Debian Codegraph @

rfc2814.html

raw
1 <pre>Network Working Group R. Yavatkar
2 Request for Comments: 2814 Intel
3 Category: Standards Track D. Hoffman
4 Teledesic
5 Y. Bernet
6 Microsoft
7 F. Baker
8 Cisco
9 M. Speer
10 Sun Microsystems
11 May 2000
12
13
14 <span class="h1">SBM (Subnet Bandwidth Manager):</span>
15 A Protocol for RSVP-based Admission Control over IEEE 802-style networks
16
17 Status of this Memo
18
19 This document specifies an Internet standards track protocol for the
20 Internet community, and requests discussion and suggestions for
21 improvements. Please refer to the current edition of the "Internet
22 Official Protocol Standards" (STD 1) for the standardization state
23 and status of this protocol. Distribution of this memo is unlimited.
24
25 Copyright Notice
26
27 Copyright (C) The Internet Society (2000). All Rights Reserved.
28
29 Abstract
30
31 This document describes a signaling method and protocol for RSVP-
32 based admission control over IEEE 802-style LANs. The protocol is
33 designed to work both with the current generation of IEEE 802 LANs as
34 well as with the recent work completed by the IEEE 802.1 committee.
35
36 <span class="h2"><a class="selflink" id="section-1" href="#section-1">1</a>. Introduction</span>
37
38 New extensions to the Internet architecture and service models have
39 been defined for an integrated services Internet [RFC-1633, <a href="./rfc2205">RFC-2205</a>,
40 <a href="./rfc2210">RFC-2210</a>] so that applications can request specific qualities or
41 levels of service from an internetwork in addition to the current IP
42 best-effort service. These extensions include RSVP, a resource
43 reservation setup protocol, and definition of new service classes to
44 be supported by Integrated Services routers. RSVP and service class
45 definitions are largely independent of the underlying networking
46 technologies and it is necessary to define the mapping of RSVP and
47 Integrated Services specifications onto specific subnetwork
48 technologies. For example, a definition of service mappings and
49
50
51
52 <span class="grey">Yavatkar, et al. Standards Track [Page 1]</span></pre>
53 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-2" ></span>
54 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
55
56
57 reservation setup protocols is needed for specific link-layer
58 technologies such as shared and switched IEEE-802-style LAN
59 technologies.
60
61 This document defines SBM, a signaling protocol for RSVP-based
62 admission control over IEEE 802-style networks. SBM provides a
63 method for mapping an internet-level setup protocol such as RSVP onto
64 IEEE 802 style networks. In particular, it describes the operation
65 of RSVP-enabled hosts/routers and link layer devices (switches,
66 bridges) to support reservation of LAN resources for RSVP-enabled
67 data flows. A framework for providing Integrated Services over
68 shared and switched IEEE-802-style LAN technologies and a definition
69 of service mappings have been described in separate documents [RFC-
70 FRAME, RFC-MAP].
71
72 <span class="h2"><a class="selflink" id="section-2" href="#section-2">2</a>. Goals and Assumptions</span>
73
74 The SBM (Subnet Bandwidth Manager) protocol and its use for admission
75 control and bandwidth management in IEEE 802 level-2 networks is
76 based on the following architectural goals and assumptions:
77
78 I. Even though the current trend is towards increased use of
79 switched LAN topologies consisting of newer switches that support
80 the priority queuing mechanisms specified by IEEE 802.1p, we
81 assume that the LAN technologies will continue to be a mix of
82 legacy shared/ switched LAN segments and newer switched segments
83 based on IEEE 802.1p specification. Therefore, we specify a
84 signaling protocol for managing bandwidth over both legacy and
85 newer LAN topologies and that takes advantage of the additional
86 functionality (such as an explicit support for different traffic
87 classes or integrated service classes) as it becomes available in
88 the new generation of switches, hubs, or bridges. As a result,
89 the SBM protocol would allow for a range of LAN bandwidth
90 management solutions that vary from one that exercises purely
91 administrative control (over the amount of bandwidth consumed by
92 RSVP-enabled traffic flows) to one that requires cooperation (and
93 enforcement) from all the end-systems or switches in a IEEE 802
94 LAN.
95
96 II. This document specifies only a signaling method and protocol
97 for LAN-based admission control over RSVP flows. We do not define
98 here any traffic control mechanisms for the link layer; the
99 protocol is designed to use any such mechanisms defined by IEEE
100 802. In addition, we assume that the Layer 3 end-systems (e.g., a
101 host or a router) will exercise traffic control by policing
102 Integrated Services traffic flows to ensure that each flow stays
103 within its traffic specifications stipulated in an earlier
104 reservation request submitted for admission control. This then
105
106
107
108 <span class="grey">Yavatkar, et al. Standards Track [Page 2]</span></pre>
109 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-3" ></span>
110 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
111
112
113 allows a system using SBM admission control combined with per flow
114 shaping at end systems and IEEE-defined traffic control at link
115 layer to realize some approximation of Controlled Load (and even
116 Guaranteed) services over IEEE 802-style LANs.
117
118 III. In the absence of any link-layer traffic control or priority
119 queuing mechanisms in the underlying LAN (such as a shared LAN
120 segment), the SBM-based admission control mechanism only limits
121 the total amount of traffic load imposed by RSVP-enabled flows on
122 a shared LAN. In such an environment, no traffic flow separation
123 mechanism exists to protect the RSVP-enabled flows from the best-
124 effort traffic on the same shared media and that raises the
125 question of the utility of such a mechanism outside a topology
126 consisting only of 802.1p-compliant switches. However, we assume
127 that the SBM-based admission control mechanism will still serve a
128 useful purpose in a legacy, shared LAN topology for two reasons.
129 First, assuming that all the nodes that generate Integrated
130 Services traffic flows utilize the SBM-based admission control
131 procedure to request reservation of resources before sending any
132 traffic, the mechanism will restrict the total amount of traffic
133 generated by Integrated Services flows within the bounds desired
134 by a LAN administrator (see discussion of the NonResvSendLimit
135 parameter in <a href="#appendix-C">Appendix C</a>). Second, the best-effort traffic
136 generated by the TCP/IP-based traffic sources is generally rate
137 adaptive (using a TCP-style "slow start" congestion avoidance
138 mechanism or a feedback-based rate adaptation mechanism used by
139 audio/video streams based on RTP/RTCP protocols) and adapts to
140 stay within the available network bandwidth. Thus, the
141 combination of admission control and rate adaptation should avoid
142 persistent traffic congestion. This does not, however, guarantee
143 that non-Integrated-Services traffic will not interfere with the
144 Integrated Services traffic in the absence of traffic control
145 support in the underlying LAN infrastructure.
146
147 <span class="h2"><a class="selflink" id="section-3" href="#section-3">3</a>. Organization of the rest of this document</span>
148
149 The rest of this document provides a detailed description of the
150 SBM-based admission control procedure(s) for IEEE 802 LAN
151 technologies. The document is organized as follows:
152
153 * <a href="#section-4">Section 4</a> first defines the various terms used in the document and
154 then provides an overview of the admission control procedure with
155 an example of its application to a sample network.
156
157 * <a href="#section-5">Section 5</a> describes the rules for processing and forwarding PATH
158 (and PATH_TEAR) messages at DSBMs (Designated Subnet Bandwidth
159 Managers), SBMs, and DSBM clients.
160
161
162
163
164 <span class="grey">Yavatkar, et al. Standards Track [Page 3]</span></pre>
165 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-4" ></span>
166 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
167
168
169 * <a href="#section-6">Section 6</a> addresses the inter-operability issues when a DSBM may
170 operate in the absence of RSVP signaling at Layer 3 or when
171 another signaling protocol (such as SNMP) is used to reserve
172 resources on a LAN segment.
