Debian Codegraph @

rfc3488.html

raw
1 <pre>Network Working Group I. Wu
2 Request for Comments: 3488 T. Eckert
3 Category: Informational Cisco Systems
4 February 2003
5
6
7 <span class="h1">Cisco Systems</span>
8 <span class="h1">Router-port Group Management Protocol (RGMP)</span>
9
10 Status of this Memo
11
12 This memo provides information for the Internet community. It does
13 not specify an Internet standard of any kind. Distribution of this
14 memo is unlimited.
15
16 Copyright Notice
17
18 Copyright (C) The Internet Society (2003). All Rights Reserved.
19
20 Abstract
21
22 This document describes the Router-port Group Management Protocol
23 (RGMP). This protocol was developed by Cisco Systems and is used
24 between multicast routers and switches to restrict multicast packet
25 forwarding in switches to those routers where the packets may be
26 needed.
27
28 RGMP is designed for backbone switched networks where multiple, high
29 speed routers are interconnected.
30
31 <span class="h2"><a class="selflink" id="section-1" href="#section-1">1</a>. Conventions used in this document</span>
32
33 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
34 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
35 document are to be interpreted as described in <a href="https://www.rfc-editor.org/bcp/bcp14">BCP 14</a>, <a href="./rfc2119">RFC 2119</a> [<a href="#ref-2" title="&quot;Key words for use in RFCs to Indicate Requirement Levels&quot;">2</a>].
36
37 <span class="h2"><a class="selflink" id="section-2" href="#section-2">2</a>. Introduction</span>
38
39 IGMP Snooping is a popular, but not well documented mechanism to
40 restrict multicast traffic, in switched networks, to those ports that
41 want to receive the multicast traffic. It dynamically establishes
42 and terminates multicast group specific forwarding in switches that
43 support this feature.
44
45
46
47
48
49
50
51
52 <span class="grey">Wu &amp; Eckert Informational [Page 1]</span></pre>
53 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-2" ></span>
54 <span class="grey"><a href="./rfc3488">RFC 3488</a> Cisco Systems RGMP February 2003</span>
55
56
57 The main limitation of IGMP Snooping is that it can only restrict
58 multicast traffic onto switch ports where receiving hosts are
59 connected directly or indirectly via other switches. IGMP Snooping
60 can not restrict multicast traffic to ports where at least one
61 multicast router is connected. It must instead flood multicast
62 traffic to these ports. Snooping on IGMP messages alone is an
63 intrinsic limitation. Through it, a switch can only learn which
64 multicast flows are being requested by hosts. A switch can not learn
65 through IGMP which traffic flows need to be received by router ports
66 to be routed because routers do not report these flows via IGMP.
67
68 In situations where multiple multicast routers are connected to a
69 switched backbone, IGMP Snooping will not reduce multicast traffic
70 load. Nor will it assist in increasing internal bandwidth through
71 the use of switches in the network.
72
73 In switched backbone networks or exchange points, where predominantly
74 routers are connected with each other, a large amount of multicast
75 traffic may lead to unexpected congestion. It also leads to more
76 resource consumption in the routers because they must discard the
77 unwanted multicast traffic.
78
79 The RGMP protocol described in this document restricts multicast
80 traffic to router ports. To effectively restrict traffic, it must be
81 supported by both the switches and the routers in the network.
82
83 The RGMP message format resembles the IGMPv2 message format so that
84 existing switches, capable of IGMP Snooping, can easily be enhanced
85 with this feature. Its messages are encapsulated in IPv4 datagrams,
86 with a protocol number of 2, the same as that of IGMP. All RGMP
87 messages are sent with TTL 1, to destination address 224.0.0.25. This
88 address has been assigned by IANA to carry messages from routers to
89 switches [<a href="#ref-3">3</a>].
90
91 RGMP is designed to work in conjunction with multicast routing
92 protocols where explicit join/prune to the distribution tree is
93 performed. PIM-SM [<a href="#ref-4" title="&quot;Protocol Independent Multicast-Sparse Mode (PIM-SM): Protocol Specification&quot;">4</a>] is an example of such a protocol.
94
95 The RGMP protocol specifies operations only for IP version 4
96 multicast routing. IP version 6 is not considered.
97
98 To keep RGMP simple, efficient and easy to implement, it is designed
99 for switches to expect RGMP messages from only one source per port.
100 For this reason, RGMP only supports a single RGMP enabled router to
101 be connected directly to a port of an RGMP enabled switch. Such a
102 topology should be customary when connecting routers to backbone
103 switches and thus not pose a limitation on the deployment of RGMP.
104
105
106
107
108 <span class="grey">Wu &amp; Eckert Informational [Page 2]</span></pre>
109 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-3" ></span>
110 <span class="grey"><a href="./rfc3488">RFC 3488</a> Cisco Systems RGMP February 2003</span>
111
112
113 All RGMP messages have the following format:
114
115 0 1 2 3
116 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
117 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
118 | Type | Reserved | Checksum |
119 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
120 | Group Address |
121 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
122
123 The reserved field in the message MUST be transmitted as zeros and
124 ignored on receipt.
