idnits 2.17.1 draft-ietf-babel-rfc6126bis-11.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 (June 29, 2019) is 1756 days in the past. Is this intentional? Checking references for intended status: Proposed Standard ---------------------------------------------------------------------------- (See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) == Outdated reference: A later version (-10) exists of draft-ietf-babel-dtls-04 == Outdated reference: A later version (-12) exists of draft-ietf-babel-hmac-04 Summary: 0 errors (**), 0 flaws (~~), 4 warnings (==), 2 comments (--).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 Google LLC 6 Expires: December 31, 2019 June 29, 2019 8 The Babel Routing Protocol 9 draft-ietf-babel-rfc6126bis-11 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 December 31, 2019. 35 Copyright Notice 37 Copyright (c) 2019 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 . . . . . . . . . . . . . . 5 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 . . . . . . . . . . . . . . . . . . . . . 13 68 3.3. Acknowledgments and acknowledgment requests . . . . . . . 17 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 . . . . . . . . . . . . . . . . . . . . 28 74 4. Protocol Encoding . . . . . . . . . . . . . . . . . . . . . . 32 75 4.1. Data Types . . . . . . . . . . . . . . . . . . . . . . . 32 76 4.2. Packet Format . . . . . . . . . . . . . . . . . . . . . . 33 77 4.3. TLV Format . . . . . . . . . . . . . . . . . . . . . . . 34 78 4.4. Sub-TLV Format . . . . . . . . . . . . . . . . . . . . . 35 79 4.5. Parser state . . . . . . . . . . . . . . . . . . . . . . 35 80 4.6. Details of Specific TLVs . . . . . . . . . . . . . . . . 36 81 4.7. Details of specific sub-TLVs . . . . . . . . . . . . . . 47 82 5. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 48 83 6. Security Considerations . . . . . . . . . . . . . . . . . . . 49 84 7. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 49 85 8. References . . . . . . . . . . . . . . . . . . . . . . . . . 50 86 8.1. Normative References . . . . . . . . . . . . . . . . . . 50 87 8.2. Informative References . . . . . . . . . . . . . . . . . 50 88 Appendix A. Cost and Metric Computation . . . . . . . . . . . . 51 89 A.1. Maintaining Hello History . . . . . . . . . . . . . . . . 51 90 A.2. Cost Computation . . . . . . . . . . . . . . . . . . . . 52 91 A.3. Metric Computation . . . . . . . . . . . . . . . . . . . 54 92 Appendix B. Constants . . . . . . . . . . . . . . . . . . . . . 54 93 Appendix C. Considerations for protocol extensions . . . . . . . 55 94 Appendix D. Stub Implementations . . . . . . . . . . . . . . . . 57 95 Appendix E. Software Availability . . . . . . . . . . . . . . . 58 96 Appendix F. Changes from previous versions . . . . . . . . . . . 58 97 F.1. Changes since RFC 6126 . . . . . . . . . . . . . . . . . 58 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 . . . . . . 59 101 F.5. Changes since draft-ietf-babel-rfc6126bis-03 . . . . . . 59 102 F.6. Changes since draft-ietf-babel-rfc6126bis-03 . . . . . . 60 103 F.7. Changes since draft-ietf-babel-rfc6126bis-04 . . . . . . 60 104 F.8. Changes since draft-ietf-babel-rfc6126bis-05 . . . . . . 60 105 F.9. Changes since draft-ietf-babel-rfc6126bis-06 . . . . . . 60 106 F.10. Changes since draft-ietf-babel-rfc6126bis-07 . . . . . . 60 107 F.11. Changes since draft-ietf-babel-rfc6126bis-08 . . . . . . 60 108 F.12. Changes since draft-ietf-babel-rfc6126bis-09 . . . . . . 61 109 F.13. Changes since draft-ietf-babel-rfc6126bis-10 . . . . . . 61 110 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 61 112 1. Introduction 114 Babel is a loop-avoiding distance-vector routing protocol that is 115 designed to be robust and efficient both in networks using prefix- 116 based routing and in networks using flat routing ("mesh networks"), 117 and both in relatively stable wired networks and in highly dynamic 118 wireless networks. 120 1.1. Features 122 The main property that makes Babel suitable for unstable networks is 123 that, unlike naive distance-vector routing protocols [RIP], it 124 strongly limits the frequency and duration of routing pathologies 125 such as routing loops and black-holes during reconvergence. Even 126 after a mobility event is detected, a Babel network usually remains 127 loop-free. Babel then quickly reconverges to a configuration that 128 preserves the loop-freedom and connectedness of the network, but is 129 not necessarily optimal; in many cases, this operation requires no 130 packet exchanges at all. Babel then slowly converges, in a time on 131 the scale of minutes, to an optimal configuration. This is achieved 132 by using sequenced routes, a technique pioneered by Destination- 133 Sequenced Distance-Vector routing [DSDV]. 135 More precisely, Babel has the following properties: 137 o when every prefix is originated by at most one router, Babel never 138 suffers from routing loops; 140 o when a single prefix is originated by multiple routers, Babel may 141 occasionally create a transient routing loop for this particular 142 prefix; this loop disappears in a time proportional to its 143 diameter, and never again (up to an arbitrary garbage-collection 144 (GC) time) will the routers involved participate in a routing loop 145 for the same prefix; 147 o assuming bounded packet loss rates, any routing black-holes that 148 may appear after a mobility event are corrected in a time at most 149 proportional to the network's diameter. 151 Babel has provisions for link quality estimation and for fairly 152 arbitrary metrics. When configured suitably, Babel can implement 153 shortest-path routing, or it may use a metric based, for example, on 154 measured packet loss. 156 Babel nodes will successfully establish an association even when they 157 are configured with different parameters. For example, a mobile node 158 that is low on battery may choose to use larger time constants (hello 159 and update intervals, etc.) than a node that has access to wall 160 power. Conversely, a node that detects high levels of mobility may 161 choose to use smaller time constants. The ability to build such 162 heterogeneous networks makes Babel particularly adapted to the 163 unmanaged and wireless environment. 165 Finally, Babel is a hybrid routing protocol, in the sense that it can 166 carry routes for multiple network-layer protocols (IPv4 and IPv6), 167 whichever protocol the Babel packets are themselves being carried 168 over. 170 1.2. Limitations 172 Babel has two limitations that make it unsuitable for use in some 173 environments. First, Babel relies on periodic routing table updates 174 rather than using a reliable transport; hence, in large, stable 175 networks it generates more traffic than protocols that only send 176 updates when the network topology changes. In such networks, 177 protocols such as OSPF [OSPF], IS-IS [IS-IS], or the Enhanced 178 Interior Gateway Routing Protocol (EIGRP) [EIGRP] might be more 179 suitable. 181 Second, unless the optional algorithm described in Section 3.5.5 is 182 implemented, Babel does impose a hold time when a prefix is 183 retracted. While this hold time does not apply to the exact prefix 184 being retracted, and hence does not prevent fast reconvergence should 185 it become available again, it does apply to any shorter prefix that 186 covers it. This may make those implementations of Babel that do not 187 implement the optional algorithm described in Section 3.5.5 188 unsuitable for use in networks that implement automatic prefix 189 aggregation. 191 1.3. Specification of Requirements 193 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 194 "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and 195 "OPTIONAL" in this document are to be interpreted as described in BCP 196 14 [RFC2119] [RFC8174] when, and only when, they appear in all 197 capitals, as shown here. 199 2. Conceptual Description of the Protocol 201 Babel is a loop-avoiding distance vector protocol: it is based on the 202 Bellman-Ford protocol, just like the venerable RIP [RIP], but 203 includes a number of refinements that either prevent loop formation 204 altogether, or ensure that a loop disappears in a timely manner and 205 doesn't form again. 207 Conceptually, Bellman-Ford is executed in parallel for every source 208 of routing information (destination of data traffic). In the 209 following discussion, we fix a source S; the reader will recall that 210 the same algorithm is executed for all sources. 212 2.1. Costs, Metrics and Neighbourship 214 For every pair of neighbouring nodes A and B, Babel computes an 215 abstract value known as the cost of the link from A to B, written 216 C(A, B). Given a route between any two (not necessarily 217 neighbouring) nodes, the metric of the route is the sum of the costs 218 of all the links along the route. The goal of the routing algorithm 219 is to compute, for every source S, the tree of routes of lowest 220 metric to S. 222 Costs and metrics need not be integers. In general, they can be 223 values in any algebra that satisfies two fairly general conditions 224 (Section 3.5.2). 226 A Babel node periodically sends Hello messages to all of its 227 neighbours; it also periodically sends an IHU ("I Heard You") message 228 to every neighbour from which it has recently heard a Hello. From 229 the information derived from Hello and IHU messages received from its 230 neighbour B, a node A computes the cost C(A, B) of the link from A to 231 B. 233 2.2. The Bellman-Ford Algorithm 235 Every node A maintains two pieces of data: its estimated distance to 236 S, written D(A), and its next-hop router to S, written NH(A). 237 Initially, D(S) = 0, D(A) is infinite, and NH(A) is undefined. 239 Periodically, every node B sends to all of its neighbours a route 240 update, a message containing D(B). When a neighbour A of B receives 241 the route update, it checks whether B is its selected next hop; if 242 that is the case, then NH(A) is set to B, and D(A) is set to C(A, B) 243 + D(B). If that is not the case, then A compares C(A, B) + D(B) to 244 its current value of D(A). If that value is smaller, meaning that 245 the received update advertises a route that is better than the 246 currently selected route, then NH(A) is set to B, and D(A) is set to 247 C(A, B) + D(B). 249 A number of refinements to this algorithm are possible, and are used 250 by Babel. In particular, convergence speed may be increased by 251 sending unscheduled "triggered updates" whenever a major change in 252 the topology is detected, in addition to the regular, scheduled 253 updates. Additionally, a node may maintain a number of alternate 254 routes, which are being advertised by neighbours other than its 255 selected neighbour, and which can be used immediately if the selected 256 route were to fail. 258 2.3. Transient Loops in Bellman-Ford 260 It is well known that a naive application of Bellman-Ford to 261 distributed routing can cause transient loops after a topology 262 change. Consider for example the following topology: 264 B 265 1 /| 266 1 / | 267 S --- A |1 268 \ | 269 1 \| 270 C 272 After convergence, D(B) = D(C) = 2, with NH(B) = NH(C) = A. 274 Suppose now that the link between S and A fails: 276 B 277 1 /| 278 / | 279 S A |1 280 \ | 281 1 \| 282 C 284 When it detects the failure of the link, A switches its next hop to B 285 (which is still advertising a route to S with metric 2), and 286 advertises a metric equal to 3, and then advertises a new route with 287 metric 3. This process of nodes changing selected neighbours and 288 increasing their metric continues until the advertised metric reaches 289 "infinity", a value larger than all the metrics that the routing 290 protocol is able to carry. 292 2.4. Feasibility Conditions 294 Bellman-Ford is a very robust algorithm: its convergence properties 295 are preserved when routers delay route acquisition or when they 296 discard some updates. Babel routers discard received route 297 announcements unless they can prove that accepting them cannot 298 possibly cause a routing loop. 300 More formally, we define a condition over route announcements, known 301 as the "feasibility condition", that guarantees the absence of 302 routing loops whenever all routers ignore route updates that do not 303 satisfy the feasibility condition. In effect, this makes Bellman- 304 Ford into a family of routing algorithms, parameterised by the 305 feasibility condition. 307 Many different feasibility conditions are possible. For example, BGP 308 can be modelled as being a distance-vector protocol with a (rather 309 drastic) feasibility condition: a routing update is only accepted 310 when the receiving node's AS number is not included in the update's 311 AS-Path attribute (note that BGP's feasibility condition does not 312 ensure the absence of transient "micro-loops" during reconvergence). 314 Another simple feasibility condition, used in the Destination- 315 Sequenced Distance-Vector (DSDV) routing protocol [DSDV] and in the 316 Ad hoc On-Demand Distance Vector (AODV) protocol, stems from the 317 following observation: a routing loop can only arise after a router 318 has switched to a route with a larger metric than the route that it 319 had previously selected. Hence, one could decide that a route is 320 feasible only when its metric at the local node would be no larger 321 than the metric of the currently selected route, i.e., an 322 announcement carrying a metric D(B) is accepted by A when C(A, B) + 323 D(B) <= D(A). If all routers obey this constraint, then the metric 324 at every router is nonincreasing, and the following invariant is 325 always preserved: if A has selected B as its successor, then D(B) < 326 D(A), which implies that the forwarding graph is loop-free. 328 Babel uses a slightly more refined feasibility condition, derived 329 from EIGRP [DUAL]. Given a router A, define the feasibility distance 330 of A, written FD(A), as the smallest metric that A has ever 331 advertised for S to any of its neighbours. An update sent by a 332 neighbour B of A is feasible when the metric D(B) advertised by B is 333 strictly smaller than A's feasibility distance, i.e., when D(B) < 334 FD(A). 336 It is easy to see that this latter condition is no more restrictive 337 than DSDV-feasibility. Suppose that node A obeys DSDV-feasibility; 338 then D(A) is nonincreasing, hence at all times D(A) <= FD(A). 339 Suppose now that A receives a DSDV-feasible update that advertises a 340 metric D(B). Since the update is DSDV-feasible, C(A, B) + D(B) <= 341 D(A), hence D(B) < D(A), and since D(A) <= FD(A), D(B) < FD(A). 343 To see that it is strictly less restrictive, consider the following 344 diagram, where A has selected the route through B, and D(A) = FD(A) = 345 2. Since D(C) = 1 < FD(A), the alternate route through C is feasible 346 for A, although its metric C(A, C) + D(C) = 5 is larger than that of 347 the currently selected route: 349 B 350 1 / \ 1 351 / \ 352 S A 353 \ / 354 1 \ / 4 355 C 357 To show that this feasibility condition still guarantees loop- 358 freedom, recall that at the time when A accepts an update from B, the 359 metric D(B) announced by B is no smaller than FD(B); since it is 360 smaller than FD(A), at that point in time FD(B) < FD(A). Since this 361 property is preserved when A sends updates, it remains true at all 362 times, which ensures that the forwarding graph has no loops. 