173
174 * <a href="#appendix-A">Appendix A</a> describes the details of the DSBM election algorithm
175 used for electing a designated SBM on a LAN segment when more than
176 one SBM is present. It also describes how DSBM clients discover
177 the presence of a DSBM on a managed segment.
178
179 * <a href="#appendix-B">Appendix B</a> specifies the formats of SBM-specific messages used and
180 the formats of new RSVP objects needed for the SBM operation.
181
182 * <a href="#appendix-C">Appendix C</a> describes usage of the DSBM to distribute configuration
183 information to senders on a managed segment.
184
185 <span class="h2"><a class="selflink" id="section-4" href="#section-4">4</a>. Overview</span>
186
187 <span class="h3"><a class="selflink" id="section-4.1" href="#section-4.1">4.1</a>. Definitions</span>
188
189 - Link Layer or Layer 2 or L2: We refer to data-link layer
190 technologies such as IEEE 802.3/Ethernet as L2 or layer 2.
191
192 - Link Layer Domain or Layer 2 domain or L2 domain: a set of nodes
193 and links interconnected without passing through a L3 forwarding
194 function. One or more IP subnets can be overlaid on a L2 domain.
195
196 - Layer 2 or L2 devices: We refer to devices that only implement
197 Layer 2 functionality as Layer 2 or L2 devices. These include
198 802.1D bridges or switches.
199
200 - Internetwork Layer or Layer 3 or L3: Layer 3 of the ISO 7 layer
201 model. This document is primarily concerned with networks that use
202 the Internet Protocol (IP) at this layer.
203
204 - Layer 3 Device or L3 Device or End-Station: these include hosts
205 and routers that use L3 and higher layer protocols or application
206 programs that need to make resource reservations.
207
208 - Segment: A L2 physical segment that is shared by one or more
209 senders. Examples of segments include (a) a shared Ethernet or
210 Token-Ring wire resolving contention for media access using CSMA
211 or token passing ("shared L2 segment"), (b) a half duplex link
212 between two stations or switches, (c) one direction of a switched
213 full-duplex link.
214
215
216
217
218
219
220 <span class="grey">Yavatkar, et al. Standards Track [Page 4]</span></pre>
221 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-5" ></span>
222 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
223
224
225 - Managed segment: A managed segment is a segment with a DSBM
226 present and responsible for exercising admission control over
227 requests for resource reservation. A managed segment includes
228 those interconnected parts of a shared LAN that are not separated
229 by DSBMs.
230
231 - Traffic Class: An aggregation of data flows which are given
232 similar service within a switched network.
233
234 - User_priority: User_priority is a value associated with the
235 transmission and reception of all frames in the IEEE 802 service
236 model: it is supplied by the sender that is using the MAC service.
237 It is provided along with the data to a receiver using the MAC
238 service. It may or may not be actually carried over the network:
239 Token-Ring/802.5 carries this value (encoded in its FC octet),
240 basic Ethernet/802.3 does not, 802.12 may or may not depending on
241 the frame format in use. 802.1p defines a consistent way to carry
242 this value over the bridged network on Ethernet, Token Ring,
243 Demand-Priority, FDDI or other MAC-layer media using an extended
244 frame format. The usage of user_priority is fully described in
245 <a href="#section-2.5">section 2.5</a> of 802.1D [<a href="#ref-IEEE8021D" title="&quot;MAC Bridges&quot;">IEEE8021D</a>] and 802.1p [IEEE8021P] "Support
246 of the Internal Layer Service by Specific MAC Procedures".
247
248 - Subnet: used in this memo to indicate a group of L3 devices
249 sharing a common L3 network address prefix along with the set of
250 segments making up the L2 domain in which they are located.
251
252 - Bridge/Switch: a layer 2 forwarding device as defined by IEEE
253 802.1D. The terms bridge and switch are used synonymously in this
254 document.
255
256 - DSBM: Designated SBM (DSBM) is a protocol entity that resides in a
257 L2 or L3 device and manages resources on a L2 segment. At most one
258 DSBM exists for each L2 segment.
259
260 - SBM: the SBM is a protocol entity that resides in a L2 or L3
261 device and is capable of managing resources on a segment. However,
262 only a DSBM manages the resources for a managed segment. When more
263 than one SBM exists on a segment, one of the SBMs is elected to be
264 the DSBM.
265
266 - Extended segment: An extended segment includes those parts of a
267 network which are members of the same IP subnet and therefore are
268 not separated by any layer 3 devices. Several managed segments,
269 interconnected by layer 2 devices, constitute an extended segment.
270
271
272
273
274
275
276 <span class="grey">Yavatkar, et al. Standards Track [Page 5]</span></pre>
277 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-6" ></span>
278 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
279
280
281 - Managed L2 domain: An L2 domain consisting of managed segments is
282 referred to as a managed L2 domain to distinguish it from a L2
283 domain with no DSBMs present for exercising admission control over
284 resources at segments in the L2 domain.
285
286 - DSBM clients: These are entities that transmit traffic onto a
287 managed segment and use the services of a DSBM for the managed
288 segment for admission control over a LAN segment. Only the layer 3
289 or higher layer entities on L3 devices such as hosts and routers
290 are expected to send traffic that requires resource reservations,
291 and, therefore, DSBM clients are L3 entities.
292
293 - SBM transparent devices: A "SBM transparent" device is unaware of
294 SBMs or DSBMs (though it may or may not be RSVP aware) and,
295 therefore, does not participate in the SBM-based admission control
296 procedure over a managed segment. Such a device uses standard
297 forwarding rules appropriate for the device and is transparent
298 with respect to SBM. An example of such a L2 device is a legacy
299 switch that does not participate in resource reservation.
300
301 - Layer 3 and layer 2 addresses: We refer to layer 3 addresses of
302 L3/L2 devices as "L3 addresses" and layer 2 addresses as "L2
303 addresses". This convention will be used in the rest of the
304 document to distinguish between Layer 3 and layer 2 addresses used
305 to refer to RSVP next hop (NHOP) and previous hop (PHOP) devices.
306 For example, in conventional RSVP message processing, RSVP_HOP
307 object in a PATH message carries the L3 address of the previous
308 hop device. We will refer to the address contained in the RSVP_HOP
309 object as the RSVP_HOP_L3 address and the corresponding MAC
310 address of the previous hop device will be referred to as the
311 RSVP_HOP_L2 address.
312
313 <span class="h3"><a class="selflink" id="section-4.2" href="#section-4.2">4.2</a>. Overview of the SBM-based Admission Control Procedure</span>
314
315 A protocol entity called "Designated SBM" (DSBM) exists for each
316 managed segment and is responsible for admission control over the
317 resource reservation requests originating from the DSBM clients in
318 that segment. Given a segment, one or more SBMs may exist on the
319 segment. For example, many SBM-capable devices may be attached to a
320 shared L2 segment whereas two SBM-capable switches may share a half-
321 duplex switched segment. In that case, a single DSBM is elected for
322 the segment. The procedure for dynamically electing the DSBM is
323 described in <a href="#appendix-A">Appendix A</a>. The only other approved method for
324 specifying a DSBM for a managed segment is static configuration at
325 SBM-capable devices.
326
327
328
329
330
331
332 <span class="grey">Yavatkar, et al. Standards Track [Page 6]</span></pre>
333 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-7" ></span>
334 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
335
336
337 The presence of a DSBM makes the segment a "managed segment".
338 Sometimes, two or more L2 segments may be interconnected by SBM
339 transparent devices. In that case, a single DSBM will manage the
340 resources for those segments treating the collection of such segments
341 as a single managed segment for the purpose of admission control.