125
126 <span class="h3"><a class="selflink" id="section-2.1" href="#section-2.1">2.1</a> Type</span>
127
128 There are four types of RGMP messages of concern to the
129 router-switch interaction. The type codes are defined to be the
130 highest values in an octet to avoid the re-use of already assigned
131 IGMP type codes.
132
133 0xFF = Hello
134 0xFE = Bye
135 0xFD = Join a group
136 0xFC = Leave a group
137
138 Unrecognized message types should be silently ignored.
139
140 Note:
141
142 RGMP and the IANA assignment of address 224.0.0.25 for it predates
143 <a href="./rfc3228">RFC 3228</a> [<a href="#ref-9" title="&quot;IANA Considerations for IPv4 Internet Group Management Protocol (IGMP)&quot;">9</a>]. RGMP defines Type values which in <a href="./rfc3228">RFC 3228</a> are
144 assigned to protocol testing and experimentation. This is not an
145 operational issue for RGMP itself because only RGMP packets use the
146 IPv4 destination address 224.0.0.25. The Type values defined above
147 are thus ONLY valid in conjunction with the RGMP destination address.
148
149 <span class="h3"><a class="selflink" id="section-2.2" href="#section-2.2">2.2</a>. Checksum</span>
150
151 Checksum covers the RGMP message (the entire IPv4 payload). The
152 algorithm and handling of checksum are the same as those for IGMP
153 messages as described in <a href="./rfc3376">RFC 3376</a> [<a href="#ref-5" title="&quot;Internet Group Management Protocol, Version 3&quot;">5</a>].
154
155
156
157
158
159
160
161
162
163
164 <span class="grey">Wu &amp; Eckert Informational [Page 3]</span></pre>
165 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-4" ></span>
166 <span class="grey"><a href="./rfc3488">RFC 3488</a> Cisco Systems RGMP February 2003</span>
167
168
169 <span class="h3"><a class="selflink" id="section-2.3" href="#section-2.3">2.3</a>. Group Address</span>
170
171 In an RGMP Hello or Bye message, the group address field is set to
172 zero.
173
174 In an RGMP Join or Leave message, the group address field holds the
175 IPv4 multicast group address of the group being joined or left.
176
177 <span class="h3"><a class="selflink" id="section-2.4" href="#section-2.4">2.4</a> IPv4 header</span>
178
179 RGMP messages are sent by routers to switches. The source IPv4
180 address of an RGMP packet is the sending interface IPv4 address of
181 the originating router. The destination IPv4 address of an RGMP
182 packet is 224.0.0.25. Switches supporting RGMP need to listen to
183 packets to this group.
184
185 <span class="h2"><a class="selflink" id="section-3" href="#section-3">3</a>. RGMP Protocol Description</span>
186
187 <span class="h3"><a class="selflink" id="section-3.1" href="#section-3.1">3.1</a> RGMP Router side Protocol Description</span>
188
189 Backbone switches use RGMP to learn which groups are desired at each
190 of their ports. Multicast routers use RGMP to pass such information
191 to the switches. Only routers send RGMP messages. They ignore
192 received RGMP messages.
193
194 A Router enabled for RGMP on an interface periodically [Hello
195 Interval] sends an RGMP Hello message to the attached network to
196 indicate that it is RGMP enabled. When RGMP is disabled on a routers
197 interface, it will send out an RGMP Bye message on that interface,
198 indicating that it again wishes to receive IPv4 multicast traffic
199 promiscuously from that interface.
200
201 When an interface is RGMP enabled, a router sends an RGMP Join
202 message out through this interface to each group that it wants to
203 receive traffic for from the interface. The router needs to
204 periodically [Join Interval] re-send an RGMP Join for a group to
205 indicate its continued desire to receive multicast traffic.
206
207 Routers supporting RGMP MUST NOT send RGMP Join or Leave messages for
208 groups 224.0.0.x (x=0...255), 224.0.1.39 and 224.0.1.40. The latter
209 two are known as cisco-rp-announce and cisco-rp-discovery [<a href="#ref-3">3</a>].
210
211 When a router no longer needs to receive traffic for a particular
212 group, it sends an RGMP Leave message for the group. For robustness,
213 the router MAY send more than one such message.
214
215
216
217
218
219
220 <span class="grey">Wu &amp; Eckert Informational [Page 4]</span></pre>
221 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-5" ></span>
222 <span class="grey"><a href="./rfc3488">RFC 3488</a> Cisco Systems RGMP February 2003</span>
223
224
225 If IPv4 multicast packets for an undesired group are received at a
226 router from a switch, the router MAY send a RGMP Leave message for
227 that group to the switch. These messages are called data-triggered
228 RGMP Leave messages and the router SHOULD rate-limit them. The
229 router MAY suppress sending a data triggered RGMP Leave message if it
230 has a desired group that has the same MAC destination address as the
231 undesired group. (See <a href="./rfc1112">RFC 1112</a> [<a href="#ref-6" title="&quot;Host Extensions for IP Multicasting&quot;">6</a>] for MAC ambiguity.) Such
232 suppression of data triggered RGMP Leave messages SHOULD be
233 configurable if supported.