364 2.5. Solving Starvation: Sequencing Routes 366 Obviously, the feasibility conditions defined above cause starvation 367 when a router runs out of feasible routes. Consider the following 368 diagram, where both A and B have selected the direct route to S: 370 A 371 1 /| D(A) = 1 372 / | FD(A) = 1 373 S |1 374 \ | D(B) = 2 375 2 \| FD(B) = 2 376 B 378 Suppose now that the link between A and S breaks: 380 A 381 | 382 | FD(A) = 1 383 S |1 384 \ | D(B) = 2 385 2 \| FD(B) = 2 386 B 388 The only route available from A to S, the one that goes through B, is 389 not feasible: A suffers from spurious starvation. At that point, the 390 whole subtree suffering from starvation must be reset, which is 391 essentially what EIGRP does when it performs a global synchronisation 392 of all the routers in the starving subtree (the "active" phase of 393 EIGRP). 395 Babel reacts to starvation in a less drastic manner, by using 396 sequenced routes, a technique introduced by DSDV and adopted by AODV. 397 In addition to a metric, every route carries a sequence number, a 398 nondecreasing integer that is propagated unchanged through the 399 network and is only ever incremented by the source; a pair (s, m), 400 where s is a sequence number and m a metric, is called a distance. 402 A received update is feasible when either it is more recent than the 403 feasibility distance maintained by the receiving node, or it is 404 equally recent and the metric is strictly smaller. More formally, if 405 FD(A) = (s, m), then an update carrying the distance (s', m') is 406 feasible when either s' > s, or s = s' and m' < m. 408 Assuming the sequence number of S is 137, the diagram above becomes: 410 A 411 | 412 | FD(A) = (137, 1) 413 S |1 414 \ | D(B) = (137, 2) 415 2 \| FD(B) = (137, 2) 416 B 418 After S increases its sequence number, and the new sequence number is 419 propagated to B, we have: 421 A 422 | 423 | FD(A) = (137, 1) 424 S |1 425 \ | D(B) = (138, 2) 426 2 \| FD(B) = (138, 2) 427 B 429 at which point the route through B becomes feasible again. 431 Note that while sequence numbers are used for determining 432 feasibility, they are not used in route selection: a node ignores the 433 sequence number when selecting the best route to a given destination 434 (Section 3.6). Doing otherwise would cause route oscillation while a 435 sequence number propagates through the network, and might even cause 436 persistent blackholes with some exotic metrics. 438 2.6. Requests 440 In DSDV, the sequence number of a source is increased periodically. 441 A route becomes feasible again after the source increases its 442 sequence number, and the new sequence number is propagated through 443 the network, which may, in general, require a significant amount of 444 time. 446 Babel takes a different approach. When a node detects that it is 447 suffering from a potentially spurious starvation, it sends an 448 explicit request to the source for a new sequence number. This 449 request is forwarded hop by hop to the source, with no regard to the 450 feasibility condition. Upon receiving the request, the source 451 increases its sequence number and broadcasts an update, which is 452 forwarded to the requesting node. 454 Note that after a change in network topology not all such requests 455 will, in general, reach the source, as some will be sent over links 456 that are now broken. However, if the network is still connected, 457 then at least one among the nodes suffering from spurious starvation 458 has an (unfeasible) route to the source; hence, in the absence of 459 packet loss, at least one such request will reach the source. 460 (Resending requests a small number of times compensates for packet 461 loss.) 463 Since requests are forwarded with no regard to the feasibility 464 condition, they may, in general, be caught in a forwarding loop; this 465 is avoided by having nodes perform duplicate detection for the 466 requests that they forward. 468 2.7. Multiple Routers 470 The above discussion assumes that each prefix is originated by a 471 single router. In real networks, however, it is often necessary to 472 have a single prefix originated by multiple routers: for example, the 473 default route will be originated by all of the edge routers of a 474 routing domain. 476 Since synchronising sequence numbers between distinct routers is 477 problematic, Babel treats routes for the same prefix as distinct 478 entities when they are originated by different routers: every route 479 announcement carries the router-id of its originating router, and 480 feasibility distances are not maintained per prefix, but per source, 481 where a source is a pair of a router-id and a prefix. In effect, 482 Babel guarantees loop-freedom for the forwarding graph to every 483 source; since the union of multiple acyclic graphs is not in general 484 acyclic, Babel does not in general guarantee loop-freedom when a 485 prefix is originated by multiple routers, but any loops will be 486 broken in a time at most proportional to the diameter of the loop -- 487 as soon as an update has "gone around" the routing loop. 489 Consider for example the following topology, where A has selected the 490 default route through S, and B has selected the one through S': 492 1 1 1 493 ::/0 -- S --- A --- B --- S' -- ::/0 495 Suppose that both default routes fail at the same time; then nothing 496 prevents A from switching to B, and B simultaneously switching to A. 497 However, as soon as A has successfully advertised the new route to B, 498 the route through A will become unfeasible for B. Conversely, as 499 soon as B will have advertised the route through A, the route through 500 B will become unfeasible for A. 502 In effect, the routing loop disappears at the latest when routing 503 information has gone around the loop. Since this process can be 504 delayed by lost packets, Babel makes certain efforts to ensure that 505 updates are sent reliably after a router-id change (Section 3.7.2). 507 Additionally, after the routers have advertised the two routes, both 508 sources will be in their source tables, which will prevent them from 509 ever again participating in a routing loop involving routes from S 510 and S' (up to the source GC time, which, available memory permitting, 511 can be set to arbitrarily large values). 513 2.8. Overlapping Prefixes 515 In the above discussion, we have assumed that all prefixes are 516 disjoint, as is the case in flat ("mesh") routing. In practice, 517 however, prefixes may overlap: for example, the default route 518 overlaps with all of the routes present in the network. 520 After a route fails, it is not correct in general to switch to a 521 route that subsumes the failed route. Consider for example the 522 following configuration: 524 1 1 525 ::/0 -- A --- B --- C 527 Suppose that node C fails. If B forwards packets destined to C by 528 following the default route, a routing loop will form, and persist 529 until A learns of B's retraction of the direct route to C. B avoids 530 this pitfall by installing an "unreachable" route after a route is 531 retracted; this route is maintained until it can be guaranteed that 532 the former route has been retracted by all of B's neighbours 533 (Section 3.5.5). 535 3. Protocol Operation 537 Every Babel speaker is assigned a router-id, which is an arbitrary 538 string of 8 octets that is assumed unique across the routing domain. 539 For example, router-ids could be assigned randomly, or they could be 540 derived from a link-layer address. (The protocol encoding is 541 slightly more compact when router-ids are assigned in the same manner 542 as the IPv6 layer assigns host IDs.) 544 3.1. Message Transmission and Reception 546 Babel protocol packets are sent in the body of a UDP datagram (as 547 described in Section 4 below). Each Babel packet consists of zero or 548 more TLVs. Most TLVs may contain sub-TLVs. 550 The source address of a Babel packet is always a unicast address, 551 link-local in the case of IPv6. Babel packets may be sent to a well- 552 known (link-local) multicast address or to a (link-local) unicast 553 address. In normal operation, a Babel speaker sends both multicast 554 and unicast packets to its neighbours. 556 With the exception of acknowledgments, all Babel TLVs can be sent to 557 either unicast or multicast addresses, and their semantics does not 558 depend on whether the destination is a unicast or a multicast 559 address. Hence, a Babel speaker does not need to determine the 560 destination address of a packet that it receives in order to 561 interpret it. 563 A moderate amount of jitter may be applied to packets sent by a Babel 564 speaker: outgoing TLVs are buffered and SHOULD be sent with a small 565 random delay. This is done for two purposes: it avoids 566 synchronisation of multiple Babel speakers across a network [JITTER], 567 and it allows for the aggregation of multiple TLVs into a single 568 packet. 570 The exact delay and amount of jitter applied to a packet depends on 571 whether it contains any urgent TLVs. Acknowledgment TLVs MUST be 572 sent before the deadline specified in the corresponding request. The 573 particular class of updates specified in Section 3.7.2 MUST be sent 574 in a timely manner. The particular class of request and update TLVs 575 specified in Section 3.8.2 SHOULD be sent in a timely manner. 577 3.2. Data Structures 579 In this section, we give a description of the data structures that 580 every Babel speaker maintains. This description is conceptual: a 581 Babel speaker may use different data structures as long as the 582 resulting protocol is the same as the one described in this document. 583 For example, rather than maintaining a single table containing both 584 selected and unselected (fallback) routes, as described in 585 Section 3.2.6 below, an actual implementation would probably use two 586 tables, one with selected routes and one with fallback routes. 588 3.2.1. Sequence number arithmetic 590 Sequence numbers (seqnos) appear in a number of Babel data 591 structures, and they are interpreted as integers modulo 2^16. For 592 the purposes of this document, arithmetic on sequence numbers is 593 defined as follows. 595 Given a seqno s and an integer n, the sum of s and n is defined by 597 s + n (modulo 2^16) = (s + n) MOD 2^16 599 or, equivalently, 601 s + n (modulo 2^16) = (s + n) AND 65535 603 where MOD is the modulo operation yielding a non-negative integer and 604 AND is the bitwise conjunction operation. 606 Given two sequence numbers s and s', the relation s is less than s' 607 (s < s') is defined by 609 s < s' (modulo 2^16) when 0 < ((s' - s) MOD 2^16) < 32768 611 or equivalently 613 s < s' (modulo 2^16) when s /= s' and ((s' - s) AND 32768) = 0. 615 3.2.2. Node Sequence Number 617 A node's sequence number is a 16-bit integer that is included in 618 route updates sent for routes originated by this node. 620 A node increments its sequence number (modulo 2^16) whenever it 621 receives a request for a new sequence number (Section 3.8.1.2). A 622 node SHOULD NOT increment its sequence number (seqno) spontaneously, 623 since increasing seqnos makes it less likely that other nodes will 624 have feasible alternate routes when their selected routes fail. 626 3.2.3. The Interface Table 628 The interface table contains the list of interfaces on which the node 629 speaks the Babel protocol. Every interface table entry contains the 630 interface's outgoing Multicast Hello seqno, a 16-bit integer that is 631 sent with each Multicast Hello TLV on this interface and is 632 incremented (modulo 2^16) whenever a Multicast Hello is sent. (Note 633 that an interface's Multicast Hello seqno is unrelated to the node's 634 seqno.) 636 There are two timers associated with each interface table entry -- 637 the multicast hello timer, which governs the sending of scheduled 638 Multicast Hello and IHU packets, and the update timer, which governs 639 the sending of periodic route updates. 641 3.2.4. The Neighbour Table 643 The neighbour table contains the list of all neighbouring interfaces 644 from which a Babel packet has been recently received. The neighbour 645 table is indexed by pairs of the form (interface, address), and every 646 neighbour table entry contains the following data: 648 o the local node's interface over which this neighbour is reachable; 650 o the address of the neighbouring interface; 652 o a history of recently received Multicast Hello packets from this 653 neighbour; this can, for example, be a sequence of n bits, for 654 some small value n, indicating which of the n hellos most recently 655 sent by this neighbour have been received by the local node; 657 o a history of recently received Unicast Hello packets from this 658 neighbour; 660 o the "transmission cost" value from the last IHU packet received 661 from this neighbour, or FFFF hexadecimal (infinity) if the IHU 662 hold timer for this neighbour has expired; 664 o the expected incoming Multicast Hello sequence number for this 665 neighbour, an integer modulo 2^16. 667 o the expected incoming Unicast Hello sequence number for this 668 neighbour, an integer modulo 2^16. 670 o the outgoing Unicast Hello sequence number for this neighbour, an 671 integer modulo 2^16 that is sent with each Unicast Hello TLV to 672 this neighbour and is incremented (modulo 2^16) whenever a Unicast 673 Hello is sent. (Note that the outgoing Unicast Hello seqno for a 674 neighbour is distinct from the interface's outgoing Multicast 675 Hello seqno.) 677 There are three timers associated with each neighbour entry -- the 678 multicast hello timer, which is initialised from the interval value 679 carried by scheduled Multicast Hello TLVs, the unicast hello timer, 680 which is initialised from the interval value carried by scheduled 681 Unicast Hello TLVs, and the IHU timer, which is initialised to a 682 small multiple of the interval carried in IHU TLVs. 684 Note that the neighbour table is indexed by IP addresses, not by 685 router-ids: neighbourship is a relationship between interfaces, not 686 between nodes. Therefore, two nodes with multiple interfaces can 687 participate in multiple neighbourship relationships, a situation that 688 can notably arise when wireless nodes with multiple radios are 689 involved. 691 3.2.5. The Source Table 693 The source table is used to record feasibility distances. It is 694 indexed by triples of the form (prefix, plen, router-id), and every 695 source table entry contains the following data: 697 o the prefix (prefix, plen), where plen is the prefix length, that 698 this entry applies to; 700 o the router-id of a router originating this prefix; 702 o a pair (seqno, metric), this source's feasibility distance. 704 There is one timer associated with each entry in the source table -- 705 the source garbage-collection timer. It is initialised to a time on 706 the order of minutes and reset as specified in Section 3.7.3. 