342
343 <span class="h4"><a class="selflink" id="section-4.2.1" href="#section-4.2.1">4.2.1</a>. Basic Algorithm</span>
344
345 Figure 1 - An Example of a Managed Segment.
346
347 +-------+ +-----+ +------+ +-----+ +--------+
348 |Router | | Host| | DSBM | | Host| | Router |
349 | R2 | | C | +------+ | B | | R3 |
350 +-------+ +-----+ / +-----+ +--------+
351 | | / | |
352 | | / | |
353 ==============================================================LAN
354 | |
355 | |
356 +------+ +-------+
357 | Host | | Router|
358 | A | | R1 |
359 +------+ +-------+
360
361 Figure 1 shows an example of a managed segment in a L2 domain that
362 interconnects a set of hosts and routers. For the purpose of this
363 discussion, we ignore the actual physical topology of the L2 domain
364 (assume it is a shared L2 segment and a single managed segment
365 represents the entire L2 domain). A single SBM device is designated
366 to be the DSBM for the managed segment. We will provide examples of
367 operation of the DSBM over switched and shared segments later in the
368 document.
369
370 The basic DSBM-based admission control procedure works as follows:
371
372 1. DSBM Initialization: As part of its initial configuration, DSBM
373 obtains information such as the limits on fraction of available
374 resources that can be reserved on each managed segment under its
375 control. For instance, bandwidth is one such resource. Even
376 though methods such as auto-negotiation of link speeds and
377 knowledge of link topology allow discovery of link capacity, the
378 configuration may be necessary to limit the fraction of link
379 capacity that can be reserved on a link. Configuration is likely
380 to be static with the current L2/L3 devices. Future work may
381 allow for dynamic discovery of this information. This document
382 does not specify the configuration mechanism.
383
384
385
386
387
388 <span class="grey">Yavatkar, et al. Standards Track [Page 7]</span></pre>
389 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-8" ></span>
390 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
391
392
393 2. DSBM Client Initialization: For each interface attached, a DSBM
394 client determines whether a DSBM exists on the interface. The
395 procedure for discovering and verifying the existence of the DSBM
396 for an attached segment is described in <a href="#appendix-A">Appendix A</a>. If the client
397 itself is capable of serving as the DSBM on the segment, it may
398 choose to participate in the election to become the DSBM. At the
399 start, a DSBM client first verifies that a DSBM exists in its L2
400 domain so that it can communicate with the DSBM for admission
401 control purposes.
402
403 In the case of a full-duplex segment, an election may not be
404 necessary as the SBM at each end will typically act as the DSBM
405 for outgoing traffic in each direction.
406
407 3. DSBM-based Admission Control: To request reservation of resources
408 (e.g., LAN bandwidth in a L2 domain), DSBM clients (RSVP-capable
409 L3 devices such as hosts and routers) follow the following steps:
410
411 a) When a DSBM client sends or forwards a RSVP PATH message over
412 an interface attached to a managed segment, it sends the PATH
413 message to the segment's DSBM instead of sending it to the RSVP
414 session destination address (as is done in conventional RSVP
415 processing). After processing (and possibly updating an
416 ADSPEC), the DSBM will forward the PATH message toward its
417 destination address. As part of its processing, the DSBM builds
418 and maintains a PATH state for the session and notes the
419 previous L2/L3 hop that sent it the PATH message.
420
421 Let us consider the managed segment in Figure 1. Assume that a
422 sender to a RSVP session (session address specifies the IP
423 address of host A on the managed segment in Figure 1) resides
424 outside the L2 domain of the managed segment and sends a PATH
425 message that arrives at router R1 which is on the path towards
426 host A.
427
428 DSBM client on Router R1 forwards the PATH message from the
429 sender to the DSBM. The DSBM processes the PATH message and
430 forwards the PATH message towards the RSVP receiver (Detailed
431 message processing and forwarding rules are described in
432 <a href="#section-5">Section 5</a>). In the process, the DSBM builds the PATH state,
433 remembers the router R1 (its L2 and l3 addresses) as the
434 previous hop for the session, puts its own L2 and L3 addresses
435 in the PHOP objects (see explanation later), and effectively
436 inserts itself as an intermediate node between the sender (or
437 R1 in Figure 1) and the receiver (host A) on the managed
438 segment.
439
440
441
442
443
444 <span class="grey">Yavatkar, et al. Standards Track [Page 8]</span></pre>
445 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-9" ></span>
446 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
447
448
449 b) When an application on host A wishes to make a reservation for
450 the RSVP session, host A follows the standard RSVP message
451 processing rules and sends a RSVP RESV message to the previous
452 hop L2/L3 address (the DSBMs address) obtained from the PHOP
453 object(s) in the previously received PATH message.
454
455 c) The DSBM processes the RSVP RESV message based on the bandwidth
456 available and returns an RESV_ERR message to the requester
457 (host A) if the request cannot be granted. If sufficient
458 resources are available and the reservation request is granted,
459 the DSBM forwards the RESV message towards the PHOP(s) based on
460 its local PATH state for the session. The DSBM merges
461 reservation requests for the same session as and when possible
462 using the rules similar to those used in the conventional RSVP
463 processing (except for an additional criterion described in
464 <a href="#section-5.8">Section 5.8</a>).
465
466 d) If the L2 domain contains more than one managed segment, the
467 requester (host A) and the forwarder (router R1) may be
468 separated by more than one managed segment. In that case, the
469 original PATH message would propagate through many DSBMs (one
470 for each managed segment on the path from R1 to A) setting up
471 PATH state at each DSBM. Therefore, the RESV message would
472 propagate hop-by-hop in reverse through the intermediate DSBMs
473 and eventually reach the original forwarder (router R1) on the
474 L2 domain if admission control at all DSBMs succeeds.
475
476 <span class="h4"><a class="selflink" id="section-4.2.2" href="#section-4.2.2">4.2.2</a>. Enhancements to the conventional RSVP operation</span>
477
478 (D)SBMs and DSBM clients implement minor additions to the standard
479 RSVP protocol. These are summarized in this section. A detailed
480 description of the message processing and forwarding rules follows in
481 <a href="#section-5">section 5</a>.
482
483 <span class="h5"><a class="selflink" id="section-4.2.2.1" href="#section-4.2.2.1">4.2.2.1</a> Sending PATH Messages to the DSBM on a Managed Segment</span>
484
485 Normal RSVP forwarding rules apply at a DSBM client when it is not
486 forwarding an outgoing PATH message over a managed segment. However,
487 outgoing PATH messages on a managed segment are sent to the DSBM for
488 the corresponding managed segment (<a href="#section-5.2">Section 5.2</a> describes how the PATH
489 messages are sent to the DSBM on a managed segment).
490
491 <span class="h5"><a class="selflink" id="section-4.2.2.2" href="#section-4.2.2.2">4.2.2.2</a> The LAN_NHOP Objects</span>
492
493 In conventional RSVP processing over point-to-point links, RSVP nodes
494 (hosts/routers) use RSVP_HOP object (NHOP and PHOP info) to keep
495 track of the next hop (downstream node in the path of data packets in
496 a traffic flow) and the previous hop (upstream nodes with respect to
497
498
499
500 <span class="grey">Yavatkar, et al. Standards Track [Page 9]</span></pre>
501 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-10" ></span>
502 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
503
504
505 the data flow) nodes on the path between a sender and a receiver.
506 Routers along the path of a PATH message forward the message towards
507 the destination address based on the L3 routing (packet forwarding)
508 tables.