234
235 <span class="h3"><a class="selflink" id="section-3.2" href="#section-3.2">3.2</a> RGMP Switch side Protocol Description</span>
236
237 A switch enabled for RGMP on a network consumes RGMP messages
238 received from ports of the network and processes them as described
239 below. If enabled for RGMP, the switch must NOT forward/flood
240 received RGMP messages out to other ports of the network.
241
242 RGMP on a switch operates on a per port basis, establishing per-group
243 forwarding state on RGMP enabled ports. A port reverts into an RGMP
244 enabled port upon receipt of an RGMP Hello message on the port, and a
245 timer [5 * Hello Interval] is started. This timer is restarted by
246 each RGMP Hello message arriving on the port. If this timer expires
247 or if it is removed by the arrival of an RGMP Bye message, then the
248 port reverts to its prior state of multicast traffic forwarding.
249
250 Correct deployment of RGMP is one RGMP enabled router directly
251 connected to a port on a switch that supports RGMP. The port on the
252 switch MAY want to keep track of the IPv4 originator address of the
253 RGMP Hello and Bye messages it receives on that port. In the event
254 it receives multiple IPv4 originating addresses in RGMP messages on
255 one port, the switch MAY generate an alert to notify the
256 administrator. The switch MAY also have a configuration option that
257 will allow for the operator to disable RGMP and have the switch fall
258 back to flooding IPv4 multicast on that port, although this is a
259 potentially dangerous option.
260
261 By default, connecting two or more RGMP enabled routers to a switch
262 port will cause intermittent black holing of IPv4 multicast traffic
263 towards these routers. Black holing occurs when a RGMP Leave is
264 received from one router while the other router is still joined.
265
266 This malfunction is not only easily recognized by the actual users
267 connected through the routers, but it also adheres to the principle
268 that a failure situation causes less traffic than more. Reverting to
269 flooding easily maintains the illusion that everything is working
270 perfectly. The exception is that the traffic constraining benefits
271
272
273
274
275
276 <span class="grey">Wu &amp; Eckert Informational [Page 5]</span></pre>
277 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-6" ></span>
278 <span class="grey"><a href="./rfc3488">RFC 3488</a> Cisco Systems RGMP February 2003</span>
279
280
281 of RGMP are not realized. This suggests that congestion happens at a
282 much later time than the misconfiguration and can then not easily be
283 correlated with the cause anymore.
284
285 Because routers supporting RGMP are not required to send RGMP Join or
286 Leave messages for groups 224.0.0.x (x=0...255), 224.0.1.39 and
287 224.0.1.40, RGMP enabled ports always need to receive traffic for
288 these groups. Traffic for other groups is initially not forwarded to
289 an RGMP enabled port.
290
291 RGMP Join and Leave messages are accepted if they arrive on an RGMP
292 enabled port, otherwise they will be discarded. Upon acceptance of
293 an RGMP Join message, the switch MUST start forwarding traffic for
294 the group to the port. Upon acceptance of an RGMP Leave message, the
295 switch SHOULD stop forwarding traffic for the group to that port.
296 The switch's ability to stop forwarding traffic for a group may be
297 limited, for example, because of destination MAC based forwarding in
298 the switch. Therefore, it is necessary for the switch to always
299 forward traffic for all MAC-ambiguous IPv4 multicast groups (see [<a href="#ref-6" title="&quot;Host Extensions for IP Multicasting&quot;">6</a>]
300 for MAC-ambiguity).
301
302 To stop forwarding of traffic to a group in the event of lost RGMP
303 Leave message(s), a switch MAY time out RGMP forwarding state on a
304 port for a group [5 * Join Interval] after the last RGMP Join for
305 that group has been received on the port.
306
307 Without any layer 2 IPv4 multicast filtering methods running, a
308 switch needs to flood multicast traffic to all ports. If a switch
309 does actually run one or more mechanisms beside RGMP to filter IPv4
310 multicast traffic, such as IGMP snooping [<a href="#ref-10" title="&quot;IGMP and MLD snooping switches&quot;">10</a>], then by default it
311 will not flood IPv4 multicast traffic to all ports anymore. Instead,
312 the switch will try to determine which ports still needs to receive
313 all IPv4 multicast traffic by default, and which ports do not.
314
315 Compliance with this specification requires that a switch MUST be
316 able to elect a port for flooding through the presence of PIM Hello
317 messages [<a href="#ref-4" title="&quot;Protocol Independent Multicast-Sparse Mode (PIM-SM): Protocol Specification&quot;">4</a>] arriving from the port and also through a manual
318 configuration option. In addition, the switch SHOULD recognize a
319 port connected to a router by other appropriate protocol packets or
320 dedicated IPv4 multicast router discovery mechanisms such as MRDISC
321 [<a href="#ref-11" title="&quot;IGMP Multicast Router Discovery&quot;">11</a>]. The manual configuration is required to support routers not
322 supporting PIM or other methods recognized by the switch.