708 3.2.6. The Route Table 710 The route table contains the routes known to this node. It is 711 indexed by triples of the form (prefix, plen, neighbour), and every 712 route table entry contains the following data: 714 o the source (prefix, plen, router-id) for which this route is 715 advertised; 717 o the neighbour that advertised this route; 719 o the metric with which this route was advertised by the neighbour, 720 or FFFF hexadecimal (infinity) for a recently retracted route; 722 o the sequence number with which this route was advertised; 724 o the next-hop address of this route; 726 o a boolean flag indicating whether this route is selected, i.e., 727 whether it is currently being used for forwarding and is being 728 advertised. 730 There is one timer associated with each route table entry -- the 731 route expiry timer. It is initialised and reset as specified in 732 Section 3.5.4. 734 Note that there are two distinct (seqno, metric) pairs associated to 735 each route: the route's distance, which is stored in the route table, 736 and the feasibility distance, stored in the source table and shared 737 between all routes with the same source. 739 3.2.7. The Table of Pending Seqno Requests 741 The table of pending seqno requests contains a list of seqno requests 742 that the local node has sent (either because they have been 743 originated locally, or because they were forwarded) and to which no 744 reply has been received yet. This table is indexed by triples of the 745 form (prefix, plen, router-id), and every entry in this table 746 contains the following data: 748 o the prefix, plen, router-id, and seqno being requested; 750 o the neighbour, if any, on behalf of which we are forwarding this 751 request; 753 o a small integer indicating the number of times that this request 754 will be resent if it remains unsatisfied. 756 There is one timer associated with each pending seqno request; it 757 governs both the resending of requests and their expiry. 759 3.3. Acknowledgments and acknowledgment requests 761 A Babel speaker may request that a neighbour receiving a given packet 762 reply with an explicit acknowledgment within a given time. While the 763 use of acknowledgment requests is optional, every Babel speaker MUST 764 be able to reply to such a request. 766 An acknowledgment MUST be sent to a unicast destination. On the 767 other hand, acknowledgment requests may be sent to either unicast or 768 multicast destinations, in which case they request an acknowledgment 769 from all of the receiving nodes. 771 When to request acknowledgments is a matter of local policy; the 772 simplest strategy is to never request acknowledgments and to rely on 773 periodic updates to ensure that any reachable routes are eventually 774 propagated throughout the routing domain. In order to improve 775 convergence speed and reduce the amount of control traffic, 776 acknowledgment requests MAY be used in order to reliably send urgent 777 updates (Section 3.7.2) and retractions (Section 3.5.5), especially 778 when the number of neighbours on a given interface is small. Since 779 Babel is designed to deal gracefully with packet loss on unreliable 780 media, sending all packets with acknowledgment requests is not 781 necessary, and NOT RECOMMENDED, as the acknowledgments cause 782 additional traffic and may force additional Address Resolution 783 Protocol (ARP) or Neighbour Discovery (ND) exchanges. 785 3.4. Neighbour Acquisition 787 Neighbour acquisition is the process by which a Babel node discovers 788 the set of neighbours heard over each of its interfaces and 789 ascertains bidirectional reachability. On unreliable media, 790 neighbour acquisition additionally provides some statistics that may 791 be useful for link quality computation. 793 Before it can exchange routing information with a neighbour, a Babel 794 node MUST create an entry for that neighbour in the neighbour table. 795 When to do that is implementation-specific; suitable strategies 796 include creating an entry when any Babel packet is received, or 797 creating an entry when a Hello TLV is parsed. Similarly, in order to 798 conserve system resources, an implementation SHOULD discard an entry 799 when it has been unused for long enough; suitable strategies include 800 dropping the neighbour after a timeout, and dropping a neighbour when 801 the associated Hello histories become empty (see Appendix A.2). 803 3.4.1. Reverse Reachability Detection 805 Every Babel node sends Hello TLVs to its neighbours to indicate that 806 it is alive, at regular or irregular intervals. Each Hello TLV 807 carries an increasing (modulo 2^16) sequence number and an upper 808 bound on the time interval until the next Hello of the same type (see 809 below). If the time interval is set to 0, then the Hello TLV does 810 not establish a new promise: the deadline carried by the previous 811 Hello of the same type still applies to the next Hello (if the most 812 recent scheduled Hello of the right kind was received at time t0 and 813 carried interval i, then the previous promise of sending another 814 Hello before time t0 + i still holds). We say that a Hello is 815 "scheduled" if it carries a non-zero interval, and "unscheduled" 816 otherwise. 818 There are two kinds of Hellos: Multicast Hellos, which use a per- 819 interface Hello counter (the Multicast Hello seqno), and Unicast 820 Hellos, which use a per-neighbour counter (the Unicast Hello seqno). 821 A Multicast Hello with a given seqno MUST be sent to all neighbours 822 on a given interface, either by sending it to a multicast address or 823 by sending it to one unicast address per neighbour (hence, the term 824 "Multicast Hello" is a slight misnomer). A Unicast Hello carrying a 825 given seqno should normally be sent to just one neighbour (over 826 unicast), since the sequence numbers of different neighbours are not 827 in general synchronised. 829 Multicast Hellos sent over multicast can be used for neighbour 830 discovery; hence, a node SHOULD send periodic (scheduled) Multicast 831 Hellos unless neighbour discovery is performed by means outside of 832 the Babel protocol. A node MAY send Unicast Hellos or unscheduled 833 Hellos of either kind for any reason, such as reducing the amount of 834 multicast traffic or improving reliability on link technologies with 835 poor support for link-layer multicast. 837 A node MAY send a scheduled Hello ahead of time. A node MAY change 838 its scheduled Hello interval. The Hello interval MAY be decreased at 839 any time; it MAY be increased immediately before sending a Hello TLV, 840 but SHOULD NOT be increased at other times. (Equivalently, a node 841 SHOULD send a scheduled Hello immediately after increasing its Hello 842 interval.) 844 How to deal with received Hello TLVs and what statistics to maintain 845 are considered local implementation matters; typically, a node will 846 maintain some sort of history of recently received Hellos. An 847 example of a suitable algorithm is described in Appendix A.1. 849 After receiving a Hello, or determining that it has missed one, the 850 node recomputes the association's cost (Section 3.4.3) and runs the 851 route selection procedure (Section 3.6). 853 3.4.2. Bidirectional Reachability Detection 855 In order to establish bidirectional reachability, every node sends 856 periodic IHU ("I Heard You") TLVs to each of its neighbours. Since 857 IHUs carry an explicit interval value, they MAY be sent less often 858 than Hellos in order to reduce the amount of routing traffic in dense 859 networks; in particular, they SHOULD be sent less often than Hellos 860 over links with little packet loss. While IHUs are conceptually 861 unicast, they MAY be sent to a multicast address in order to avoid an 862 ARP or Neighbour Discovery exchange and to aggregate multiple IHUs 863 into a single packet. 865 In addition to the periodic IHUs, a node MAY, at any time, send an 866 unscheduled IHU packet. It MAY also, at any time, decrease its IHU 867 interval, and it MAY increase its IHU interval immediately before 868 sending an IHU, but SHOULD NOT increase it at any other time. 869 (Equivalently, a node SHOULD send an extra IHU immediately after 870 increasing its Hello interval.) 872 Every IHU TLV contains two pieces of data: the link's rxcost 873 (reception cost) from the sender's perspective, used by the neighbour 874 for computing link costs (Section 3.4.3), and the interval between 875 periodic IHU packets. A node receiving an IHU sets the value of the 876 txcost (transmission cost) maintained in the neighbour table to the 877 value contained in the IHU, and resets the IHU timer associated to 878 this neighbour to a small multiple of the interval value received in 879 the IHU. When a neighbour's IHU timer expires, the neighbour's 880 txcost is set to infinity. 882 After updating a neighbour's txcost, the receiving node recomputes 883 the neighbour's cost (Section 3.4.3) and runs the route selection 884 procedure (Section 3.6). 886 3.4.3. Cost Computation 888 A neighbourship association's link cost is computed from the values 889 maintained in the neighbour table: the statistics kept in the 890 neighbour table about the reception of Hellos, and the txcost 891 computed from received IHU packets. 893 For every neighbour, a Babel node computes a value known as this 894 neighbour's rxcost. This value is usually derived from the Hello 895 history, which may be combined with other data, such as statistics 896 maintained by the link layer. The rxcost is sent to a neighbour in 897 each IHU. 899 Since nodes do not necessarily send periodic Unicast Hellos but do 900 usually send periodic Multicast Hellos (Section 3.4.1), a node SHOULD 901 use an algorithm that yields a finite rxcost when only Multicast 902 Hellos are received, unless interoperability with nodes that only 903 send Multicast Hellos is not required. 905 How the txcost and rxcost are combined in order to compute a link's 906 cost is a matter of local policy; as far as Babel's correctness is 907 concerned, only the following conditions MUST be satisfied: 909 o the cost is strictly positive; 911 o if no Hello TLVs of either kind were received recently, then the 912 cost is infinite; 914 o if the txcost is infinite, then the cost is infinite. 916 Note that while this document does not constrain cost computation any 917 further, not all cost computation strategies will give good results. 918 See Appendix A.2 for examples of strategies for computing a link's 919 cost that are known to work well in practice. 921 3.5. Routing Table Maintenance 923 Conceptually, a Babel update is a quintuple (prefix, plen, router-id, 924 seqno, metric), where (prefix, plen) is the prefix for which a route 925 is being advertised, router-id is the router-id of the router 926 originating this update, seqno is a nondecreasing (modulo 2^16) 927 integer that carries the originating router seqno, and metric is the 928 announced metric. 930 Before being accepted, an update is checked against the feasibility 931 condition (Section 3.5.1), which ensures that the route does not 932 create a routing loop. If the feasibility condition is not 933 satisfied, the update is either ignored or prevents the route from 934 being selected, as described in Section 3.5.4. If the feasibility 935 condition is satisfied, then the update cannot possibly cause a 936 routing loop. 938 3.5.1. The Feasibility Condition 940 The feasibility condition is applied to all received updates. The 941 feasibility condition compares the metric in the received update with 942 the metrics of the updates previously sent by the receiving node; 943 updates that fail the feasibility condition, and therefore have 944 metrics large enough to cause a routing loop, are either ignored or 945 prevent the resulting route from being selected. 947 A feasibility distance is a pair (seqno, metric), where seqno is an 948 integer modulo 2^16 and metric is a positive integer. Feasibility 949 distances are compared lexicographically, with the first component 950 inverted: we say that a distance (seqno, metric) is strictly better 951 than a distance (seqno', metric'), written 953 (seqno, metric) < (seqno', metric') 955 when 957 seqno > seqno' or (seqno = seqno' and metric < metric') 959 where sequence numbers are compared modulo 2^16. 961 Given a source (prefix, plen, router-id), a node's feasibility 962 distance for this source is the minimum, according to the ordering 963 defined above, of the distances of all the finite updates ever sent 964 by this particular node for the prefix (prefix, plen) and the given 965 router-id. Feasibility distances are maintained in the source table, 966 the exact procedure is given in Section 3.7.3. 968 A received update is feasible when either it is a retraction (its 969 metric is FFFF hexadecimal), or the advertised distance is strictly 970 better, in the sense defined above, than the feasibility distance for 971 the corresponding source. More precisely, a route advertisement 972 carrying the quintuple (prefix, plen, router-id, seqno, metric) is 973 feasible if one of the following conditions holds: 975 o metric is infinite; or 977 o no entry exists in the source table indexed by (prefix, plen, 978 router-id); or 980 o an entry (prefix, plen, router-id, seqno', metric') exists in the 981 source table, and either 983 * seqno' < seqno or 985 * seqno = seqno' and metric < metric'. 987 Note that the feasibility condition considers the metric advertised 988 by the neighbour, not the route's metric; hence, a fluctuation in a 989 neighbour's cost cannot render a selected route unfeasible. Note 990 further that retractions (updates with infinite metric) are always 991 feasible, since they cannot possibly cause a routing loop. 993 3.5.2. Metric Computation 995 A route's metric is computed from the metric advertised by the 996 neighbour and the neighbour's link cost. Just like cost computation, 997 metric computation is considered a local policy matter; as far as 998 Babel is concerned, the function M(c, m) used for computing a metric 999 from a locally computed link cost and the metric advertised by a 1000 neighbour MUST only satisfy the following conditions: 1002 o if c is infinite, then M(c, m) is infinite; 1004 o M is strictly monotonic: M(c, m) > m. 1006 Additionally, the metric SHOULD satisfy the following condition: 1008 o M is left-distributive: if m <= m', then M(c, m) <= M(c, m'). 1010 Note that while strict monotonicity is essential to the integrity of 1011 the network (persistent routing loops may arise if it is not 1012 satisfied), left distributivity is not: if it is not satisfied, Babel 1013 will still converge to a loop-free configuration, but might not reach 1014 a global optimum (in fact, a global optimum may not even exist). 1016 As with cost computation, not all strategies for computing route 1017 metrics will give good results. In particular, some metrics are more 1018 likely than others to lead to routing instabilities (route flapping). 1019 In Appendix A.3, we give a number of examples of strictly monotonic, 1020 left-distributive routing metrics that are known to work well in 1021 practice. 