509
510 For example, consider the L2 domain in Figure 1. Assume that both the
511 sender (some host X) and the receiver (some host Y) in a RSVP session
512 reside outside the L2 domain shown in the Figure, but PATH messages
513 from the sender to its receiver pass through the routers in the L2
514 domain using it as a transit subnet. Assume that the PATH message
515 from the sender X arrives at the router R1. R1 uses its local routing
516 information to decide which next hop router (either router R2 or
517 router R3) to use to forward the PATH message towards host Y.
518 However, when the path traverses a managed L2 domain, we require the
519 PATH and RESV messages to go through a DSBM for each managed segment.
520 Such a L2 domain may span many managed segments (and DSBMs) and,
521 typically, SBM protocol entities on L2 devices (such as a switch)
522 will serve as the DSBMs for the managed segments in a switched
523 topology. When R1 forwards the PATH message to the DSBM (an L2
524 device), the DSBM may not have the L3 routing information necessary
525 to select the egress router (between R2 and R3) before forwarding the
526 PATH message. To ensure correct operation and routing of RSVP
527 messages, we must provide additional forwarding information to DSBMs.
528
529 For this purpose, we introduce new RSVP objects called LAN_NHOP
530 address objects that keep track of the next L3 hop as the PATH
531 message traverses an L2 domain between two L3 entities (RSVP PHOP and
532 NHOP nodes).
533
534 <span class="h5"><a class="selflink" id="section-4.2.2.3" href="#section-4.2.2.3">4.2.2.3</a> Including Both Layer-2 and Layer-3 Addresses in the LAN_NHOP</span>
535
536 When a DSBM client (a host or a router acting as the originator of a
537 PATH message) sends out a PATH message to the DSBM, it must include
538 LAN_NHOP information in the message. In the case of a unicast
539 destination, the LAN_NHOP address specifies the destination address
540 (if the destination is local to its L2 domain) or the address of the
541 next hop router towards the destination. In our example of an RSVP
542 session involving the sender X and receiver Y with L2 domain in
543 Figure 1 acting as the transit subnet, R1 is the ingress node that
544 receives the PATH message. R1 first determines that R2 is the next
545 hop router (or the egress node in the L2 domain for the session
546 address) and then inserts a LAN_NHOP object that specifies R2's IP
547 address. When a DSBM receives a PATH message, it can now look at the
548 address in the LAN_NHOP object and forward the PATH message towards
549 the egress node after processing the PATH message. However, we
550 expect the L2 devices (such as switches) to act as DSBMs on the path
551 within the L2 domain and it may not be reasonable to expect these
552 devices to have an ARP capability to determine the MAC address (we
553
554
555
556 <span class="grey">Yavatkar, et al. Standards Track [Page 10]</span></pre>
557 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-11" ></span>
558 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
559
560
561 call it L2ADDR for Layer 2 address) corresponding to the IP address
562 in the LAN_NHOP object.
563
564 Therefore, we require that the LAN_NHOP information (generated by the
565 L3 device) include both the IP address (LAN_NHOP_L3 address) and the
566 corresponding MAC address (LAN_NHOP_L2 address ) for the next L3 hop
567 over the L2 domain. The LAN_NHOP_L3 address is used by SBM protocol
568 entities on L3 devices to forward the PATH message towards its
569 destination whereas the L2 address is used by the SBM protocol
570 entities on L2 devices to determine how to forward the PATH message
571 towards the L3 NHOP (egress point from the L2 domain). The exact
572 format of the LAN_NHOP information and relevant objects is described
573 later in <a href="#appendix-B">Appendix B</a>.
574
575 <span class="h5"><a class="selflink" id="section-4.2.2.4" href="#section-4.2.2.4">4.2.2.4</a> Similarities to Standard RSVP Message Processing</span>
576
577 - When a DSBM receives a RSVP PATH message, it processes the PATH
578 message according to the PATH processing rules described in the
579 RSVP specification. In particular, the DSBM retrieves the IP
580 address of the previous hop from the RSVP_HOP object in the PATH
581 message and stores the PHOP address in its PATH state. It then
582 forwards the PATH message with the PHOP (RSVP_HOP) object modified
583 to reflect its own IP address (RSVP_HOP_L3 address). Thus, the
584 DSBM inserts itself as an intermediate hop in the chain of nodes
585 in the path between two L3 nodes across the L2 domain.
586
587 - The PATH state in a DSBM is used for forwarding subsequent RESV
588 messages as per the standard RSVP message processing rules. When
589 the DSBM receives a RESV message, it processes the message and
590 forwards it to appropriate PHOP(s) based on its PATH state.
591
592 - Because a DSBM inserts itself as a hop between two RSVP nodes in
593 the path of a RSVP flow, all RSVP related messages (such as PATH,
594 PATH_TEAR, RESV, RESV_CONF, RESV_TEAR, and RESV_ERR) now flow
595 through the DSBM. In particular, a PATH_TEAR message is routed
596 exactly through the intermediate DSBM(s) as its corresponding PATH
597 message and the local PATH state is first cleaned up at each
598 intermediate hop before the PATH_TEAR message gets forwarded.
599
600 - So far, we have described how the PATH message propagates through
601 the L2 domain establishing PATH state at each DSBM along the
602 managed segments in the path. The layer 2 address (LAN_NHOP_L2
603 address) in the LAN_NHOP object should be used by the L2 devices
604 along the path to decide how to forward the PATH message toward
605 the next L3 hop. Such devices will apply the standard IEEE 802.1D
606 forwarding rules (e.g., send it on a single port based on its
607 filtering database, or flood it on all ports active in the
608 spanning tree if the L2 address does not appear in the filtering
609
610
611
612 <span class="grey">Yavatkar, et al. Standards Track [Page 11]</span></pre>
613 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-12" ></span>
614 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
615
616
617 database) to the LAN_NHOP_L2 address as are applied normally to
618 data packets destined to the address.
619
620 <span class="h5"><a class="selflink" id="section-4.2.2.5" href="#section-4.2.2.5">4.2.2.5</a> Including Both Layer-2 and Layer-3 Addresses in the RSVP_HOP</span>
621 <span class="h5"> Objects</span>
622
623 In the conventional RSVP message processing, the PATH state
624 established along the nodes on a path is used to route the RESV
625 message from a receiver to a sender in an RSVP session. As each
626 intermediate node builds the path state, it remembers the previous
627 hop (stores the PHOP IP address available in the RSVP_HOP object of
628 an incoming message) that sent it the PATH message and, when the RESV
629 message arrives, the intermediate node simply uses the stored PHOP
630 address to forward the RESV after processing it successfully.
631
632 In our case, we expect the SBM entities residing at L2 devices to act
633 as DSBMs (and, therefore, intermediate RSVP hops in an L2 domain)
634 along the path between a sender (PHOP) and receiver (NHOP). Thus,
635 when a RESV message arrives at a DSBM, it must use the stored PHOP IP
636 address to forward the RESV message to its previous hop. However, it
637 may not be reasonable to expect the L2 devices to have an ARP cache
638 or the ARP capability to map the PHOP IP address to its corresponding
639 L2 address before forwarding the RESV message.
640
641 To obviate the need for such address mapping at L2 devices, we use a
642 RSVP_HOP_L2 object in the PATH message. The RSVP_HOP_L2 object
643 includes the Layer 2 address (L2ADDR) of the previous hop and
644 complements the L3 address information included in the RSVP_HOP
645 object (RSVP_HOP_L3 address).