323
324 Further mechanisms for IPv4 multicast traffic restriction may also be
325 used on RGMP enabled ports. In this case, forwarding for a group on
326 the port must be established if either mechanism requires it, and it
327 must only be removed if no mechanism requires it anymore.
328
329
330
331
332 <span class="grey">Wu &amp; Eckert Informational [Page 6]</span></pre>
333 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-7" ></span>
334 <span class="grey"><a href="./rfc3488">RFC 3488</a> Cisco Systems RGMP February 2003</span>
335
336
337 <span class="h2"><a class="selflink" id="section-4" href="#section-4">4</a>. Operational Notes</span>
338
339 <span class="h3"><a class="selflink" id="section-4.1" href="#section-4.1">4.1</a>. Support for networks with multiple switches</span>
340
341 To be simple to implement on switches and resilient in face of
342 potential layer 2 network topology changes, RGMP does not specify how
343 to restrict multicast traffic on links connecting switches amongst
344 each other. With just RGMP being used, multicast traffic will thus
345 be flooded on inter-switch links within a network if at least one
346 router is connected to each of the switches.
347
348 This happens implicitly because the switch will not flood/forward
349 received RGMP messages out to the inter-switch link and thus the
350 switch on the other end will only recognize the port as a router port
351 via the PIM Hello messages flooded by the switches. Manual
352 configuration for inter-switch links may be required if non-PIM
353 routers are being used, depending on the other capabilities of the
354 switch.
355
356 If appropriate, a switch can send out RGMP messages on ports to make
357 it look like an RGMP enabled router to a potential switch at the
358 other end of the link. This would constrain IPv4 multicast traffic
359 between switches, but this type of "RGMP Spoofing" by the switch is
360 outside the scope of this specification.
361
362 <span class="h3"><a class="selflink" id="section-4.2" href="#section-4.2">4.2</a>. Interoperability with RGMP-incapable routers</span>
363
364 Since RGMP messages received at a switch only affect the state of
365 their ingress ports, the traffic restriction is applied there only.
366 RGMP-incapable routers will receive multicast traffic for all
367 multicast groups.
368
369 <span class="h3"><a class="selflink" id="section-4.3" href="#section-4.3">4.3</a>. RGMP and multicast routing protocols</span>
370
371 One result of the simplicity of RGMP are its restrictions in
372 supporting specific routing protocols. The following paragraphs list
373 a few known restrictions.
374
375 A router running RGMP on a switched network will not receive traffic
376 for a multicast group unless it explicitly requests it via RGMP Join
377 messages (besides those group ranges specified to be flooded in 3.1).
378 For this reason, it is not possible to run a protocol like PIM
379 Dense-Mode or DVMRP across an RGMP enabled network with RGMP enabled
380 routers.
381
382
383
384
385
386
387
388 <span class="grey">Wu &amp; Eckert Informational [Page 7]</span></pre>
389 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-8" ></span>
390 <span class="grey"><a href="./rfc3488">RFC 3488</a> Cisco Systems RGMP February 2003</span>
391
392
393 In Bidir-PIM, a router elected to be the DF must not be enabled for
394 RGMP on the network, because it unconditionally needs to forward
395 traffic received from the network towards the RP. If a router is not
396 the DF for any group on the network, it can be enabled for RGMP on
397 that network.
398
399 In PIM-SM, directly connected sources on the network can not be
400 supported if the elected DR is running RGMP, because this DR needs to
401 unconditionally receive traffic from directly connected sources to
402 trigger the PIM-SM registering process on the DR. In PIM-SSM,
403 directly connected sources can be supported with RGMP enabled
404 routers.
405
406 Both in PIM-SM and PIM-SSM, upstream routers forwarding traffic into
407 the switched network need to send RGMP Joins for the group in support
408 of the PIM assert process.
409
410 <span class="h2"><a class="selflink" id="section-5" href="#section-5">5</a>. List of timers and default values</span>
411
412 <span class="h3"><a class="selflink" id="section-5.1" href="#section-5.1">5.1</a>. Hello Interval</span>
413
414 The Hello Interval is the interval between RGMP Hello messages sent
415 by an RGMP-enabled router to an RGMP-enabled switch. Default: 60
416 seconds.
417
418 <span class="h3"><a class="selflink" id="section-5.2" href="#section-5.2">5.2</a>. Join Interval</span>
419
420 The Join Interval is the interval between periodic RGMP Join messages
421 sent by an RGMP-enabled router to an RGMP-enabled switch for a given
422 group address. Default: 60 seconds.
423
424 <span class="h2"><a class="selflink" id="section-6" href="#section-6">6</a>. Security Considerations</span>
425
426 The RGMP protocol assumes that physical port security can be
427 guaranteed for switch ports from which RGMP messages are accepted.
428 Physical port security for RGMP means that physical measures will
429 ensure that such ports are dedicatedly connected to one system which
430 acts as an RGMP capable router. This is also the recommended
431 configuration to best leverage the benefits of the RGMP protocol
432 (e.g., avoiding unwanted third-party IPv4 multicast traffic arriving
433 on said ports).