1023 3.5.3. Encoding of Updates 1025 In a large network, the bulk of Babel traffic consists of route 1026 updates; hence, some care has been given to encoding them 1027 efficiently. An Update TLV itself only contains the prefix, seqno, 1028 and metric, while the next hop is derived either from the network- 1029 layer source address of the packet or from an explicit Next Hop TLV 1030 in the same packet. The router-id is derived from a separate Router- 1031 Id TLV in the same packet, which optimises the case when multiple 1032 updates are sent with the same router-id. 1034 Additionally, a prefix of the advertised prefix can be omitted in an 1035 Update TLV, in which case it is copied from a previous Update TLV in 1036 the same packet -- this is known as address compression 1037 (Section 4.6.9). 1039 Finally, as a special optimisation for the case when a router-id 1040 coincides with the interface-id part of an IPv6 address, the router- 1041 id can optionally be derived from the low-order bits of the 1042 advertised prefix. 1044 The encoding of updates is described in detail in Section 4.6. 1046 3.5.4. Route Acquisition 1048 When a Babel node receives an update (prefix, plen, router-id, seqno, 1049 metric) from a neighbour neigh with a link cost value equal to cost, 1050 it checks whether it already has a route table entry indexed by 1051 (prefix, plen, neigh). 1053 If no such entry exists: 1055 o if the update is unfeasible, it MAY be ignored; 1057 o if the metric is infinite (the update is a retraction of a route 1058 we do not know about), the update is ignored; 1060 o otherwise, a new entry is created in the route table, indexed by 1061 (prefix, plen, neigh), with source equal to (prefix, plen, router- 1062 id), seqno equal to seqno and an advertised metric equal to the 1063 metric carried by the update. 1065 If such an entry exists: 1067 o if the entry is currently selected, the update is unfeasible, and 1068 the router-id of the update is equal to the router-id of the 1069 entry, then the update MAY be ignored; 1071 o otherwise, the entry's sequence number, advertised metric, metric, 1072 and router-id are updated and, if the advertised metric is not 1073 infinite, the route's expiry timer is reset to a small multiple of 1074 the Interval value included in the update. If the update is 1075 unfeasible, then the (now unfeasible) entry MUST be immediately 1076 unselected. If the update caused the router-id of the entry to 1077 change, an update (possibly a retraction) MUST be sent in a timely 1078 manner (see Section 3.7.2). 1080 Note that the route table may contain unfeasible routes, either 1081 because they were created by an unfeasible update or due to a metric 1082 fluctuation. Such routes are never selected, since they are not 1083 known to be loop-free; should all the feasible routes become 1084 unusable, however, the unfeasible routes can be made feasible and 1085 therefore possible to select by sending requests along them (see 1086 Section 3.8.2). 1088 When a route's expiry timer triggers, the behaviour depends on 1089 whether the route's metric is finite. If the metric is finite, it is 1090 set to infinity and the expiry timer is reset. If the metric is 1091 already infinite, the route is flushed from the route table. 1093 After the route table is updated, the route selection procedure 1094 (Section 3.6) is run. 1096 3.5.5. Hold Time 1098 When a prefix P is retracted, because all routes are unfeasible or 1099 have an infinite metric (whether due to the expiry timer or to other 1100 reasons), and a shorter prefix P' that covers P is reachable, P' 1101 cannot in general be used for routing packets destined to P without 1102 running the risk of creating a routing loop (Section 2.8). 1104 To avoid this issue, whenever a prefix P is retracted, a route table 1105 entry with infinite metric is maintained as described in 1106 Section 3.5.4 above. As long as this entry is maintained, packets 1107 destined to an address within P MUST NOT be forwarded by following a 1108 route for a shorter prefix. This entry is removed as soon as a 1109 finite-metric update for prefix P is received and the resulting route 1110 selected. If no such update is forthcoming, the infinite metric 1111 entry SHOULD be maintained at least until it is guaranteed that no 1112 neighbour has selected the current node as next-hop for prefix P. 1113 This can be achieved by either: 1115 o waiting until the route's expiry timer has expired 1116 (Section 3.5.4); 1118 o sending a retraction with an acknowledgment request (Section 3.3) 1119 to every reachable neighbour that has not explicitly retracted 1120 prefix P and waiting for all acknowledgments. 1122 The former option is simpler and ensures that at that point, any 1123 routes for prefix P pointing at the current node have expired. 1124 However, since the expiry time can be as high as a few minutes, doing 1125 that prevents automatic aggregation by creating spurious black-holes 1126 for aggregated routes. The latter option is RECOMMENDED as it 1127 dramatically reduces the time for which a prefix is unreachable in 1128 the presence of aggregated routes. 1130 3.6. Route Selection 1132 Route selection is the process by which a single route for a given 1133 prefix is selected to be used for forwarding packets and to be re- 1134 advertised to a node's neighbours. 1136 Babel is designed to allow flexible route selection policies. As far 1137 as the protocol's correctness is concerned, the route selection 1138 policy MUST only satisfy the following properties: 1140 o a route with infinite metric (a retracted route) is never 1141 selected; 1143 o an unfeasible route is never selected. 1145 Note, however, that Babel does not naturally guarantee the stability 1146 of routing, and configuring conflicting route selection policies on 1147 different routers may lead to persistent route oscillation. 1149 Route selection is a difficult problem, since a good route selection 1150 policy needs to take into account multiple mutually contradictory 1151 criteria; in roughly decreasing order of importance, these are: 1153 o routes with a small metric should be preferred to routes with a 1154 large metric; 1156 o switching router-ids should be avoided; 1158 o routes through stable neighbours should be preferred to routes 1159 through unstable ones; 1161 o stable routes should be preferred to unstable ones; 1163 o switching next hops should be avoided. 1165 Route selection MUST NOT take seqnos into account: a route MUST NOT 1166 be preferred just because it carries a higher (more recent) seqno. 1167 Doing otherwise would cause route oscillation while a new seqno 1168 propagates through the network, possibly following multiple paths of 1169 different latency, and might even create persistent blackholes if the 1170 metric being used is not left-distributive Section 3.5.2. 1172 A simple but useful strategy is to choose the feasible route with the 1173 smallest metric, with a small amount of hysteresis applied to avoid 1174 switching router-ids too often. 1176 After the route selection procedure is run, triggered updates 1177 (Section 3.7.2) and requests (Section 3.8.2) are sent. 1179 3.7. Sending Updates 1181 A Babel speaker advertises to its neighbours its set of selected 1182 routes. Normally, this is done by sending one or more multicast 1183 packets containing Update TLVs on all of its connected interfaces; 1184 however, on link technologies where multicast is significantly more 1185 expensive than unicast, a node MAY choose to send multiple copies of 1186 updates in unicast packets, especially when the number of neighbours 1187 is small. 1189 Additionally, in order to ensure that any black-holes are reliably 1190 cleared in a timely manner, a Babel node sends retractions (updates 1191 with an infinite metric) for any recently retracted prefixes. 1193 If an update is for a route injected into the Babel domain by the 1194 local node (e.g., it carries the address of a local interface, the 1195 prefix of a directly attached network, or a prefix redistributed from 1196 a different routing protocol), the router-id is set to the local 1197 node's router-id, the metric is set to some arbitrary finite value 1198 (typically 0), and the seqno is set to the local router's sequence 1199 number. 1201 If an update is for a route learned from another Babel speaker, the 1202 router-id and sequence number are copied from the route table entry, 1203 and the metric is computed as specified in Section 3.5.2. 1205 3.7.1. Periodic Updates 1207 Every Babel speaker periodically advertises all of its selected 1208 routes on all of its interfaces, including any recently retracted 1209 routes. Since Babel doesn't suffer from routing loops (there is no 1210 "counting to infinity") and relies heavily on triggered updates 1211 (Section 3.7.2), this full dump only needs to happen infrequently. 1213 3.7.2. Triggered Updates 1215 In addition to periodic routing updates, a Babel speaker sends 1216 unscheduled, or triggered, updates in order to inform its neighbours 1217 of a significant change in the network topology. 1219 A change of router-id for the selected route to a given prefix may be 1220 indicative of a routing loop in formation; hence, a node MUST send a 1221 triggered update in a timely manner whenever it changes the selected 1222 router-id for a given destination. Additionally, it SHOULD make a 1223 reasonable attempt at ensuring that all reachable neighbours receive 1224 this update. 1226 There are two strategies for ensuring that. If the number of 1227 neighbours is small, then it is reasonable to send the update 1228 together with an acknowledgment request; the update is resent until 1229 all neighbours have acknowledged the packet, up to some number of 1230 times. If the number of neighbours is large, however, requesting 1231 acknowledgments from all of them might cause a non-negligible amount 1232 of network traffic; in that case, it may be preferable to simply 1233 repeat the update some reasonable number of times (say, 5 for 1234 wireless and 2 for wired links). 1236 A route retraction is somewhat less worrying: if the route retraction 1237 doesn't reach all neighbours, a black-hole might be created, which, 1238 unlike a routing loop, does not endanger the integrity of the 1239 network. When a route is retracted, a node SHOULD send a triggered 1240 update and SHOULD make a reasonable attempt at ensuring that all 1241 neighbours receive this retraction. 1243 Finally, a node MAY send a triggered update when the metric for a 1244 given prefix changes in a significant manner, due to a received 1245 update, because a link's cost has changed, or because a different 1246 next hop has been selected. A node SHOULD NOT send triggered updates 1247 for other reasons, such as when there is a minor fluctuation in a 1248 route's metric, when the selected next hop changes, or to propagate a 1249 new sequence number (except to satisfy a request, as specified in 1250 Section 3.8). 1252 3.7.3. Maintaining Feasibility Distances 1254 Before sending an update (prefix, plen, router-id, seqno, metric) 1255 with finite metric (i.e., not a route retraction), a Babel node 1256 updates the feasibility distance maintained in the source table. 1257 This is done as follows. 1259 If no entry indexed by (prefix, plen, router-id) exists in the source 1260 table, then one is created with value (prefix, plen, router-id, 1261 seqno, metric). 1263 If an entry (prefix, plen, router-id, seqno', metric') exists, then 1264 it is updated as follows: 1266 o if seqno > seqno', then seqno' := seqno, metric' := metric; 1268 o if seqno = seqno' and metric' > metric, then metric' := metric; 1270 o otherwise, nothing needs to be done. 1272 The garbage-collection timer for the entry is then reset. Note that 1273 the feasibility distance is not updated and the garbage-collection 1274 timer is not reset when a retraction (an update with infinite metric) 1275 is sent. 1277 When the garbage-collection timer expires, the entry is removed from 1278 the source table. 1280 3.7.4. Split Horizon 1282 When running over a transitive, symmetric link technology, e.g., a 1283 point-to-point link or a wired LAN technology such as Ethernet, a 1284 Babel node SHOULD use an optimisation known as split horizon. When 1285 split horizon is used on a given interface, a routing update for 1286 prefix P is not sent on the particular interface over which the 1287 selected route towards prefix P was learnt. 1289 Split horizon SHOULD NOT be applied to an interface unless the 1290 interface is known to be symmetric and transitive; in particular, 1291 split horizon is not applicable to decentralised wireless link 1292 technologies (e.g., IEEE 802.11 in ad hoc mode) when routing updates 1293 are sent over multicast. 1295 3.8. Explicit Requests 1297 In normal operation, a node's route table is populated by the regular 1298 and triggered updates sent by its neighbours. Under some 1299 circumstances, however, a node sends explicit requests in order to 1300 cause a resynchronisation with the source after a mobility event or 1301 to prevent a route from spuriously expiring. 1303 The Babel protocol provides two kinds of explicit requests: route 1304 requests, which simply request an update for a given prefix, and 1305 seqno requests, which request an update for a given prefix with a 1306 specific sequence number. The former are never forwarded; the latter 1307 are forwarded if they cannot be satisfied by the receiver. 1309 3.8.1. Handling Requests 1311 Upon receiving a request, a node either forwards the request or sends 1312 an update in reply to the request, as described in the following 1313 sections. If this causes an update to be sent, the update is either 1314 sent to a multicast address on the interface on which the request was 1315 received, or to the unicast address of the neighbour that sent the 1316 request. 1318 The exact behaviour is different for route requests and seqno 1319 requests. 1321 3.8.1.1. Route Requests 1323 When a node receives a route request for a given prefix, it checks 1324 its route table for a selected route to this exact prefix. If such a 1325 route exists, it MUST send an update (over unicast or over 1326 multicast); if such a route does not exist, it MUST send a retraction 1327 for that prefix. 1329 When a node receives a wildcard route request, it SHOULD send a full 1330 route table dump. Full route dumps MAY be rate-limited, especially 1331 if they are sent over multicast. 1333 3.8.1.2. Seqno Requests 1335 When a node receives a seqno request for a given router-id and 1336 sequence number, it checks whether its route table contains a 1337 selected entry for that prefix. If a selected route for the given 1338 prefix exists, it has finite metric, and either the router-ids are 1339 different or the router-ids are equal and the entry's sequence number 1340 is no smaller (modulo 2^16) than the requested sequence number, the 1341 node MUST send an update for the given prefix. If the router-ids 1342 match but the requested seqno is larger (modulo 2^16) than the route 1343 entry's, the node compares the router-id against its own router-id. 1344 If the router-id is its own, then it increases its sequence number by 1345 1 (modulo 2^16) and sends an update. A node MUST NOT increase its 1346 sequence number by more than 1 in response to a seqno request. 