 2694 of the Router-Id field. 2696 o Made router-ids all-0 and all-1 forbidden. 2698 F.2. Changes since draft-ietf-babel-rfc6126bis-00 2700 o Added security considerations. 2702 F.3. Changes since draft-ietf-babel-rfc6126bis-01 2704 o Integrated the format of sub-TLVs. 2706 o Mentioned for each TLV whether it supports sub-TLVs. 2708 o Added Appendix C. 2710 o Added a mandatory bit in sub-TLVs. 2712 o Changed compression state to be per-AF rather than per-AE. 2714 o Added implementation hint for the routing table. 2716 o Clarified how router-ids are computed when bit 0x40 is set in 2717 Updates. 2719 o Relaxed the conditions for sending requests, and tightened the 2720 conditions for forwarding requests. 2722 o Clarified that neighbours should be acquired at some point, but it 2723 doesn't matter when. 2725 F.4. Changes since draft-ietf-babel-rfc6126bis-02 2727 o Added Unicast Hellos. 2729 o Added unscheduled (interval-less) Hellos. 2731 o Changed Appendix A to consider Unicast and unscheduled Hellos. 2733 o Changed Appendix B to agree with the reference implementation. 2735 o Added optional algorithm to avoid the hold time. 2737 o Changed the table of pending seqno requests to be indexed by 2738 router-id in addition to prefixes. 2740 o Relaxed the route acquisition algorithm. 2742 o Replaced minimal implementations by stub implementations. 2744 o Added acknowledgments section. 2746 F.5. Changes since draft-ietf-babel-rfc6126bis-03 2748 o Clarified that all the data structures are conceptual. 2750 o Made sending and receiving Multicast Hellos a SHOULD, avoids 2751 expressing any opinion about Unicast Hellos. 2753 o Removed opinion about Multicast vs. Unicast Hellos (Appendix A.4). 2755 o Made hold-time into a SHOULD rather than MUST. 2757 o Clarified that Seqno Requests are for a finite-metric Update. 2759 o Clarified that sub-TLVs Pad1 and PadN are allowed within any TLV 2760 that allows sub-TLVs. 2762 o Updated IANA Considerations. 2764 o Updated Security Considerations. 2766 o Renamed routing table back to route table. 2768 o Made buffering outgoing updates a SHOULD. 2770 o Weakened advice to use modified EUI-64 in router-ids. 2772 o Added information about sending requests to Appendix B. 2774 o A number of minor wording changes and clarifications. 2776 F.6. Changes since draft-ietf-babel-rfc6126bis-03 2778 Minor editorial changes. 2780 F.7. Changes since draft-ietf-babel-rfc6126bis-04 2782 o Renamed isotonicity to left-distributivity. 2784 o Minor clarifications to unicast hellos. 2786 o Updated requirements boilerplate to RFC 8174. 2788 o Minor editorial changes. 2790 F.8. Changes since draft-ietf-babel-rfc6126bis-05 2792 o Added information about the packet trailer, now that it is used by 2793 draft-ietf-babel-hmac. 2795 Authors' Addresses 2797 Juliusz Chroboczek 2798 IRIF, University of Paris-Diderot 2799 Case 7014 2800 75205 Paris Cedex 13 2801 France 2803 Email: jch@irif.fr 2805 David Schinazi 2806 Apple Inc. 2807 1 Infinite Loop 2808 Cupertino, California 95014 2809 US 2811 Email: dschinazi@apple.com