434
435 RGMP specific DoS attacks arise from forged RGMP messages. If more
436 than one system is connected to a port of the RGMP switch, then one
437 system may forge RGMP messages and affect the operations of the other
438 system(s) on the same port. This is a potential security risk.
439
440
441
442
443
444 <span class="grey">Wu &amp; Eckert Informational [Page 8]</span></pre>
445 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-9" ></span>
446 <span class="grey"><a href="./rfc3488">RFC 3488</a> Cisco Systems RGMP February 2003</span>
447
448
449 When physical security ensures that only one system is connected to a
450 RGMP capable port on a switch, then forged messages from this system
451 itself can take effect. Such forged messages can always be avoided
452 by system local measures.
453
454 We consider the ramifications of a forged message of each type:
455
456 Hello Message:
457
458 A forged RGMP Hello message can restrict multicast data towards a
459 non-RGMP enabled router on the same port. This effectively
460 introduces a blackholing DoS attack.
461
462 Leave Message:
463
464 A forged RGMP Leave message can restrict IPv4 multicast traffic
465 for individual groups toward the port. The effect is a possible
466 blackholing DoS attack similar to an RGMP Hello Message except
467 that it does not affect all IPv4 multicast traffic but only that
468 of the groups indicated in the forged messages. It will also only
469 affect a port if there officially is only one RGMP enabled router
470 connected to it (i.e., if the port is RGMP enabled).
471
472 Bye Message:
473
474 A forged RGMP Bye message can turn the port into being
475 RGMP-disabled. This could, indirectly, cause a DoS attack based
476 on the port getting overloaded with IPv4 multicast traffic if the
477 network bandwidth of the port was provisioned with the expectation
478 that RGMP will suppress unwanted IPv4 multicast messages.
479
480 This type of DoS attack simply re-establishes a port behavior as
481 if RGMP was not configured and invalidates the benefit of RGMP.
482 This, however, does not introduce an issue that would not have
483 been there without RGMP in the first place.
484
485 Join Message:
486
487 A forged RGMP Join message could attract undesired multicast
488 packets to the port where it is received from. The effect is
489 similar to an RGMP Bye Message except that it does not affect all
490 IPv4 multicast traffic only the groups indicated in the forged
491 messages. The message will affect a port only if there officially
492 is only one RGMP enabled router connected to it (i.e., if the port
493 is RGMP enabled).
494
495
496
497
498
499
500 <span class="grey">Wu &amp; Eckert Informational [Page 9]</span></pre>
501 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-10" ></span>
502 <span class="grey"><a href="./rfc3488">RFC 3488</a> Cisco Systems RGMP February 2003</span>
503
504
505 <span class="h2"><a class="selflink" id="section-7" href="#section-7">7</a>. Normative References</span>
506
507 [<a id="ref-1">1</a>] Bradner, S., "The Internet Standards Process -- Revision 3", <a href="https://www.rfc-editor.org/bcp/bcp9">BCP</a>
508 <a href="https://www.rfc-editor.org/bcp/bcp9">9</a>, <a href="./rfc2026">RFC 2026</a>, October 1996.
509
510 [<a id="ref-2">2</a>] Bradner, S., "Key words for use in RFCs to Indicate Requirement
511 Levels", <a href="https://www.rfc-editor.org/bcp/bcp14">BCP 14</a>, <a href="./rfc2119">RFC 2119</a>, March 1997.
512
513 [<a id="ref-4">4</a>] Estrin, D., Farinacci, D., Helmy, A., Thaler, D., Deering, S.,
514 Handley, M., Jacobson, V., Liu, C., Sharma, P. and L. Wei,
515 "Protocol Independent Multicast-Sparse Mode (PIM-SM): Protocol
516 Specification", <a href="./rfc2362">RFC 2362</a>, June 1998.
517
518 [<a id="ref-5">5</a>] Cain, B., Deering, S., Kouvelas, I., Fenner, W. and A.
519 Thyagarajan, "Internet Group Management Protocol, Version 3",
520 <a href="./rfc3376">RFC 3376</a>, October 2002.
521
522 [<a id="ref-6">6</a>] Deering, S., "Host Extensions for IP Multicasting", STD 5, <a href="./rfc1112">RFC</a>
523 <a href="./rfc1112">1112</a>, August 1989.
524
525 [<a id="ref-7">7</a>] ANSI/IEEE Std 802.1D 1998 Edition, "Media Access Control (MAC)
526 Bridges", 1998.
527
528 <span class="h2"><a class="selflink" id="section-8" href="#section-8">8</a>. Informative References</span>
529
530 [<a id="ref-3">3</a>] Internet Multicast Addresses,
531 <a href="http://www.iana.org/assignments/multicast-addresses">http://www.iana.org/assignments/multicast-addresses</a>
532
533 [<a id="ref-8">8</a>] Farinacci D., Tweedly D., Speakman T., "Cisco Group Management
534 Protocol (CGMP)", 1996/1997
535 <a href="ftp://ftpeng.cisco.com/ipmulticast/specs/cgmp.txt">ftp://ftpeng.cisco.com/ipmulticast/specs/cgmp.txt</a>
536
537 [<a id="ref-9">9</a>] Fenner, B., "IANA Considerations for IPv4 Internet Group
538 Management Protocol (IGMP)", <a href="./rfc3228">RFC 3228</a>, February 2002.
539
540 [<a id="ref-10">10</a>] Christensen, M. and F. Solensky, "IGMP and MLD snooping
541 switches", Work In Progress.