1348 Otherwise, if the requested router-id is not its own, the received 1349 request's hop count is 2 or more, and the node is advertising the 1350 prefix to its neighbours, the node selects a neighbour to forward the 1351 request to as follows: 1353 o if the node has one or more feasible routes toward the requested 1354 prefix with a next hop that is not the requesting node, then the 1355 node MUST forward the request to the next hop of one such route; 1357 o otherwise, if the node has one or more (not necessarily feasible) 1358 routes to the requested prefix with a next hop that is not the 1359 requesting node, then the node SHOULD forward the request to the 1360 next hop of one such route. 1362 In order to actually forward the request, the node decrements the hop 1363 count and sends the request in a unicast packet destined to the 1364 selected neighbour. 1366 A node SHOULD maintain a list of recently forwarded seqno requests 1367 and forward the reply (an update with a seqno sufficiently large to 1368 satisfy the request) in a timely manner. A node SHOULD compare every 1369 incoming seqno request against its list of recently forwarded seqno 1370 requests and avoid forwarding it if it is redundant (i.e., if it has 1371 recently sent a request with the same prefix, router-id and a seqno 1372 that is not smaller modulo 2^16). 1374 Since the request-forwarding mechanism does not necessarily obey the 1375 feasibility condition, it may get caught in routing loops; hence, 1376 requests carry a hop count to limit the time during which they remain 1377 in the network. However, since requests are only ever forwarded as 1378 unicast packets, the initial hop count need not be kept particularly 1379 low, and performing an expanding horizon search is not necessary. A 1380 single request MUST NOT be duplicated: it MUST NOT be forwarded to a 1381 multicast address, and it MUST NOT be forwarded to multiple 1382 neighbours. However, if a seqno request is resent by its originator, 1383 the subsequent copies MAY be forwarded to a different neighbour than 1384 the initial one. 1386 3.8.2. Sending Requests 1388 A Babel node MAY send a route or seqno request at any time, to a 1389 multicast or a unicast address; there is only one case when 1390 originating requests is required (Section 3.8.2.1). 1392 3.8.2.1. Avoiding Starvation 1394 When a route is retracted or expires, a Babel node usually switches 1395 to another feasible route for the same prefix. It may be the case, 1396 however, that no such routes are available. 1398 A node that has lost all feasible routes to a given destination but 1399 still has unexpired unfeasible routes to that destination MUST send a 1400 seqno request; if it doesn't have any such routes, it MAY still send 1401 a seqno request. The router-id of the request is set to the router- 1402 id of the route that it has just lost, and the requested seqno is the 1403 value contained in the source table plus 1. 1405 If the node has any (unfeasible) routes to the requested destination, 1406 then it MUST send the request to at least one of the next-hop 1407 neighbours that advertised these routes, and SHOULD send it to all of 1408 them; in any case, it MAY send the request to any other neighbours, 1409 whether they advertise a route to the requested destination or not. 1410 A simple implementation strategy is therefore to unconditionally 1411 multicast the request over all interfaces. 1413 Similar requests will be sent by other nodes that are affected by the 1414 route's loss. If the network is still connected, and assuming no 1415 packet loss, then at least one of these requests will be forwarded to 1416 the source, resulting in a route being advertised with a new sequence 1417 number. (Due to duplicate suppression, only a small number of such 1418 requests will actually reach the source.) 1420 In order to compensate for packet loss, a node SHOULD repeat such a 1421 request a small number of times if no route becomes feasible within a 1422 short time. In the presence of heavy packet loss, however, all such 1423 requests might be lost; in that case, the mechanism in the next 1424 section will eventually ensure that a new seqno is received. 1426 3.8.2.2. Dealing with Unfeasible Updates 1428 When a route's metric increases, a node might receive an unfeasible 1429 update for a route that it has currently selected. As specified in 1430 Section 3.5.1, the receiving node will either ignore the update or 1431 unselect the route. 1433 In order to keep routes from spuriously expiring because they have 1434 become unfeasible, a node SHOULD send a unicast seqno request when it 1435 receives an unfeasible update for a route that is currently selected. 1436 The requested sequence number is computed from the source table as in 1437 Section 3.8.2.1 above. 1439 Additionally, since metric computation does not necessarily coincide 1440 with the delay in propagating updates, a node might receive an 1441 unfeasible update from a currently unselected neighbour that is 1442 preferable to the currently selected route (e.g., because it has a 1443 much smaller metric); in that case, the node SHOULD send a unicast 1444 seqno request to the neighbour that advertised the preferable update. 1446 3.8.2.3. Preventing Routes from Expiring 1448 In normal operation, a route's expiry timer never triggers: since a 1449 route's hold time is computed from an explicit interval included in 1450 Update TLVs, a new update (possibly a retraction) should arrive in 1451 time to prevent a route from expiring. 1453 In the presence of packet loss, however, it may be the case that no 1454 update is successfully received for an extended period of time, 1455 causing a route to expire. In order to avoid such spurious expiry, 1456 shortly before a selected route expires, a Babel node SHOULD send a 1457 unicast route request to the neighbour that advertised this route; 1458 since nodes always send either updates or retractions in response to 1459 non-wildcard route requests (Section 3.8.1.1), this will usually 1460 result in the route being either refreshed or retracted. 1462 3.8.2.4. Acquiring New Neighbours 1464 In order to speed up convergence after a mobility event, a node MAY 1465 send a unicast wildcard request after acquiring a new neighbour. 1466 Additionally, a node MAY send a small number of multicast wildcard 1467 requests shortly after booting. Note however that doing that 1468 carelessly can cause serious congestion when a whole network is 1469 rebooted, especially on link layers with high per-packet overhead 1470 (e.g., IEEE 802.11). 1472 4. Protocol Encoding 1474 A Babel packet MUST be sent as the body of a UDP datagram, with 1475 network-layer hop count set to 1, destined to a well-known multicast 1476 address or to a unicast address, over IPv4 or IPv6; in the case of 1477 IPv6, these addresses are link-local. Both the source and 1478 destination UDP port are set to a well-known port number. A Babel 1479 packet MUST be silently ignored unless its source address is either a 1480 link-local IPv6 address or an IPv4 address belonging to the local 1481 network, and its source port is the well-known Babel port. It MAY be 1482 silently ignored if its destination address is a global IPv6 address. 1484 In order to minimise the number of packets being sent while avoiding 1485 lower-layer fragmentation, a Babel node SHOULD attempt to maximise 1486 the size of the packets it sends, up to the outgoing interface's MTU 1487 adjusted for lower-layer headers (28 octets for UDP over IPv4, 48 1488 octets for UDP over IPv6). It MUST NOT send packets larger than the 1489 attached interface's MTU adjusted for lower-layer headers or 512 1490 octets, whichever is larger, but not exceeding 2^16 - 1 adjusted for 1491 lower-layer headers. Every Babel speaker MUST be able to receive 1492 packets that are as large as any attached interface's MTU adjusted 1493 for lower-layer headers or 512 octets, whichever is larger. Babel 1494 packets MUST NOT be sent in IPv6 Jumbograms. 1496 In order to avoid global synchronisation of a Babel network and to 1497 aggregate multiple TLVs into large packets, a Babel node SHOULD 1498 buffer every TLV and delay sending a packet by a small, randomly 1499 chosen delay [JITTER]. In order to allow accurate computation of 1500 packet loss rates, this delay MUST NOT be larger than half the 1501 advertised Hello interval. 1503 4.1. Data Types 1505 4.1.1. Interval 1507 Relative times are carried as 16-bit values specifying a number of 1508 centiseconds (hundredths of a second). This allows times up to 1509 roughly 11 minutes with a granularity of 10ms, which should cover all 1510 reasonable applications of Babel. 1512 4.1.2. Router-Id 1514 A router-id is an arbitrary 8-octet value. A router-id MUST NOT 1515 consist of either all binary zeroes (0000000000000000 hexadecimal) or 1516 all binary ones ones (ffffffffffffffff hexadecimal). 1518 4.1.3. Address 1520 Since the bulk of the protocol is taken by addresses, multiple ways 1521 of encoding addresses are defined. Additionally, a common subnet 1522 prefix may be omitted when multiple addresses are sent in a single 1523 packet -- this is known as address compression (Section 4.6.9). 1525 Address encodings: 1527 o AE 0: wildcard address. The value is 0 octets long. 1529 o AE 1: IPv4 address. Compression is allowed. 4 octets or less. 1531 o AE 2: IPv6 address. Compression is allowed. 16 octets or less. 1533 o AE 3: link-local IPv6 address. Compression is not allowed. The 1534 value is 8 octets long, a prefix of fe80::/64 is implied. 1536 The address family associated to an address encoding is either IPv4 1537 or IPv6; it is undefined for AE 0, IPv4 for AE 1, and IPv6 for AEs 2 1538 and 3. 1540 4.1.4. Prefixes 1542 A network prefix is encoded just like a network address, but it is 1543 stored in the smallest number of octets that are enough to hold the 1544 significant bits (up to the prefix length). 1546 4.2. Packet Format 1548 A Babel packet consists of a 4-octet header, followed by a sequence 1549 of TLVs (the packet body), optionally followed by a second sequence 1550 of TLVs (the packet trailer). 1552 0 1 2 3 1553 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 1554 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1555 | Magic | Version | Body length | 1556 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1557 | Packet Body ... 1558 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+- 1559 | Packet Trailer... 1560 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+- 1562 Fields : 1564 Magic The arbitrary but carefully chosen value 42 (decimal); 1565 packets with a first octet different from 42 MUST be 1566 silently ignored. 1568 Version This document specifies version 2 of the Babel protocol. 1569 Packets with a second octet different from 2 MUST be 1570 silently ignored. 1572 Body length The length in octets of the body following the packet 1573 header (excluding the Magic, Version and Body length 1574 fields, and excluding the packet trailer). 1576 Packet Body The packet body; a sequence of TLVs. 1578 Packet Trailer The packet trailer; another sequence of TLVs. 1580 The packet body and trailer are both sequences of TLVs. The packet 1581 body is the normal place to store TLVs; the packet trailer only 1582 contains specialised TLVs that do not need to be protected by 1583 cryptographic security mechanisms. When parsing the trailer, the 1584 receiver MUST ignore any TLV unless its definition explicitly states 1585 that it is allowed to appear there. Among the TLVs defined in this 1586 document, only Pad1 and PadN are allowed in the trailer; since these 1587 TLVs are ignored in any case, an implementation MAY silently ignore 1588 the packet trailer without even parsing it, unless it implements at 1589 least one extension that defines TLVs that are allowed to appear in 1590 the trailer. 1592 4.3. TLV Format 1594 With the exception of Pad1, all TLVs have the following structure: 1596 0 1 2 3 1597 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 1598 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1599 | Type | Length | Payload... 1600 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+- 1602 Fields : 1604 Type The type of the TLV. 1606 Length The length of the body, exclusive of the Type and Length 1607 fields. If the body is longer than the expected length of 1608 a given type of TLV, any extra data MUST be silently 1609 ignored. 1611 Payload The TLV payload, which consists of a body and, for selected 1612 TLV types, an optional list of sub-TLVs. 1614 TLVs with an unknown type value MUST be silently ignored. 1616 4.4. Sub-TLV Format 1618 Every TLV carries an explicit length in its header; however, most 1619 TLVs are self-terminating, in the sense that it is possible to 1620 determine the length of the body without reference to the explicit 1621 Length field. If a TLV has a self-terminating format, then it MAY 1622 allow a sequence of sub-TLVs to follow the body. 1624 Sub-TLVs have the same structure as TLVs. With the exception of 1625 PAD1, all TLVs have the following structure: 1627 0 1 2 3 1628 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 1629 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1630 | Type | Length | Body... 1631 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+- 1633 Fields : 1635 Type The type of the sub-TLV. 1637 Length The length of the body, in octets, exclusive of the Type 1638 and Length fields. 1640 Body The sub-TLV body, the interpretation of which depends on 1641 both the type of the sub-TLV and the type of the TLV within 1642 which it is embedded. 1644 The most-significant bit of the sub-TLV, called the mandatory bit, 1645 indicates how to handle unknown sub-TLVs. If the mandatory bit is 1646 not set, then an unknown sub-TLV MUST be silently ignored, and the 1647 rest of the TLV processed normally. If the mandatory bit is set, 1648 then the whole enclosing TLV MUST be silently ignored (except for 1649 updating the parser state by a Router-Id, Next-Hop or Update TLV, see 1650 Section 4.6.7, Section 4.6.8, and Section 4.6.9). 1652 4.5. Parser state 1654 Babel uses a stateful parser: a TLV may refer to data from a previous 1655 TLV. The parser state consists of the following pieces of data: 1657 o for each address encoding that allows compression, the current 1658 default prefix; this is undefined at the start of the packet, and 1659 is updated by each Update TLV with the Prefix flag set 1660 (Section 4.6.9); 1662 o for each address family (IPv4 or IPv6), the current next-hop; this 1663 is the source address of the enclosing packet for the matching 1664 address family at the start of a packet, and is updated by each 1665 Next-Hop TLV (Section 4.6.8); 1667 o the current router-id; this is undefined at the start of the 1668 packet, and is updated by each Router-ID TLV (Section 4.6.7) and 1669 by each Update TLV with Router-Id flag set. 1671 Since the parser state is separate from the bulk of Babel's state, 1672 and since for correct parsing it must be identical across 1673 implementations, it is updated before checking for mandatory TLVs: 1674 parsing a TLV MUST update the parser state even if the TLV is 1675 otherwise ignored due to an unknown mandatory sub-TLV. 1677 None of the TLVs that modify the parser state are allowed in the 1678 packet trailer; hence, an implementation may choose to use a 1679 dedicated stateless parser to parse the packet trailer. 