646
647 When a L3 device constructs and forwards a PATH message over a
648 managed segment, it includes its IP address (IP address of the
649 interface over which PATH is sent) in the RSVP_HOP object and adds a
650 RSVP_HOP_L2 object that includes the corresponding L2 address for the
651 interface. When a device in the L2 domain receives such a PATH
652 message, it remembers the addresses in the RSVP_HOP and RSVP_HOP_L2
653 objects in its PATH state and then overwrites the RSVP_HOP and
654 RSVP_HOP_L2 objects with its own addresses before forwarding the PATH
655 message over a managed segment.
656
657 The exact format of RSVP_HOP_L2 object is specified in <a href="#appendix-B">Appendix B</a>.
658
659 <span class="h5"><a class="selflink" id="section-4.2.2.6" href="#section-4.2.2.6">4.2.2.6</a> Loop Detection</span>
660
661 When an RSVP session address is a multicast address and a SBM, DSBM,
662 and DSBM clients share the same L2 segment (a shared segment), it is
663 possible for a SBM or a DSBM client to receive one or more copies of
664 a PATH message that it forwarded earlier when a DSBM on the same wire
665
666
667
668 <span class="grey">Yavatkar, et al. Standards Track [Page 12]</span></pre>
669 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-13" ></span>
670 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
671
672
673 forwards it (See <a href="#section-5.7">Section 5.7</a> for an example of such a case). To
674 facilitate detection of such loops, we use a new RSVP object called
675 the LAN_LOOPBACK object. DSBM clients or SBMs (but not the DSBMs
676 reflecting a PATH message onto the interface over which it arrived
677 earlier) must overwrite (or add if the PATH message does NOT already
678 include a LAN_LOOPBACK object) the LAN_LOOPBACK object in the PATH
679 message with their own unicast IP address.
680
681 Now, a SBM or a DSBM client can easily detect and discard the
682 duplicates by checking the contents of the LAN_LOOPBACK object (a
683 duplicate PATH message will list a device's own interface address in
684 the LAN_LOOPBACK object). <a href="#appendix-B">Appendix B</a> specifies the exact format of
685 the LAN_LOOPBACK object.
686
687 <span class="h5"><a class="selflink" id="section-4.2.2.7" href="#section-4.2.2.7">4.2.2.7</a> 802.1p, User Priority and TCLASS</span>
688
689 The model proposed by the Integrated Services working group requires
690 isolation of traffic flows from each other during their transit
691 across a network. The motivation for traffic flow separation is to
692 provide Integrated Services flows protection from misbehaving flows
693 and other best-effort traffic that share the same path. The basic
694 IEEE 802.3/Ethernet networks do not provide any notion of traffic
695 classes to discriminate among different flows that request different
696 services. However, IEEE 802.1p defines a way for switches to
697 differentiate among several "user_priority" values encoded in packets
698 representing different traffic classes (see [<a href="#ref-IEEE802Q" title="&quot;IEEE Standards for Local and Metropolitan Area Networks: Virtual Bridged Local Area Networks&quot;">IEEE802Q</a>, IEEE8021p] for
699 further details). The user_priority values can be encoded either in
700 native LAN packets (e.g., in IEEE 802.5's FC octet) or by using an
701 encapsulation above the MAC layer (e.g., in the case of Ethernet, the
702 user_priority value assigned to each packet will be carried in the
703 frame header using the new, extended frame format defined by IEEE
704 802.1Q [IEEE8021Q]. IEEE, however, makes no recommendations about how
705 a sender or network should use the user_priority values. An
706 accompanying document makes recommendations on the usage of the
707 user_priority values (see [<a href="#ref-RFC-MAP" title="&quot;Integrated Service Mappings on IEEE 802 Networks&quot;">RFC-MAP</a>] for details).
708
709 Under the Integrated Services model, L3 (or higher) entities that
710 transmit traffic flows onto a L2 segment should perform per-flow
711 policing to ensure that the flows do not exceed their traffic
712 specification as specified during admission control. In addition, L3
713 devices may label the frames in such flows with a user_priority value
714 to identify their service class.
715
716 For the purpose of this discussion, we will refer to the
717 user_priority value carried in the extended frame header as the
718 "traffic class" of a packet. Under the ISSLL model, the L3 entities,
719 that send traffic and that use the SBM protocol, may select the
720 appropriate traffic class of outgoing packets [<a href="#ref-RFC-MAP" title="&quot;Integrated Service Mappings on IEEE 802 Networks&quot;">RFC-MAP</a>]. This
721
722
723
724 <span class="grey">Yavatkar, et al. Standards Track [Page 13]</span></pre>
725 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-14" ></span>
726 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
727
728
729 selection may be overridden by DSBM devices, in the following manner.
730 once a sender sends a PATH message, downstream DSBMs will insert a
731 new traffic class object (TCLASS object) in the PATH message that
732 travels to the next L3 device (L3 NHOP for the PATH message). To some
733 extent, the TCLASS object contents are treated like the ADSPEC object
734 in the RSVP PATH messages. The L3 device that receives the PATH
735 message must remove and store the TCLASS object as part of its PATH
736 state for the session. Later, when the same L3 device needs to
737 forward a RSVP RESV message towards the original sender, it must
738 include the TCLASS object in the RESV message. When the RESV message
739 arrives at the original sender, the sender must use the user_priority
740 value from the TCLASS object to override its selection for the
741 traffic class marked in outgoing packets.
742
743 The format of the TCLASS object is specified in <a href="#appendix-B">Appendix B</a>. Note
744 that TCLASS and other SBM-specific objects are carried in a RSVP
745 message in addition to all the other, normal RSVP objects per <a href="./rfc2205">RFC</a>
746 <a href="./rfc2205">2205</a>.
747
748 <span class="h5"><a class="selflink" id="section-4.2.2.8" href="#section-4.2.2.8">4.2.2.8</a> Processing the TCLASS Object</span>
749
750 In summary, use of TCLASS objects requires following additions to the
751 conventional RSVP message processing at DSBMs, SBMs, and DSBM
752 clients:
753
754 * When a DSBM receives a PATH message over a managed segment and the
755 PATH message does not include a TCLASS object, the DSBM MAY add a
756 TCLASS object to the PATH message before forwarding it. The DSBM
757 determines the appropriate user_priority value for the TCLASS
758 object. A mechanism for selecting the appropriate user_priority
759 value is described in an accompanying document [<a href="#ref-RFC-MAP" title="&quot;Integrated Service Mappings on IEEE 802 Networks&quot;">RFC-MAP</a>].
760
761 * When SBM or DSBM receives a PATH message with a TCLASS object over
762 a managed segment in a L2 domain and needs to forward it over a
763 managed segment in the same L2 domain, it will store it in its
764 path state and typically forward the message without changing the
765 contents of the TCLASS object. However, if the DSBM/SBM cannot
766 support the service class represented by the user_priority value
767 specified by the TCLASS object in the PATH message, it may change
768 the priority value in the TCLASS to a semantically "lower" service
769 value to reflect its capability and store the changed TCLASS value
770 in its path state.
771
772
773
774
775
776
777
778
779
780 <span class="grey">Yavatkar, et al. Standards Track [Page 14]</span></pre>
781 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-15" ></span>
782 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
783
784
785 [NOTE: An accompanying document defines the int-serv mappings over
786 IEEE 802 networks [<a href="#ref-RFC-MAP" title="&quot;Integrated Service Mappings on IEEE 802 Networks&quot;">RFC-MAP</a>] provides a precise definition of
787 user_priority values and describes how the user_priority values
788 are compared to determine "lower" of the two values or the
789 "lowest" among all the user_priority values.]
790
791 * When a DSBM receives a RESV message with a TCLASS object, it may
792 use the traffic class information (in addition to the usual
793 flowspec information in the RSVP message) for its own admission
794 control for the managed segment.