542
543 [<a id="ref-11">11</a>] Biswas, S., Cain, B. and B. Haberman, "IGMP Multicast Router
544 Discovery", Work In Progress.
545
546 <span class="h2"><a class="selflink" id="section-9" href="#section-9">9</a>. Acknowledgments</span>
547
548 The authors would like to thank Gorry Fairhurst, Bill Fenner,
549 Giovanni Meo, Mike Norton, Pavlin Radoslavov and Alex Zinin for their
550 review of the document and their suggestions.
551
552
553
554
555
556 <span class="grey">Wu &amp; Eckert Informational [Page 10]</span></pre>
557 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-11" ></span>
558 <span class="grey"><a href="./rfc3488">RFC 3488</a> Cisco Systems RGMP February 2003</span>
559
560
561 <span class="h2"><a class="selflink" id="appendix-A" href="#appendix-A">Appendix A</a>. Intellectual Property Rights</span>
562
563 The IETF has been notified of intellectual property rights claimed in
564 regard to some or all of the specification contained in this
565 document. For more information consult the online list of claimed
566 rights.
567
568 <span class="h2"><a class="selflink" id="appendix-B" href="#appendix-B">Appendix B</a>. Comparison with GARP/GMRP</span>
569
570 This appendix is not part of the RGMP specification but is provided
571 for information only.
572
573 GARP/GMRP (defined in IEEE 802.1D [<a href="#ref-7" title="&quot;Media Access Control (MAC) Bridges&quot;">7</a>]) is the ANSI/ISO/IEC/IEEE
574 protocol suite to constrain ethernet multicast traffic in bridged
575 ethernet networks. As such it is also a possible alternative to RGMP
576 for the purpose of constraining multicast traffic towards router
577 ports. This appendix will explain the motivation not to rely on
578 GARP/GMRP and how GARP/GMRP and RGMP differ.
579
580 The key factor in rolling out GARP/GMRP would have been to completely
581 replace IGMP Snooping. This was the design goal of GARP/GMRP. For
582 efficient operations, IGMP Snooping requires hardware filtering
583 support in the switch (to differentiate between hosts membership
584 reports and actual IPv4 multicast traffic). Especially in many older
585 switches this support does not exist. Vendors tried to find a way
586 around this issue to provide the benefit of constraining IPv4
587 multicast traffic in a switched LAN without having to build more
588 expensive switch hardware. GARP/GMRP is one protocol resulting from
589 this. CGMP from Cisco is another one. While CGMP solves the problem
590 without requiring changes to the host stack software, GARP/GMRP
591 requires support for it by the host stack.
592
593 Up to date GARP/GMRP has so far not made significant inroads into
594 deployed solutions. IGMP Snooping (and CGMP) are the norm for this
595 environment. In result, GARP/GMRP can not necessarily be expected to
596 be supported by layer 2 switches. In addition, GARP/GMRP does not
597 address clearly the issues RGMP tries to solve. On one hand,
598 GARP/GMRP provides much more functionality and as such complexity as
599 immediately required. On the other hand, GARP/GMRP is limited by
600 being a standard predominantly for the Ethernet scope.
601
602
603
604
605
606
607
608
609
610
611
612 <span class="grey">Wu &amp; Eckert Informational [Page 11]</span></pre>
613 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-12" ></span>
614 <span class="grey"><a href="./rfc3488">RFC 3488</a> Cisco Systems RGMP February 2003</span>
615
616
617 Beyond the process and applicability reasons, the main differences
618 between GARP/GMRP and RGMP are as follows:
619
620 o GARP/GMRP switches/systems need to send and listen/react to
621 GARP/GMRP messages. In RGMP, routers only need to send RGMP
622 messages and switches only need to listen to them. This protocol
623 approach is meant to simplify implementation, operations and
624 troubleshooting.
625
626 o The same switch running RGMP in a backbone network will likely see
627 more states then running on the edge only doing IGMP Snooping,
628 making it preferable to keep the amount of per group processing
629 and memory requirements in RGMP more in bounds than possible in
630 IGMP Snooping and GARP/GMRP: In GARP/GMRP, a (multiple) timer
631 based state-machines needs to be maintained on a per ethernet
632 group address, in RGMP timer maintenance is completely optional
633 and there are only two states per group (joined or not joined).
634
635 o GARP/GMRP is an ethernet level protocol from the IEEE. It
636 supports to constrain traffic for ethernet addresses (groups).