1681 4.6. Details of Specific TLVs 1683 4.6.1. Pad1 1685 0 1686 0 1 2 3 4 5 6 7 1687 +-+-+-+-+-+-+-+-+ 1688 | Type = 0 | 1689 +-+-+-+-+-+-+-+-+ 1691 Fields : 1693 Type Set to 0 to indicate a Pad1 TLV. 1695 This TLV is silently ignored on reception. It is allowed in the 1696 packet trailer. 1698 4.6.2. PadN 1700 0 1 2 3 1701 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 1702 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1703 | Type = 1 | Length | MBZ... 1704 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+- 1706 Fields : 1708 Type Set to 1 to indicate a PadN TLV. 1710 Length The length of the body, exclusive of the Type and Length 1711 fields. 1713 MBZ Set to 0 on transmission. 1715 This TLV is silently ignored on reception. It is allowed in the 1716 packet trailer. 1718 4.6.3. Acknowledgment Request 1720 0 1 2 3 1721 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 1722 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1723 | Type = 2 | Length | Reserved | 1724 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1725 | Nonce | Interval | 1726 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1728 This TLV requests that the receiver send an Acknowledgment TLV within 1729 the number of centiseconds specified by the Interval field. 1731 Fields : 1733 Type Set to 2 to indicate an Acknowledgment Request TLV. 1735 Length The length of the body, exclusive of the Type and Length 1736 fields. 1738 Reserved Sent as 0 and MUST be ignored on reception. 1740 Nonce An arbitrary value that will be echoed in the receiver's 1741 Acknowledgment TLV. 1743 Interval A time interval in centiseconds after which the sender will 1744 assume that this packet has been lost. This MUST NOT be 0. 1745 The receiver MUST send an Acknowledgment TLV before this 1746 time has elapsed (with a margin allowing for propagation 1747 time). 1749 This TLV is self-terminating, and allows sub-TLVs. 1751 4.6.4. Acknowledgment 1752 0 1 2 3 1753 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 1754 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1755 | Type = 3 | Length | Nonce | 1756 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1758 This TLV is sent by a node upon receiving an Acknowledgment Request. 1760 Fields : 1762 Type Set to 3 to indicate an Acknowledgment TLV. 1764 Length The length of the body, exclusive of the Type and Length 1765 fields. 1767 Nonce Set to the Nonce value of the Acknowledgment Request that 1768 prompted this Acknowledgment. 1770 Since nonce values are not globally unique, this TLV MUST be sent to 1771 a unicast address. 1773 This TLV is self-terminating, and allows sub-TLVs. 1775 4.6.5. Hello 1777 0 1 2 3 1778 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 1779 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1780 | Type = 4 | Length | Flags | 1781 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1782 | Seqno | Interval | 1783 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1785 This TLV is used for neighbour discovery and for determining a 1786 neighbour's reception cost. 1788 Fields : 1790 Type Set to 4 to indicate a Hello TLV. 1792 Length The length of the body, exclusive of the Type and Length 1793 fields. 1795 Flags The individual bits of this field specify special handling 1796 of this TLV (see below). 1798 Seqno If the Unicast flag is set, this is the value of the 1799 sending node's outgoing Unicast Hello seqno for this 1800 neighbour. Otherwise, it is the sending node's outgoing 1801 Multicast Hello seqno for this interface. 1803 Interval If non-zero, this is an upper bound, expressed in 1804 centiseconds, on the time after which the sending node will 1805 send a new scheduled Hello TLV with the same setting of the 1806 Unicast flag. If this is 0, then this Hello represents an 1807 unscheduled Hello, and doesn't carry any new information 1808 about times at which Hellos are sent. 1810 The Flags field is interpreted as follows: 1812 0 1 1813 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 1814 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1815 |U|X|X|X|X|X|X|X|X|X|X|X|X|X|X|X| 1816 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1818 o U (Unicast) flag (8000 hexadecimal): if set, then this Hello 1819 represents a Unicast Hello, otherwise it represents a Multicast 1820 Hello; 1822 o X: all other bits MUST be sent as 0 and silently ignored on 1823 reception. 1825 Every time a Hello is sent, the corresponding seqno counter MUST be 1826 incremented. Since there is a single seqno counter for all the 1827 Multicast Hellos sent by a given node over a given interface, if the 1828 Unicast flag is not set, this TLV MUST be sent to all neighbors on 1829 this link, which can be achieved by sending to a multicast 1830 destination, or by sending multiple packets to the unicast addresses 1831 of all reachable neighbours. Conversely, if the Unicast flag is set, 1832 this TLV MUST be sent to a single neighbour, which can achieved by 1833 sending to a unicast destination. In order to avoid large 1834 discontinuities in link quality, multiple Hello TLVs SHOULD NOT be 1835 sent in the same packet. 1837 This TLV is self-terminating, and allows sub-TLVs. 1839 4.6.6. IHU 1840 0 1 2 3 1841 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 1842 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1843 | Type = 5 | Length | AE | Reserved | 1844 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1845 | Rxcost | Interval | 1846 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1847 | Address... 1848 +-+-+-+-+-+-+-+-+-+-+-+- 1850 An IHU ("I Heard You") TLV is used for confirming bidirectional 1851 reachability and carrying a link's transmission cost. 1853 Fields : 1855 Type Set to 5 to indicate an IHU TLV. 1857 Length The length of the body, exclusive of the Type and Length 1858 fields. 1860 AE The encoding of the Address field. This should be 1 or 3 1861 in most cases. As an optimisation, it MAY be 0 if the TLV 1862 is sent to a unicast address, if the association is over a 1863 point-to-point link, or when bidirectional reachability is 1864 ascertained by means outside of the Babel protocol. 1866 Reserved Sent as 0 and MUST be ignored on reception. 1868 Rxcost The rxcost according to the sending node of the interface 1869 whose address is specified in the Address field. The value 1870 FFFF hexadecimal (infinity) indicates that this interface 1871 is unreachable. 1873 Interval An upper bound, expressed in centiseconds, on the time 1874 after which the sending node will send a new IHU; this MUST 1875 NOT be 0. The receiving node will use this value in order 1876 to compute a hold time for this symmetric association. 1878 Address The address of the destination node, in the format 1879 specified by the AE field. Address compression is not 1880 allowed. 1882 Conceptually, an IHU is destined to a single neighbour. However, IHU 1883 TLVs contain an explicit destination address, and MAY be sent to a 1884 multicast address, as this allows aggregation of IHUs destined to 1885 distinct neighbours into a single packet and avoids the need for an 1886 ARP or Neighbour Discovery exchange when a neighbour is not being 1887 used for data traffic. 1889 IHU TLVs with an unknown value in the AE field MUST be silently 1890 ignored. 1892 This TLV is self-terminating, and allows sub-TLVs. 1894 4.6.7. Router-Id 1896 0 1 2 3 1897 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 1898 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1899 | Type = 6 | Length | Reserved | 1900 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1901 | | 1902 + Router-Id + 1903 | | 1904 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1906 A Router-Id TLV establishes a router-id that is implied by subsequent 1907 Update TLVs. This TLV sets the router-id even if it is otherwise 1908 ignored due to an unknown mandatory sub-TLV. 1910 Fields : 1912 Type Set to 6 to indicate a Router-Id TLV. 1914 Length The length of the body, exclusive of the Type and Length 1915 fields. 1917 Reserved Sent as 0 and MUST be ignored on reception. 1919 Router-Id The router-id for routes advertised in subsequent Update 1920 TLVs. This MUST NOT consist of all zeroes or all ones. 1922 This TLV is self-terminating, and allows sub-TLVs. 1924 4.6.8. Next Hop 1926 0 1 2 3 1927 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 1928 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1929 | Type = 7 | Length | AE | Reserved | 1930 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1931 | Next hop... 1932 +-+-+-+-+-+-+-+-+-+-+-+- 1934 A Next Hop TLV establishes a next-hop address for a given address 1935 family (IPv4 or IPv6) that is implied in subsequent Update TLVs. 1937 This TLV sets up the next-hop for subsequent Update TLVs even if it 1938 is otherwise ignored due to an unknown mandatory sub-TLV. 1940 Fields : 1942 Type Set to 7 to indicate a Next Hop TLV. 1944 Length The length of the body, exclusive of the Type and Length 1945 fields. 1947 AE The encoding of the Address field. This SHOULD be 1 (IPv4) 1948 or 3 (link-local IPv6), and MUST NOT be 0. 1950 Reserved Sent as 0 and MUST be ignored on reception. 1952 Next hop The next-hop address advertised by subsequent Update TLVs, 1953 for this address family. 1955 When the address family matches the network-layer protocol that this 1956 packet is transported over, a Next Hop TLV is not needed: in the 1957 absence of a Next Hop TLV in a given address family, the next hop 1958 address is taken to be the source address of the packet. 1960 Next Hop TLVs with an unknown value for the AE field MUST be silently 1961 ignored. 1963 This TLV is self-terminating, and allows sub-TLVs. 1965 4.6.9. Update 1967 0 1 2 3 1968 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 1969 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1970 | Type = 8 | Length | AE | Flags | 1971 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1972 | Plen | Omitted | Interval | 1973 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1974 | Seqno | Metric | 1975 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 1976 | Prefix... 1977 +-+-+-+-+-+-+-+-+-+-+-+- 1979 An Update TLV advertises or retracts a route. As an optimisation, it 1980 can optionally have the side effect of establishing a new implied 1981 router-id and a new default prefix. 1983 Fields : 1985 Type Set to 8 to indicate an Update TLV. 1987 Length The length of the body, exclusive of the Type and Length 1988 fields. 1990 AE The encoding of the Prefix field. 1992 Flags The individual bits of this field specify special handling 1993 of this TLV (see below). 1995 Plen The length of the advertised prefix. 1997 Omitted The number of octets that have been omitted at the 1998 beginning of the advertised prefix and that should be taken 1999 from a preceding Update TLV in the same address family with 2000 the Prefix flag set. 2002 Interval An upper bound, expressed in centiseconds, on the time 2003 after which the sending node will send a new update for 2004 this prefix. This MUST NOT be 0. The receiving node will 2005 use this value to compute a hold time for the route table 2006 entry. The value FFFF hexadecimal (infinity) expresses 2007 that this announcement will not be repeated unless a 2008 request is received (Section 3.8.2.3). 2010 Seqno The originator's sequence number for this update. 2012 Metric The sender's metric for this route. The value FFFF 2013 hexadecimal (infinity) means that this is a route 2014 retraction. 2016 Prefix The prefix being advertised. This field's size is 2017 (Plen/8 - Omitted) rounded upwards. 2019 The Flags field is interpreted as follows: 2021 0 1 2 3 4 5 6 7 2022 +-+-+-+-+-+-+-+-+ 2023 |P|R|X|X|X|X|X|X| 2024 +-+-+-+-+-+-+-+-+ 2026 o P (Prefix) flag (80 hexadecimal): if set, then this Update 2027 establishes a new default prefix for subsequent Update TLVs with a 2028 matching address encoding within the same packet, even if this TLV 2029 is otherwise ignored due to an unknown mandatory sub-TLV; 2031 o R (Router-Id) flag (40 hexadecimal): if set, then this TLV 2032 establishes a new default router-id for this TLV and subsequent 2033 Update TLVs in the same packet, even if this TLV is otherwise 2034 ignored due to an unknown mandatory sub-TLV. This router-id is 2035 computed from the first address of the advertised prefix as 2036 follows: 2038 * if the length of the address is 8 octets or more, then the new 2039 router-id is taken from the 8 last octets of the address; 2041 * if the length of the address is smaller than 8 octets, then the 2042 new router-id consists of the required number of zero octets 2043 followed by the address, i.e., the address is stored on the 2044 right of the router-id. For example, for an IPv4 address, the 2045 router-id consists of 4 octets of zeroes followed by the IPv4 2046 address. 2048 o X: all other bits MUST be sent as 0 and silently ignored on 2049 reception. 2051 The prefix being advertised by an Update TLV is computed as follows: 2053 o the first Omitted octets of the prefix are taken from the previous 2054 Update TLV with the Prefix flag set and the same address encoding, 2055 even if it was ignored due to an unknown mandatory sub-TLV; 2057 o the next (Plen/8 - Omitted) rounded upwards octets are taken from 2058 the Prefix field; 2060 o the remaining octets are set to 0. If AE is 3 (link-local IPv6), 2061 Omitted MUST be 0) 2063 If the Metric field is finite, the router-id of the originating node 2064 for this announcement is taken from the prefix advertised by this 2065 Update if the Router-Id flag is set, computed as described above. 2066 Otherwise, it is taken either from the preceding Router-Id packet, or 2067 the preceding Update packet with the Router-Id flag set, whichever 2068 comes last, even if that TLV is otherwise ignored due to an unknown 2069 mandatory sub-TLV. 2071 The next-hop address for this update is taken from the last preceding 2072 Next Hop TLV with a matching address family (IPv4 or IPv6) in the 2073 same packet even if it was otherwise ignored due to an unknown 2074 mandatory sub-TLV; if no such TLV exists, it is taken from the 2075 network-layer source address of this packet. 2077 If the metric field is FFFF hexadecimal, this TLV specifies a 2078 retraction. In that case, the router-id, next-hop and seqno are not 2079 used. AE MAY then be 0, in which case this Update retracts all of 2080 the routes previously advertised by the sending interface. If the 2081 metric is finite, AE MUST NOT be 0. If the metric is infinite and AE 2082 is 0, Plen and Omitted MUST both be 0. 2084 Update TLVs with an unknown value in the AE field MUST be silently 2085 ignored. 2087 This TLV is self-terminating, and allows sub-TLVs. 2089 4.6.10. Route Request 2091 0 1 2 3 2092 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 2093 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 2094 | Type = 9 | Length | AE | Plen | 2095 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 2096 | Prefix... 2097 +-+-+-+-+-+-+-+-+-+-+-+- 2099 A Route Request TLV prompts the receiver to send an update for a 2100 given prefix, or a full route table dump. 2102 Fields : 2104 Type Set to 9 to indicate a Route Request TLV. 2106 Length The length of the body, exclusive of the Type and Length 2107 fields. 