795
796 Note that this document does not specify the actual algorithm or
797 policy used for admission control. At one extreme, a DSBM may use
798 per-flow reservation request as specified by the flowspec for a
799 fine grain admission control. At the other extreme, a DSBM may
800 only consider the traffic class information for a very coarse-
801 grain admission control based on some static allocation of link
802 capacity for each traffic class. Any combination of the options
803 represented by these two extremes may also be used.
804
805 * When a DSBM (at an L2 or L3) device receives a RESV message
806 without a TCLASS object and it needs to forward the RESV message
807 over a managed segment within the same L2 domain, it should first
808 check its path state and check whether it has stored a TCLASS
809 value. If so, it should include the TCLASS object in the outgoing
810 RESV message after performing its own admission control. If no
811 TCLASS value is stored, it must forward the RESV message without
812 inserting a TCLASS object.
813
814 * When a DSBM client (residing at an L3 device such as a host or an
815 edge router) receives the TCLASS object in a PATH message that it
816 accepts over an interface, it should store the TCLASS object as
817 part of its PATH state for the interface. Later, when the client
818 forwards a RESV message for the same session on the interface, the
819 client must include the TCLASS object (unchanged from what was
820 received in the previous PATH message) in the RESV message it
821 forwards over the interface.
822
823 * When a DSBM client receives a TCLASS object in an incoming RESV
824 message over a managed segment and local admission control
825 succeeds for the session for the outgoing interface over the
826 managed segment, the client must pass the user_priority value in
827 the TCLASS object to its local packet classifier. This will ensure
828 that the data packets in the admitted RSVP flow that are
829 subsequently forwarded over the outgoing interface will contain
830 the appropriate value encoded in their frame header.
831
832
833
834
835
836 <span class="grey">Yavatkar, et al. Standards Track [Page 15]</span></pre>
837 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-16" ></span>
838 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
839
840
841 * When an L3 device receives a PATH or RESV message over a managed
842 segment in one L2 domain and it needs to forward the PATH/RESV
843 message over an interface outside that domain, the L3 device must
844 remove the TCLASS object (along with LAN_NHOP, RSVP_HOP_L2, and
845 LAN_LOOPBACK objects in the case of the PATH message) before
846 forwarding the PATH/RESV message. If the outgoing interface is on
847 a separate L2 domain, these objects may be regenerated according
848 to the processing rules applicable to that interface.
849
850 <span class="h2"><a class="selflink" id="section-5" href="#section-5">5</a>. Detailed Message Processing Rules</span>
851
852 <span class="h3"><a class="selflink" id="section-5.1" href="#section-5.1">5.1</a>. Additional Notes on Terminology</span>
853
854 * An L2 device may have several interfaces with attached segments
855 that are part of the same L2 domain. A switch in a L2 domain is an
856 example of such a device. A device which has several interfaces
857 may contain a SBM protocol entity that acts in different
858 capacities on each interface. For example, a SBM protocol entity
859 could act as a SBM on interface A, and act as a DSBM on interface
860 B.
861
862 * A SBM protocol entity on a layer 3 device can be a DSBM client,
863 and SBM, a DSBM, or none of the above (SBM transparent). Non-
864 transparent L3 devices can implement any combination of these
865 roles simultaneously. DSBM clients always reside at L3 devices.
866
867 * A SBM protocol entity residing at a layer 2 device can be a SBM, a
868 DSBM or none of the above (SBM transparent). A layer 2 device will
869 never host a DSBM client.
870
871 <span class="h3"><a class="selflink" id="section-5.2" href="#section-5.2">5.2</a>. Use Of Reserved IP Multicast Addresses</span>
872
873 As stated earlier, we require that the DSBM clients forward the RSVP
874 PATH messages to their DSBMs in a L2 domain before they reach the
875 next L3 hop in the path. RSVP PATH messages are addressed, according
876 to <a href="./rfc2205">RFC-2205</a>, to their destination address (which can be either an IP
877 unicast or multicast address). When a L2 device hosts a DSBM, a
878 simple-to-implement mechanism must be provided for the device to
879 capture an incoming PATH message and hand it over to the local DSBM
880 agent without requiring the L2 device to snoop for L3 RSVP messages.
881
882 In addition, DSBM clients need to know how to address SBM messages to
883 the DSBM. For the ease of operation and to allow dynamic DSBM-client
884 binding, it should be possible to easily detect and address the
885 existing DSBM on a managed segment.
886
887
888
889
890
891
892 <span class="grey">Yavatkar, et al. Standards Track [Page 16]</span></pre>
893 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-17" ></span>
894 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
895
896
897 To facilitate dynamic DSBM-client binding as well as to enable easy
898 detection and capture of PATH messages at L2 devices, we require that
899 a DSBM be addressed using a logical address rather than a physical
900 address. We make use of reserved IP multicast address(es) for the
901 purpose of communication with a DSBM. In particular, we require that
902 when a DSBM client or a SBM forwards a PATH message over a managed
903 segment, it is addressed to a reserved IP multicast address. Thus, a
904 DSBM on a L2 device needs to be configured in a way to make it easy
905 to intercept the PATH message and forward it to the local SBM
906 protocol entity. For example, this may involve simply adding a static
907 entry in the device's filtering database (FDB) for the corresponding
908 MAC multicast address to ensure the PATH messages get intercepted and
909 are not forwarded further without the DSBM intervention.
910
911 Similarly, a DSBM always sends the PATH messages over a managed
912 segment using a reserved IP multicast address and, thus, the SBMs or
913 DSBM clients on the managed segments must simply be configured to
914 intercept messages addressed to the reserved multicast address on the
915 appropriate interfaces to easily receive PATH messages.
916
917 RSVP RESV messages continue to be unicast to the previous hop address
918 stored as part of the PATH state at each intermediate hop.
919
920 We define use of two reserved IP multicast addresses. We call these
921 the "AllSBM Address" and the "DSBMLogicalAddress". These are chosen
922 from the range of local multicast addresses, such that:
923
924 * They are not passed through layer 3 devices.
925
926 * They are passed transparently through layer 2 devices which are
927 SBM transparent.
928
929 * They are configured in the permanent database of layer 2 devices
930 which host SBMs or DSBMs, such that they are directed to the SBM
931 management entity in these devices. This obviates the need for
932 these devices to explicitly snoop for SBM related control packets.
933
934 * The two reserved addresses are 224.0.0.16 (DSBMLogicalAddress) and
935 224.0.0.17 (AllSBMAddress).
936
937 These addresses are used as described in the following table:
938
939 Type DSBMLogicaladdress AllSBMAddress
940
941 DSBM * Sends PATH messages * Monitors this address to detect
942 Client to this address the presence of a DSBM
943
944
945
946
947
948 <span class="grey">Yavatkar, et al. Standards Track [Page 17]</span></pre>
949 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-18" ></span>
950 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
951
952
953 * Monitors this address to
954 receive PATH messages
955 forwarded by the DSBM
956
957 SBM * Sends PATH messages * Monitors and sends on this
958 to this address address to participate in
959 election of the DSBM
960 * Monitors this address to
961 receive PATH messages
962 forwarded by the DSBM
963
964 DSBM * Monitors this address * Monitors and sends on this
965 for PATH messages to participate in election
966 directed to it of the DSBM
967 * Sends PATH messages to this
968 address
969
970 The L2 or MAC addresses corresponding to IP multicast addresses are
971 computed algorithmically using a reserved L2 address block (the high
972 order 24-bits are 00:00:5e). The Assigned Numbers RFC [<a href="./rfc1700">RFC-1700</a>]
973 gives additional details.