637 RGMP does constrain traffic for IPv4 multicast groups. Today this
638 is even beyond the capabilities of typical switch platforms used
639 as layer2 switches. Extensions to support further entities are
640 likely easier to come by through extensions to RGMP than to
641 GARP/GMRP.
642
643 o RGMP shares the basic packet format with IGMP (version 2) and is
644 as such easy to add to router and switch platforms that already
645 support IGMP and IGMP Snooping respectively. This is especially
646 true for switches that in hardware can differentiate between IGMP
647 protocol type packets and other IPv4 multicast traffic sent to the
648 same (or a MAC ambiguous) group. In addition, due to the state
649 simplicity of RGMP it is easy to integrate IGMP Snooping and RGMP
650 operations in the IPv4 multicast control and forwarding plane of a
651 switch.
652
653 o GARP/GMRP supports more than one system (host/router) on a switch
654 port which is one reason for its complexity. In RGMP, this
655 configuration is explicitly not supported: More than one router
656 per switched port is not only not a common scenario in today's
657 switches layer 2 networks, it is also an undesired configuration
658 when unwanted IPv4 multicast traffic is to be kept away from
659 routers.
660
661
662
663
664
665
666
667
668 <span class="grey">Wu &amp; Eckert Informational [Page 12]</span></pre>
669 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-13" ></span>
670 <span class="grey"><a href="./rfc3488">RFC 3488</a> Cisco Systems RGMP February 2003</span>
671
672
673 o GARP/GMRP defines how to constrain multicast traffic between
674 switches, another reason for its complexity. RGMP does not
675 explicitly support this as part of the protocol because of the
676 following reasons:
677
678 o It is not necessary to include this function as part of the
679 RGMP protocol description because switch implementations can
680 transparently decide to support this function (see 4.1 about
681 this "RGMP Spoofing").
682
683 o Important deployments through which large amounts of IPv4
684 multicast are moved today are typically single switch
685 MIX - Multicast Internet eXchange points.
686
687 o Avoiding congestion on inter-switch links in general is more
688 complex than simply constraining IPv4 multicast traffic to
689 paths where it is needed. With or without IPv4 multicast, the
690 aggregate bandwidth needed between switches can easily be the
691 aggregate required bandwidth to routers on either sides. For
692 this reason, inter-switch bandwidth is most often appropriately
693 over provisioned. In addition, the likelihood for receiving
694 routers to be only on the sources side of an inter-switch link
695 is in general deployments rather low. The cases where traffic
696 constrainment on inter-switch links is required and helpful is
697 thus limited and can in most cases be avoided or worked around.
698 Moving the network to a layer 3 routed network is often the
699 best solution, supporting RGMP-Spoofing (see <a href="#section-4.1">section 4.1</a>) is
700 another one.
701
702 <span class="h2"><a class="selflink" id="appendix-C" href="#appendix-C">Appendix C</a>. Possible future extensions / comparison to PIM Snooping</span>
703
704 This appendix is not part of the RGMP specification but is provided
705 for information only.
706
707 This appendix presents a discussion of possible extensions to RGMP.
708 Included are points on why the extensions are not included and in
709 addition a motivation for RGMP in comparison to (PIM) snooping.
710
711 o Support for multiple switches
712
713 As discussed in "RGMP Spoofing", chapter 4.1 and GARP/GMRP
714 comparison in <a href="#appendix-B">Appendix B</a>.
715
716 o Support for SSM
717
718 While RGMP works with PIM-SSM, it does not have explicit messages
719 for the router to selectively join to (S,G) channels individually.
720 Instead the router must RGMP join to all (Si,G) channels by
721
722
723
724 <span class="grey">Wu &amp; Eckert Informational [Page 13]</span></pre>
725 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-14" ></span>
726 <span class="grey"><a href="./rfc3488">RFC 3488</a> Cisco Systems RGMP February 2003</span>
727
728
729 joining to G. Extending RGMP to include (S,G) Join/Leaves is
730 feasible. However, currently the majority of switches do not
731 support actual traffic constraining on a per channel basis. In
732 addition, the likelihood for actual channel collision (two SSM
733 channels using the same group) will only become an issue when SSM
734 is fully deployed.
735
736 o Support for IPv6
737
738 RGMP could easily be extended to support IPv6 by mapping the RGMP
739 packet format into the MLD/IPv6 packet format. This was not done
740 for this specification because most switches today do not even
741 support MLD snooping.
742
743 o Support for multiple routers per port
744
745 As discussed in <a href="#appendix-B">Appendix B</a>. This is probably one extension that
746 should be avoided. Multiple RGMP router per port are
747 inappropriate for efficient multicast traffic constrainment.
748
749 o Support for non-join based protocols / protocol elements
750
751 For protocols like PIM dense-mode, DVMRP or Bidir-PIM DF routers,
752 additional RGMP messages may be added to allow routers to indicate
753 that certain group (ranges) traffic need to be flooded from
754 (dense-mode) or to (Bidir-PIM) them.
755
756 o Support for multi-policy switching
757
758 In Multicast Exchange Points (MIXes) environments situations exist
759 where different downstream routers for policy reasons need to
760 receive the same traffic flow from different upstream routers.