2109 AE The encoding of the Prefix field. The value 0 specifies 2110 that this is a request for a full route table dump (a 2111 wildcard request). 2113 Plen The length of the requested prefix. 2115 Prefix The prefix being requested. This field's size is Plen/8 2116 rounded upwards. 2118 A Request TLV prompts the receiver to send an update message 2119 (possibly a retraction) for the prefix specified by the AE, Plen, and 2120 Prefix fields, or a full dump of its route table if AE is 0 (in which 2121 case Plen MUST be 0 and Prefix is of length 0). 2123 This TLV is self-terminating, and allows sub-TLVs. 2125 4.6.11. Seqno Request 2127 0 1 2 3 2128 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 2129 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 2130 | Type = 10 | Length | AE | Plen | 2131 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 2132 | Seqno | Hop Count | Reserved | 2133 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 2134 | | 2135 + Router-Id + 2136 | | 2137 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 2138 | Prefix... 2139 +-+-+-+-+-+-+-+-+-+-+ 2141 A Seqno Request TLV prompts the receiver to send an Update for a 2142 given prefix with a given sequence number, or to forward the request 2143 further if it cannot be satisfied locally. 2145 Fields : 2147 Type Set to 10 to indicate a Seqno Request TLV. 2149 Length The length of the body, exclusive of the Type and Length 2150 fields. 2152 AE The encoding of the Prefix field. This MUST NOT be 0. 2154 Plen The length of the requested prefix. 2156 Seqno The sequence number that is being requested. 2158 Hop Count The maximum number of times that this TLV may be forwarded, 2159 plus 1. This MUST NOT be 0. 2161 Reserved Sent as 0 and MUST be ignored on reception. 2163 Router-Id The Router-Id that is being requested. This MUST NOT 2164 consist of all zeroes or all ones. 2166 Prefix The prefix being requested. This field's size is Plen/8 2167 rounded upwards. 2169 A Seqno Request TLV prompts the receiving node to send a finite- 2170 metric Update for the prefix specified by the AE, Plen, and Prefix 2171 fields, with either a router-id different from what is specified by 2172 the Router-Id field, or a Seqno no less (modulo 2^16) than what is 2173 specified by the Seqno field. If this request cannot be satisfied 2174 locally, then it is forwarded according to the rules set out in 2175 Section 3.8.1.2. 2177 While a Seqno Request MAY be sent to a multicast address, it MUST NOT 2178 be forwarded to a multicast address and MUST NOT be forwarded to more 2179 than one neighbour. A request MUST NOT be forwarded if its Hop Count 2180 field is 1. 2182 This TLV is self-terminating, and allows sub-TLVs. 2184 4.7. Details of specific sub-TLVs 2186 4.7.1. Pad1 2188 0 1 2 3 4 5 6 7 2189 +-+-+-+-+-+-+-+-+ 2190 | Type = 0 | 2191 +-+-+-+-+-+-+-+-+ 2193 Fields : 2195 Type Set to 0 to indicate a Pad1 sub-TLV. 2197 This sub-TLV is silently ignored on reception. It is allowed within 2198 any TLV that allows sub-TLVs. 2200 4.7.2. PadN 2202 0 1 2 3 2203 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 2204 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 2205 | Type = 1 | Length | MBZ... 2206 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+- 2208 Fields : 2210 Type Set to 1 to indicate a PadN sub-TLV. 2212 Length The length of the body, in octets, exclusive of the Type 2213 and Length fields. 2215 MBZ Set to 0 on transmission. 2217 This sub-TLV is silently ignored on reception. It is allowed within 2218 any TLV that allows sub-TLVs. 2220 5. IANA Considerations 2222 IANA has registered the UDP port number 6696, called "babel", for use 2223 by the Babel protocol. 2225 IANA has registered the IPv6 multicast group ff02::1:6 and the IPv4 2226 multicast group 224.0.0.111 for use by the Babel protocol. 2228 IANA has created a registry called "Babel TLV Types". The values in 2229 this registry are not changed by this specification. 2231 IANA has created a registry called "Babel sub-TLV Types". Due to the 2232 addition of a Mandatory bit to the Babel protocol, the values in the 2233 "Babel sub-TLV Types" registry are amended as follows: 2235 +---------+-----------------------------------------+---------------+ 2236 | Type | Name | Reference | 2237 +---------+-----------------------------------------+---------------+ 2238 | 0 | Pad1 | this document | 2239 | | | | 2240 | 1 | PadN | this document | 2241 | | | | 2242 | 112-126 | Reserved for Experimental Use | this document | 2243 | | | | 2244 | 127 | Reserved for expansion of the type | this document | 2245 | | space | | 2246 | | | | 2247 | 240-254 | Reserved for Experimental Use | this document | 2248 | | | | 2249 | 255 | Reserved for expansion of the type | this document | 2250 | | space | | 2251 +---------+-----------------------------------------+---------------+ 2253 Existing assignments in the "Babel sub-TLV Types" registry in the 2254 range 2 to 111 are not changed by this specification. The values 224 2255 through 239, previously reserved for Experimental Use, are now 2256 unassigned. 2258 IANA has created a registry called "Babel Flags Values". IANA is 2259 instructed to rename this registry to "Babel Update Flags Values", 2260 with its contents unchanged. 2262 IANA is instructed to create a new registry called "Babel Hello Flags 2263 Values". The allocation policy for this registry is Specification 2264 Required [RFC8126]. The initial values in this registry are as 2265 follows: 2267 +------+------------+---------------+ 2268 | Bit | Name | Reference | 2269 +------+------------+---------------+ 2270 | 0 | Unicast | this document | 2271 | | | | 2272 | 1-15 | Unassigned | | 2273 +------+------------+---------------+ 2275 IANA is instructed to replace all references to RFCs 6126 and 7557 in 2276 all of the registries mentioned above by references to this document. 2278 6. Security Considerations 2280 As defined in this document, Babel is a completely insecure protocol. 2281 Any attacker can misdirect data traffic by advertising routes with a 2282 low metric or a high seqno. This issue can be solved either by a 2283 lower-layer security mechanism (e.g., link-layer security or IPsec), 2284 or by deploying a suitable authentication mechanism within Babel 2285 itself. There are currently two such mechanisms: Babel over DTLS 2286 [BABEL-DTLS] and HMAC-based authentication [BABEL-HMAC]. Both 2287 mechanisms ensure integrity of messages and prevent message replay, 2288 but only DTLS supports asymmetric keying and message confidentiality. 2289 HMAC is simpler and does not depend on DTLS, and therefore its use is 2290 RECOMMENDED whenever both mechanisms are applicable. 2292 The information that a Babel node announces to the whole routing 2293 domain is often sufficient to determine a mobile node's physical 2294 location with reasonable precision. The privacy issues that this 2295 causes can be mitigated somewhat by using randomly chosen router-ids 2296 and randomly chosen IP addresses, and changing them periodically. 2298 When carried over IPv6, Babel packets are ignored unless they are 2299 sent from a link-local IPv6 address; since routers don't forward 2300 link-local IPv6 packets, this provides protection against spoofed 2301 Babel packets being sent from the global Internet. No such natural 2302 protection exists when Babel packets are carried over IPv4. 2304 7. Acknowledgments 2306 A number of people have contributed text and ideas to this 2307 specification. The authors are particularly indebted to Matthieu 2308 Boutier, Gwendoline Chouasne, Margaret Cullen, Donald Eastlake, Toke 2309 Hoiland-Jorgensen, Joao Sobrinho and Martin Vigoureux. Earlier 2310 versions of this document greatly benefited from the input of Joel 2311 Halpern. The address compression technique was inspired by 2312 [PACKETBB]. 2314 8. References 2316 8.1. Normative References 2318 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 2319 Requirement Levels", BCP 14, RFC 2119, 2320 DOI 10.17487/RFC2119, March 1997. 2322 [RFC8174] Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC 2323 2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174, 2324 May 2017. 2326 8.2. Informative References 2328 [BABEL-DTLS] 2329 Decimo, A., Schinazi, D., and J. Chroboczek, "Babel 2330 Routing Protocol over Datagram Transport Layer Security", 2331 Internet Draft draft-ietf-babel-dtls-04, February 2019. 2333 [BABEL-HMAC] 2334 Do, C., Kolodziejak, W., and J. Chroboczek, "HMAC 2335 authentication for the Babel routing protocol", Internet 2336 Draft draft-ietf-babel-hmac-04, March 2019. 2338 [DSDV] Perkins, C. and P. Bhagwat, "Highly Dynamic Destination- 2339 Sequenced Distance-Vector Routing (DSDV) for Mobile 2340 Computers", ACM SIGCOMM'94 Conference on Communications 2341 Architectures, Protocols and Applications 234-244, 1994. 2343 [DUAL] Garcia Luna Aceves, J., "Loop-Free Routing Using Diffusing 2344 Computations", IEEE/ACM Transactions on Networking 1:1, 2345 February 1993. 2347 [EIGRP] Albrightson, B., Garcia Luna Aceves, J., and J. Boyle, 2348 "EIGRP -- a Fast Routing Protocol Based on Distance 2349 Vectors", Proc. Interop 94, 1994. 2351 [ETX] De Couto, D., Aguayo, D., Bicket, J., and R. Morris, "A 2352 high-throughput path metric for multi-hop wireless 2353 networks", Proc. MobiCom 2003, 2003. 2355 [IS-IS] "Information technology -- Telecommunications and 2356 information exchange between systems -- Intermediate 2357 System to Intermediate System intra-domain routeing 2358 information exchange protocol for use in conjunction with 2359 the protocol for providing the connectionless-mode network 2360 service (ISO 8473)", ISO/IEC 10589:2002, 2002. 2362 [JITTER] Floyd, S. and V. Jacobson, "The synchronization of 2363 periodic routing messages", IEEE/ACM Transactions on 2364 Networking 2, 2, 122-136, April 1994. 2366 [OSPF] Moy, J., "OSPF Version 2", RFC 2328, April 1998. 2368 [PACKETBB] 2369 Clausen, T., Dearlove, C., Dean, J., and C. Adjih, 2370 "Generalized Mobile Ad Hoc Network (MANET) Packet/Message 2371 Format", RFC 5444, February 2009. 2373 [RFC8126] Cotton, M., Leiba, B., and T. Narten, "Guidelines for 2374 Writing an IANA Considerations Section in RFCs", BCP 26, 2375 RFC 8126, June 2017. 2377 [RIP] Malkin, G., "RIP Version 2", RFC 2453, November 1998. 2379 Appendix A. Cost and Metric Computation 2381 The strategy for computing link costs and route metrics is a local 2382 matter; Babel itself only requires that it comply with the conditions 2383 given in Section 3.4.3 and Section 3.5.2. Different nodes may use 2384 different strategies in a single network and may use different 2385 strategies on different interface types. This section describes the 2386 strategies used by the sample implementation of Babel. 2388 The sample implementation of Babel sends periodic Multicast Hellos, 2389 and never sends Unicast Hellos. It maintains statistics about the 2390 last 16 received Hello TLVs of each kind (Appendix A.1), computes 2391 costs by using the 2-out-of-3 strategy (Appendix A.2.1) on wired 2392 links, and ETX (Appendix A.2.2) on wireless links. It uses an 2393 additive algebra for metric computation (Appendix A.3.1). 2395 A.1. Maintaining Hello History 2397 For each neighbour, the sample implementation of Babel maintains two 2398 sets of Hello history, one for each kind of Hello, and an expected 2399 sequence number, one for Multicast and one for Unicast Hellos. Each 2400 Hello history is a vector of 16 bits, where a 1 value represents a 2401 received Hello, and a 0 value a missed Hello. For each kind of 2402 Hello, the expected sequence number, written ne, is the sequence 2403 number that is expected to be carried by the next received Hello from 2404 this neighbour. 2406 Whenever it receives a Hello packet of a given kind from a neighbour, 2407 a node compares the received sequence number nr for that kind of 2408 Hello with its expected sequence number ne. Depending on the outcome 2409 of this comparison, one of the following actions is taken: 2411 o if the two differ by more than 16 (modulo 2^16), then the sending 2412 node has probably rebooted and lost its sequence number; the whole 2413 associated neighbour table entry is flushed and a new one is 2414 created; 2416 o otherwise, if the received nr is smaller (modulo 2^16) than the 2417 expected sequence number ne, then the sending node has increased 2418 its Hello interval without us noticing; the receiving node removes 2419 the last (ne - nr) entries from this neighbour's Hello history (we 2420 "undo history"); 2422 o otherwise, if nr is larger (modulo 2^16) than ne, then the sending 2423 node has decreased its Hello interval, and some Hellos were lost; 2424 the receiving node adds (nr - ne) 0 bits to the Hello history (we 2425 "fast-forward"). 2427 The receiving node then appends a 1 bit to the Hello history and sets 2428 ne to (nr + 1). If the Interval field of the received Hello is not 2429 zero, it resets the neighbour's hello timer to 1.5 times the 2430 advertised Interval (the extra margin allows for delay due to 2431 jitter). 2433 Whenever either Hello timer associated to a neighbour expires, the 2434 local node adds a 0 bit to this neighbour's Hello history, and 2435 increments the expected Hello number. If both Hello histories are 2436 empty (they contain 0 bits only), the neighbour entry is flushed; 2437 otherwise, the relevant hello timer is reset to the value advertised 2438 in the last Hello of that kind received from this neighbour (no extra 2439 margin is necessary in this case, since jitter was already taken into 2440 account when computing the timeout that has just expired). 2442 A.2. Cost Computation 2444 This section discusses how to compute costs based on Hello history. 2446 A.2.1. k-out-of-j 2448 K-out-of-j link sensing is suitable for wired links that are either 2449 up, in which case they only occasionally drop a packet, or down, in 2450 which case they drop all packets. 2452 The k-out-of-j strategy is parameterised by two small integers k and 2453 j, such that 0 < k <= j, and the nominal link cost, a constant K >= 2454 1. A node keeps a history of the last j hellos; if k or more of 2455 those have been correctly received, the link is assumed to be up, and 2456 the rxcost is set to K; otherwise, the link is assumed to be down, 2457 and the rxcost is set to infinity. 2459 Since Babel supports two kinds of Hellos, a Babel node performs k- 2460 out-of-j twice for each neighbour, once on the Unicast and once on 2461 the Multicast Hello history. If either of the instances of k-out- 2462 of-j indicates that the link is up, then the link is assumed to be 2463 up, and the rxcost is set to K; if both instances indicate that the 2464 link is down, then the link is assumed to be down, and the rxcost is 2465 set to infinity. In other words, the resulting rxcost is the minimum 2466 of the rxcosts yielded by the two instances of k-out-of-j link 2467 sensing. 2469 The cost of a link performing k-out-of-j link sensing is defined as 2470 follows: 2472 o cost = FFFF hexadecimal if rxcost = FFFF hexadecimal; 2474 o cost = txcost otherwise. 