974
975 <span class="h3"><a class="selflink" id="section-5.3" href="#section-5.3">5.3</a>. Layer 3 to Layer 2 Address Mapping</span>
976
977 As stated earlier, DSBMs or DSBM clients residing at a L3 device must
978 include a LAN_NHOP_L2 address in the LAN_NHOP information so that L2
979 devices along the path of a PATH message do not need to separately
980 determine the mapping between the LAN_NHOP_L3 address in the LAN_NHOP
981 object and its corresponding L2 address (for example, using ARP).
982
983 For the purpose of such mapping at L3 devices, we assume a mapping
984 function called "map_address" that performs the necessary mapping:
985
986 L2ADDR object = map_addr(L3Addr)
987
988 We do not specify how the function is implemented; the implementation
989 may simply involve access to the local ARP cache entry or may require
990 performing an ARP function. The function returns a L2ADDR object
991 that need not be interpreted by an L3 device and can be treated as an
992 opaque object. The format of the L2ADDR object is specified in
993 <a href="#appendix-B">Appendix B</a>.
994
995 <span class="h3"><a class="selflink" id="section-5.4" href="#section-5.4">5.4</a>. Raw vs. UDP Encapsulation</span>
996
997 We assume that the DSBMs, DSBM clients, and SBMs use only raw IP for
998 encapsulating RSVP messages that are forwarded onto a L2 domain.
999 Thus, when a SBM protocol entity on a L3 device forwards a RSVP
1000 message onto a L2 segment, it will only use RAW IP encapsulation.
1001
1002
1003
1004 <span class="grey">Yavatkar, et al. Standards Track [Page 18]</span></pre>
1005 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-19" ></span>
1006 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
1007
1008
1009 <span class="h3"><a class="selflink" id="section-5.5" href="#section-5.5">5.5</a>. The Forwarding Rules</span>
1010
1011 The message processing and forwarding rules will be described in the
1012 context of the sample network illustrated in Figure 2.
1013
1014 Figure 2 - A sample network or L2 domain consisting of switched and
1015 shared L2 segments
1016
1017 ..........
1018 .
1019 +------+ . +------+ seg A +------+ seg C +------+ seg D +------+
1020 | H1 |_______| R1 |_________| S1 |_________| S2 |_______| H2 |
1021 | | . | | | | | | | |
1022 +------+ . +------+ +------+ +------+ +------+
1023 . | /
1024 <span class="h5"><a class="selflink" id="section-1.0.0.0" href="#section-1.0.0.0">1.0.0.0</a> . </span> | /
1025 . |___ /
1026 . seg B | / seg E
1027 .......... | /
1028 2.0.0.0 | /
1029 +-----------+
1030 | S3 |
1031 | |
1032 +-----------+
1033 |
1034 |
1035 |
1036 |
1037 seg F | .................
1038 ------------------------------ .
1039 | | | .
1040 +------+ +------+ +------+ . +------+
1041 | H3 | | H4 | | R2 |____________| H5 |
1042 | | | | | | . | |
1043 +------+ +------+ +------+ . +------+
1044 .
1045 . 3.0.0.0
1046 .................
1047
1048 Figure 2 illustrates a sample network topology consisting of three IP
1049 subnets (1.0.0.0, 2.0.0.0, and 3.0.0.0) interconnected using two
1050 routers. The subnet 2.0.0.0 is an example of a L2 domain consisting
1051 of switches, hosts, and routers interconnected using switched
1052 segments and a shared L2 segment. The sample network contains the
1053 following devices:
1054
1055
1056
1057
1058
1059
1060 <span class="grey">Yavatkar, et al. Standards Track [Page 19]</span></pre>
1061 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-20" ></span>
1062 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
1063
1064
1065 Device Type SBM Type
1066
1067 H1, H5 Host (layer 3) SBM Transparent
1068 H2-H4 Host (layer 3) DSBM Client
1069 R1 Router (layer 3) SBM
1070 R2 Router (layer 3) DSBM for segment F
1071 S1 Switch (layer 2) DSBM for segments A, B
1072 S2 Switch (layer 2) DSBM for segments C, D, E
1073 S3 Switch (layer 2) SBM
1074
1075 The following paragraphs describe the rules, which each of these
1076 devices should use to forward PATH messages (rules apply to PATH_TEAR
1077 messages as well). They are described in the context of the general
1078 network illustrated above. While the examples do not address every
1079 scenario, they do address most of the interesting scenarios.
1080 Exceptions can be discussed separately.
1081
1082 The forwarding rules are applied to received PATH messages (routers
1083 and switches) or originating PATH messages (hosts), as follows:
1084
1085 1. Determine the interface(s) on which to forward the PATH message
1086 using standard forwarding rules:
1087
1088 * If there is a LAN_LOOPBACK object in the PATH message, and it
1089 carries the address of this device, silently discard the
1090 message. (See the section below on "Additional notes on
1091 forwarding the PATH message onto a managed segment).
1092
1093 * Layer 3 devices use the RSVP session address and perform a
1094 routing lookup to determine the forwarding interface(s).
1095
1096 * Layer 2 devices use the LAN_NHOP_L2 address in the LAN_NHOP
1097 information and MAC forwarding tables to determine the
1098 forwarding interface(s). (See the section below on "Additional
1099 notes on forwarding the PATH message onto a managed segment")
1100
1101 2. For each forwarding interface:
1102
1103 * If the device is a layer 3 device, determine whether the
1104 interface is on a managed segment managed by a DSBM, based on
1105 the presence or absence of I_AM_DSBM messages. If the interface
1106 is not on a managed segment, strip out RSVP_HOP_L2, LAN_NHOP,
1107 LAN_LOOPBACK, and TCLASS objects (if present), and forward to
1108 the unicast or multicast destination.
1109
1110
1111
1112
1113
1114
1115
1116 <span class="grey">Yavatkar, et al. Standards Track [Page 20]</span></pre>
1117 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-21" ></span>
1118 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
1119
1120
1121 (Note that the RSVP Class Numbers for these new objects are
1122 chosen so that if an RSVP message includes these objects, the
1123 nodes that are RSVP-aware, but do not participate in the SBM
1124 protocol, will ignore and silently discard such objects.)
1125
1126 * If the device is a layer 2 device or it is a layer 3 device
1127 *and* the interface is on a managed segment, proceed to rule
1128 #3.
1129
1130 3. Forward the PATH message onto the managed segment:
1131
1132 * If the device is a layer 3 device, insert LAN_NHOP address
1133 objects, a LAN_LOOPBACK, and a RSVP_HOP_L2 object into the PATH
1134 message. The LAN_NHOP objects carry the LAN_NHOP_L3 and
1135 LAN_NHOP_L2 addresses of the next layer 3 hop. The RSVP_HOP_L2
1136 object carries the device's own L2 address, and the
1137 LAN_LOOPBACK object contains the IP address of the outgoing
1138 interface.
1139
1140 An L3 device should use the map_addr() function described
1141 earlier to obtain an L2 address corresponding to an IP address.
1142
1143 * If the device hosts the DSBM for the segment to which the
1144 forwarding interface is attached, do the following:
1145
1146 - Retrieve the PHOP information from the standard RSVP HOP
1147 object in the PATH message, and store it. This will be used
1148 to route RESV messages back through the L2 network. If the
1149 PATH message arrived over a managed segment, it will also
1150 contain the RSVP_HOP_L2 object; then retrieve and store also
1151 the previous hop's L2 address in the PATH state.
1152
1153 - Copy the IP address of the forwarding interface (layer 2
1154 devices must also have IP addresses) into the standard RSVP
1155 HOP object and the L2 address of the forwarding interface
1156 into the RSVP_HOP_L2 object.