761
762 This problem could be solved by actually providing an upstream
763 neighbor field in RGMP Join/Leave messages. The RGMP switch would
764 then forward traffic from one upstream router only to those
765 downstream routers who want to have the traffic from exactly this
766 upstream router. This extension would best go in hand with
767 changes to the layer 3 routing protocol run between the routers.
768
769 As previously mentioned, RGMP was designed to be easy to implement
770 and to support simple layer2 switches. Implementations could also be
771 applied to switches beyond layer 2. If all the above possible future
772 extensions were to be supported by an evolution of RGMP, it would be
773 questionable whether such a protocol could be any less complex than
774 actually snooping into the layer3 IPv4 routing protocol run between
775 routers in a switched LAN.
776
777
778
779
780 <span class="grey">Wu &amp; Eckert Informational [Page 14]</span></pre>
781 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-15" ></span>
782 <span class="grey"><a href="./rfc3488">RFC 3488</a> Cisco Systems RGMP February 2003</span>
783
784
785 From the perspective of protocol architecture it is certainly more
786 appropriate to have a separate protocol like RGMP or GARP/GMRP for
787 this purpose. Then again, the more complex the requirements are, the
788 more duplication of effort is involved and snooping seems to become a
789 more attractive option.
790
791 Even though there exists one predominant routing Protocol, PIM, in
792 IPv4 multicast, routing with PIM in itself is extremely complex for a
793 switch to snoop into. PIM has two main versions, different
794 modes - sparse, dense, bidir, ssm, join / prune / graft messages
795 (depending on the mode of the group), various PIM Hello options,
796 different versions of asserts, two dynamic mode announcement
797 protocols (BSR, AutoRP), and finally supports both IPv4 and IPv6.
798
799 A switch snooping into PIM is very likely to implement just a subset
800 of this feature set, making it very hard for the user to determine
801 what level of actual traffic constrainment is achieved unless a clear
802 specification exists for the implementation (or better the method per
803 se.). In addition, there is always the danger that such a snooping
804 implementation may break newer features of the routing protocol that
805 it was not designed to handle (likely because they could not have
806 been predicted). For example, this can happen with switches using
807 IGMP (v2) snooping implementations that are being subjected to IGMP
808 version 3 messages - they break IGMPv3.
809
810 In summary, with PIM still evolving, the approach taken by RGMP is
811 the safest one for the immediate problems at hand, and extensions
812 like those listed should be considered in time for actual demand.
813 (PIM) snooping is a valid alternative once the total amount of
814 features that need to be supported makes it an equally attractive
815 solution (with respect to complexity) to a dedicated protocol and if
816 its functions are well defined to allow predicting its effects - but
817 always at the price of possible incompatibilities with upcoming PIM
818 protocol extensions unless support for layer 2 switches is explicitly
819 considered in moving PIM protocols forward.
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836 <span class="grey">Wu &amp; Eckert Informational [Page 15]</span></pre>
837 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-16" ></span>
838 <span class="grey"><a href="./rfc3488">RFC 3488</a> Cisco Systems RGMP February 2003</span>
839
840
841 Authors' Addresses
842
843 Ishan Wu
844 cisco Systems
845 170 West Tasman Drive
846 San Jose, CA 95134
847
848 Phone: (408) 526-5673
849 EMail: [email protected]
850
851
852 Toerless Eckert
853 cisco Systems
854 170 West Tasman Drive
855 San Jose, CA 95134
856
857 Phone: (408) 853-5856
858 Email: [email protected]
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892 <span class="grey">Wu &amp; Eckert Informational [Page 16]</span></pre>
893 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-17" ></span>
894 <span class="grey"><a href="./rfc3488">RFC 3488</a> Cisco Systems RGMP February 2003</span>
895
896
897 Full Copyright Statement
898
899 Copyright (C) The Internet Society (2003). All Rights Reserved.
900
901 This document and translations of it may be copied and furnished to
902 others, and derivative works that comment on or otherwise explain it
903 or assist in its implementation may be prepared, copied, published
904 and distributed, in whole or in part, without restriction of any
905 kind, provided that the above copyright notice and this paragraph are
906 included on all such copies and derivative works. However, this
907 document itself may not be modified in any way, such as by removing
908 the copyright notice or references to the Internet Society or other
909 Internet organizations, except as needed for the purpose of
910 developing Internet standards in which case the procedures for
911 copyrights defined in the Internet Standards process must be
912 followed, or as required to translate it into languages other than
913 English.
914
915 The limited permissions granted above are perpetual and will not be
916 revoked by the Internet Society or its successors or assigns.
917
918 This document and the information contained herein is provided on an
919 "AS IS" basis and THE INTERNET SOCIETY AND THE INTERNET ENGINEERING
920 TASK FORCE DISCLAIMS ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING
921 BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION
922 HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF
923 MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.
924
925 Acknowledgement
926
927 Funding for the RFC Editor function is currently provided by the
928 Internet Society.
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948 Wu &amp; Eckert Informational [Page 17]
949 </pre>