2476 A.2.2. ETX 2478 Unlike wired links, which are bimodal (either up or down), wireless 2479 links exhibit continuous variation of the link quality. Naive 2480 application of hop-count routing in networks that use wireless links 2481 for transit tends to select long, lossy links in preference to 2482 shorter, lossless links, which can dramatically reduce throughput. 2483 For that reason, a routing protocol designed to support wireless 2484 links must perform some form of link-quality estimation. 2486 ETX [ETX] is a simple link-quality estimation algorithm that is 2487 designed to work well with the IEEE 802.11 MAC. By default, the 2488 IEEE 802.11 MAC performs ARQ and rate adaptation on unicast frames, 2489 but not on multicast frames, which are sent at a fixed rate with no 2490 ARQ; therefore, measuring the loss rate of multicast frames yields a 2491 useful estimate of a link's quality. 2493 A node performing ETX link quality estimation uses a neighbour's 2494 Multicast Hello history to compute an estimate, written beta, of the 2495 probability that a Hello TLV is successfully received. Beta can be 2496 computed as the fraction of 1 bits within a small number (say, 6) of 2497 the most recent entries in the Multicast Hello history, or it can be 2498 an exponential average, or some combination of both approaches. 2500 Let alpha be MIN(1, 256/txcost), an estimate of the probability of 2501 successfully sending a Hello TLV. The cost is then computed by 2503 cost = 256/(alpha * beta) 2505 or, equivalently, 2506 cost = (MAX(txcost, 256) * rxcost) / 256. 2508 Since the IEEE 802.11 MAC performs ARQ on unicast frames, unicast 2509 frames do not provide a useful measure of link quality, and therefore 2510 ETX ignores the Unicast Hello history. Thus, a node performing ETX 2511 link-quality estimation will not route through neighbouring nodes 2512 unless they send periodic Multicast Hellos (possibly in addition to 2513 Unicast Hellos). 2515 A.3. Metric Computation 2517 As described in Section 3.5.2, the metric advertised by a neighbour 2518 is combined with the link cost to yield a metric. 2520 A.3.1. Additive Metrics 2522 The simplest approach for obtaining a monotonic, left-distributive 2523 metric is to define the metric of a route as the sum of the costs of 2524 the component links. More formally, if a neighbour advertises a 2525 route with metric m over a link with cost c, then the resulting route 2526 has metric M(c, m) = c + m. 2528 A multiplicative metric can be converted into an additive one by 2529 taking the logarithm (in some suitable base) of the link costs. 2531 A.3.2. External Sources of Willingness 2533 A node may want to vary its willingness to forward packets by taking 2534 into account information that is external to the Babel protocol, such 2535 as the monetary cost of a link, the node's battery status, CPU load, 2536 etc. This can be done by adding to every route's metric a value k 2537 that depends on the external data. For example, if a battery-powered 2538 node receives an update with metric m over a link with cost c, it 2539 might compute a metric M(c, m) = k + c + m, where k depends on the 2540 battery status. 2542 In order to preserve strict monotonicity (Section 3.5.2), the value k 2543 must be greater than -c. 2545 Appendix B. Constants 2547 The choice of time constants is a trade-off between fast detection of 2548 mobility events and protocol overhead. Two implementations of Babel 2549 with different time constants will interoperate, although the 2550 resulting convergence time will most likely be dictated by the slower 2551 of the two. 2553 Experience with the sample implementation of Babel indicates that the 2554 Hello interval is the most important time constant: a mobility event 2555 is detected within 1.5 to 3 Hello intervals. Due to Babel's reliance 2556 on triggered updates and explicit requests, the Update interval only 2557 has an effect on the time it takes for accurate metrics to be 2558 propagated after variations in link costs too small to trigger an 2559 unscheduled update or in the presence of packet loss. 2561 At the time of writing, the sample implementation of Babel uses the 2562 following default values: 2564 Multicast Hello Interval: 4 seconds. 2566 IHU Interval: the advertised IHU interval is always 3 times the 2567 Multicast Hello interval. IHUs are actually sent with each Hello 2568 on lossy links (as determined from the Hello history), but only 2569 with every third Multicast Hello on lossless links. 2571 Unicast Hello Interval: the sample implementation never sends 2572 scheduled Unicast Hellos; 2574 Update Interval: 4 times the Multicast Hello interval. 2576 IHU Hold Time: 3.5 times the advertised IHU interval. 2578 Route Expiry Time: 3.5 times the advertised update interval. 2580 Source GC time: 3 minutes. 2582 Request timeout: initially 2 seconds, doubled every time a request 2583 is resent, up to a maximum of three times. 2585 The amount of jitter applied to a packet depends on whether it 2586 contains any urgent TLVs or not (Section 3.1). Urgent triggered 2587 updates and urgent requests are delayed by no more than 200ms; 2588 acknowledgments, by no more than the associated deadline; and other 2589 TLVs by no more than one-half the Multicast Hello interval. 2591 Appendix C. Considerations for protocol extensions 2593 Babel is an extensible protocol, and this document defines a number 2594 of mechanisms that can be used to extend the protocol in a backwards 2595 compatible manner: 2597 o increasing the version number in the packet header; 2599 o defining new TLVs; 2600 o defining new sub-TLVs (with or without the mandatory bit set); 2602 o defining new AEs; 2604 o using the packet trailer. 2606 This appendix is intended to guide designers of protocol extensions 2607 in chosing a particular encoding. 2609 The version number in the Babel header should only be increased if 2610 the new version is not backwards compatible with the original 2611 protocol. 2613 In many cases, an extension could be implemented either by defining a 2614 new TLV, or by adding a new sub-TLV to an existing TLV. For example, 2615 an extension whose purpose is to attach additional data to route 2616 updates can be implemented either by creating a new "enriched" Update 2617 TLV, by adding a non-mandatory sub-TLV to the Update TLV, or by 2618 adding a mandatory sub-TLV. 2620 The various encodings are treated differently by implementations that 2621 do not understand the extension. In the case of a new TLV or of a 2622 sub-TLV with the mandatory bit set, the whole TLV is ignored by 2623 implementations that do not implement the extension, while in the 2624 case of a non-mandatory sub-TLV, the TLV is parsed and acted upon, 2625 and only the unknown sub-TLV is silently ignored. Therefore, a non- 2626 mandatory sub-TLV should be used by extensions that extend the Update 2627 in a compatible manner (the extension data may be silently ignored), 2628 while a mandatory sub-TLV or a new TLV must be used by extensions 2629 that make incompatible extensions to the meaning of the TLV (the 2630 whole TLV must be thrown away if the extension data is not 2631 understood). 2633 Experience shows that the need for additional data tends to crop up 2634 in the most unexpected places. Hence, it is recommended that 2635 extensions that define new TLVs should make them self-terminating, 2636 and allow attaching sub-TLVs to them. 2638 Adding a new AE is essentially equivalent to adding a new TLV: Update 2639 TLVs with an unknown AE are ignored, just like unknown TLVs. 2640 However, adding a new AE is more involved than adding a new TLV, 2641 since it creates a new set of compression state. Additionally, since 2642 the Next Hop TLV creates state specific to a given address family, as 2643 opposed to a given AE, a new AE for a previously defined address 2644 family must not be used in the Next Hop TLV if backwards 2645 compatibility is required. A similar issue arises with Update TLVs 2646 with unknown AEs establishing a new router-id (due to the Router-Id 2647 flag being set). Therefore, defining new AEs must be done with care 2648 if compatibility with unextended implementations is required. 2650 The packet trailer is intended to carry cryptographic signatures that 2651 only cover the packet body; storing the cryptographic signatures in 2652 the packet trailer avoids clearing the signature before computing a 2653 hash of the packet body, and makes it possible to check a 2654 cryptographic signature before running the full, stateful TLV parser. 2655 Hence, only TLVs that don't need to be protected by cryptographic 2656 security protocols should be allowed in the packet trailer. Any such 2657 TLVs should be easy to parse, and in particular should not require 2658 stateful parsing. 2660 Appendix D. Stub Implementations 2662 Babel is a fairly economic protocol. Updates take between 12 and 40 2663 octets per destination, depending on the address family and how 2664 successful compression is; in a double-stack flat network, an average 2665 of less than 24 octets per update is typical. The route table 2666 occupies about 35 octets per IPv6 entry. To put these values into 2667 perspective, a single full-size Ethernet frame can carry some 65 2668 route updates, and a megabyte of memory can contain a 20000-entry 2669 route table and the associated source table. 2671 Babel is also a reasonably simple protocol. The sample 2672 implementation consists of less than 12 000 lines of C code, and it 2673 compiles to less than 120 kB of text on a 32-bit CISC architecture; 2674 about half of this figure is due to protocol extensions and user- 2675 interface code. 2677 Nonetheless, in some very constrained environments, such as PDAs, 2678 microwave ovens, or abacuses, it may be desirable to have subset 2679 implementations of the protocol. 2681 There are many different definitions of a stub router, but for the 2682 needs of this section a stub implementation of Babel is one that 2683 announces one or more directly attached prefixes into a Babel network 2684 but doesn't reannounce any routes that it has learnt from its 2685 neighbours. It may either maintain a full routing table, or simply 2686 select a default gateway amongst any one of its neighbours that 2687 announces a default route. Since a stub implementation never 2688 forwards packets except from or to directly attached links, it cannot 2689 possibly participate in a routing loop, and hence it need not 2690 evaluate the feasibility condition or maintain a source table. 2692 No matter how primitive, a stub implementation MUST parse sub-TLVs 2693 attached to any TLVs that it understands and check the mandatory bit. 2694 It MUST answer acknowledgment requests and MUST participate in the 2695 Hello/IHU protocol. It MUST also be able to reply to seqno requests 2696 for routes that it announces and SHOULD be able to reply to route 2697 requests. 2699 Experience shows that an IPv6-only stub implementation of Babel can 2700 be written in less than 1000 lines of C code and compile to 13 kB of 2701 text on 32-bit CISC architecture. 2703 Appendix E. Software Availability 2705 The sample implementation of Babel is available from 2706 . 2708 Appendix F. Changes from previous versions 2710 F.1. Changes since RFC 6126 2712 o Changed UDP port number to 6696. 2714 o Consistently use router-id rather than id. 2716 o Clarified that the source garbage collection timer is reset after 2717 sending an update even if the entry was not modified. 2719 o In section "Seqno Requests", fixed an erroneous "route request". 2721 o In the description of the Seqno Request TLV, added the description 2722 of the Router-Id field. 2724 o Made router-ids all-0 and all-1 forbidden. 2726 F.2. Changes since draft-ietf-babel-rfc6126bis-00 2728 o Added security considerations. 2730 F.3. Changes since draft-ietf-babel-rfc6126bis-01 2732 o Integrated the format of sub-TLVs. 2734 o Mentioned for each TLV whether it supports sub-TLVs. 2736 o Added Appendix C. 2738 o Added a mandatory bit in sub-TLVs. 2740 o Changed compression state to be per-AF rather than per-AE. 2742 o Added implementation hint for the routing table. 2744 o Clarified how router-ids are computed when bit 0x40 is set in 2745 Updates. 2747 o Relaxed the conditions for sending requests, and tightened the 2748 conditions for forwarding requests. 2750 o Clarified that neighbours should be acquired at some point, but it 2751 doesn't matter when. 2753 F.4. Changes since draft-ietf-babel-rfc6126bis-02 2755 o Added Unicast Hellos. 2757 o Added unscheduled (interval-less) Hellos. 2759 o Changed Appendix A to consider Unicast and unscheduled Hellos. 2761 o Changed Appendix B to agree with the reference implementation. 2763 o Added optional algorithm to avoid the hold time. 2765 o Changed the table of pending seqno requests to be indexed by 2766 router-id in addition to prefixes. 2768 o Relaxed the route acquisition algorithm. 2770 o Replaced minimal implementations by stub implementations. 2772 o Added acknowledgments section. 2774 F.5. Changes since draft-ietf-babel-rfc6126bis-03 2776 o Clarified that all the data structures are conceptual. 2778 o Made sending and receiving Multicast Hellos a SHOULD, avoids 2779 expressing any opinion about Unicast Hellos. 2781 o Removed opinion about Multicast vs. Unicast Hellos (Appendix A.4). 2783 o Made hold-time into a SHOULD rather than MUST. 2785 o Clarified that Seqno Requests are for a finite-metric Update. 2787 o Clarified that sub-TLVs Pad1 and PadN are allowed within any TLV 2788 that allows sub-TLVs. 2790 o Updated IANA Considerations. 2792 o Updated Security Considerations. 2794 o Renamed routing table back to route table. 2796 o Made buffering outgoing updates a SHOULD. 2798 o Weakened advice to use modified EUI-64 in router-ids. 2800 o Added information about sending requests to Appendix B. 2802 o A number of minor wording changes and clarifications. 2804 F.6. Changes since draft-ietf-babel-rfc6126bis-03 2806 Minor editorial changes. 2808 F.7. Changes since draft-ietf-babel-rfc6126bis-04 2810 o Renamed isotonicity to left-distributivity. 2812 o Minor clarifications to unicast hellos. 2814 o Updated requirements boilerplate to RFC 8174. 2816 o Minor editorial changes. 2818 F.8. Changes since draft-ietf-babel-rfc6126bis-05 2820 o Added information about the packet trailer, now that it is used by 2821 draft-ietf-babel-hmac. 2823 F.9. Changes since draft-ietf-babel-rfc6126bis-06 2825 o Added references to security documents. 2827 F.10. Changes since draft-ietf-babel-rfc6126bis-07 2829 o Added list of obsoleted drafts to the abstract. 2831 o Updated references. 2833 F.11. Changes since draft-ietf-babel-rfc6126bis-08 2835 o Added recommendation that route selection should not take seqnos 2836 into account. 2838 F.12. Changes since draft-ietf-babel-rfc6126bis-09 2840 o Editorial changes only. 2842 F.13. Changes since draft-ietf-babel-rfc6126bis-10 2844 o Editorial changes only. 2846 Authors' Addresses 2848 Juliusz Chroboczek 2849 IRIF, University of Paris-Diderot 2850 Case 7014 2851 75205 Paris Cedex 13 2852 France 2854 Email: jch@irif.fr 2856 David Schinazi 2857 Google LLC 2858 1600 Amphitheatre Parkway 2859 Mountain View, California 94043 2860 USA 2862 Email: dschinazi.ietf@gmail.com