1157
1158 - If the PATH message received does not contain the TCLASS
1159 object, insert a TCLASS object. The user_priority value
1160 inserted in the TCLASS object is based on service mappings
1161 internal to the device that are configured according to the
1162 guidelines listed in [<a href="#ref-RFC-MAP" title="&quot;Integrated Service Mappings on IEEE 802 Networks&quot;">RFC-MAP</a>]. If the message already
1163 contains the TCLASS object, the user_priority value may be
1164 changed based again on the service mappings internal to the
1165 device.
1166
1167
1168
1169
1170
1171
1172 <span class="grey">Yavatkar, et al. Standards Track [Page 21]</span></pre>
1173 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-22" ></span>
1174 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
1175
1176
1177 * If the device is a layer 3 device and hosts a SBM for the
1178 segment to which the forwarding interface is attached, it *is
1179 required* to retrieve and store the PHOP info.
1180
1181 If the device is a layer 2 device and hosts a SBM for the
1182 segment to which the forwarding interface is attached, it is
1183 *not* required to retrieve and store the PHOP info. If it does
1184 not do so, the SBM must leave the standard RSVP HOP object and
1185 the RSVP_HOP_L2 objects in the PATH message intact and it will
1186 not receive RESV messages.
1187
1188 If the SBM on a L2 device chooses to overwrite the RSVP HOP and
1189 RSVP_HOP_L2 objects with the IP and L2 addresses of its
1190 forwarding interface, it will receive RESV messages. In this
1191 case, it must store the PHOP address info received in the
1192 standard RSVP_HOP field and RSVP_HOP_L2 objects of the incident
1193 PATH message.
1194
1195 In both the cases mentioned above (L2 or L3 devices), the SBM
1196 must forward the TCLASS object in the received PATH message
1197 unchanged.
1198
1199 * Copy the IP address of the forwarding interface into the
1200 LAN_LOOPBACK object, unless the SBM protocol entity is a DSBM
1201 reflecting a PATH message back onto the incident interface.
1202 (See the section below on "Additional notes on forwarding a
1203 PATH message onto a managed segment").
1204
1205 * If the SBM protocol entity is the DSBM for the segment to which
1206 the forwarding interface is attached, it must send the PATH
1207 message to the AllSBMAddress.
1208
1209 * If the SBM protocol entity is a SBM or a DSBM Client on the
1210 segment to which the forwarding interface is attached, it must
1211 send the PATH message to the DSBMLogicalAddress.
1212
1213 <span class="h4"><a class="selflink" id="section-5.5.1" href="#section-5.5.1">5.5.1</a>. Additional notes on forwarding a PATH message onto a managed</span>
1214 <span class="h4"> segment</span>
1215
1216 Rule #1 states that normal IEEE 802.1D forwarding rules should be
1217 used to determine the interfaces on which the PATH message should be
1218 forwarded. In the case of data packets, standard forwarding rules at
1219 a L2 device dictate that the packet should not be forwarded on the
1220 interface from which it was received. However, in the case of a DSBM
1221 that receives a PATH message over a managed segment, the following
1222 exception applies:
1223
1224
1225
1226
1227
1228 <span class="grey">Yavatkar, et al. Standards Track [Page 22]</span></pre>
1229 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-23" ></span>
1230 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
1231
1232
1233 E1. If the address in the LAN_NHOP object is a unicast address,
1234 consult the filtering database (FDB) to determine whether the
1235 destination address is listed on the same interface over which
1236 the message was received. If yes, follow the rule below on
1237 "reflecting a PATH message back onto an interface" described
1238 below; otherwise, proceed with the rest of the message
1239 processing as usual.
1240
1241 E2. If there are members of the multicast group address (specified
1242 by the addresses in the LAN_NHOP object), on the segment from
1243 which the message was received, the message should be
1244 forwarded back onto the interface from which it was received
1245 and follow the rule on "reflecting a PATH message back onto an
1246 interface" described below.
1247
1248 *** Reflecting a PATH message back onto an interface ***
1249
1250 Under the circumstances described above, when a DSBM reflects the
1251 PATH message back onto an interface over which it was received, it
1252 must address it using the AllSBMAddress.
1253
1254 Since it is possible for a DSBM to reflect a PATH message back
1255 onto the interface from which it was received, precautions must be
1256 taken to avoid looping these messages indefinitely. The
1257 LAN_LOOPBACK object addresses this issue. All SBM protocol
1258 entities (except DSBMs reflecting a PATH message) overwrite the
1259 LAN_LOOPBACK object in the PATH message with the IP address of the
1260 outgoing interface. DSBMs which are reflecting a PATH message,
1261 leave the LAN_LOOPBACK object unchanged. Thus, SBM protocol
1262 entities will always be able to recognize a reflected multicast
1263 message by the presence of their own address in the LAN_LOOPBACK
1264 object. These messages should be silently discarded.
1265
1266 <span class="h3"><a class="selflink" id="section-5.6" href="#section-5.6">5.6</a>. Applying the Rules -- Unicast Session</span>
1267
1268 Let's see how the rules are applied in the general network
1269 illustrated previously (see Figure 2).
1270
1271 Assume that H1 is sending a PATH for a unicast session for which H5
1272 is the receiver. The following PATH message is composed by H1:
1273
1274 RSVP Contents
1275 RSVP session IP address IP address of H5 (3.0.0.35)
1276 Sender Template IP address of H1 (1.0.0.11)
1277 PHOP IP address of H1 (1.0.0.11)
1278 RSVP_HOP_L2 n/a (H1 is not sending onto a managed
1279 segment)
1280 LAN_NHOP n/a (H1 is not sending onto a managed
1281
1282
1283
1284 <span class="grey">Yavatkar, et al. Standards Track [Page 23]</span></pre>
1285 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-24" ></span>
1286 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
1287
1288
1289 segment)
1290 LAN_LOOPBACK n/a (H1 is not sending onto a managed
1291 segment)
1292
1293 IP Header
1294 Source address IP address of H1 (1.0.0.11)
1295 Destn address IP addr of H5 (3.0.0.35, assuming raw mode
1296 &amp; router alert)
1297
1298 MAC Header
1299 Destn address The L2 addr corresponding to R1 (determined
1300 by map_addr() and routing tables at H1)
1301
1302 Since H1 is not sending onto a managed segment, the PATH message is
1303 composed and forwarded according to standard RSVP processing rules.
1304
1305 Upon receipt of the PATH message, R1 composes and forwards a PATH
1306 message as follows:
1307
1308 RSVP Contents
1309 RSVP session IP address IP address of H5
1310 Sender Template IP address of H1
1311 PHOP IP address of R1 (2.0.0.1)
1312 (seed the return path for RESV messages)
1313 RSVP_HOP_L2 L2 address of R1
1314 LAN_NHOP LAN_NHOP_L3 (2.0.0.2) and
1315 LAN_NHOP_L2 address of R2 (L2ADDR)
1316 (this is the next layer 3 hop)
1317 LAN_LOOPBACK IP address of R1 (2.0.0.1)
1318
1319 IP Header
1320 Source address IP address of H1
1321 Destn address DSBMLogical IP address (224.0.0.16)
1322
1323 MAC Header
1324 Destn address DSBMLogical MAC address
1325
1326 * R1 does a routing lookup on the RSVP session address, to
1327 determine the IP address of the next layer 3 hop, R2.
1328
1329 * It determines that R2 is accessible via seg A and that seg A
1330 is managed by a DSBM, S1.
1331
1332 * Therefore, it concludes that it is sending onto a managed
1333 segment, and composes LAN_NHOP objects to carry the layer 3
1334 and layer 2 next hop addresses. To compose the LAN_NHOP
1335 L2ADDR object, it invokes the L3 to L2 address mapping function
1336
1337
1338
1339
1340 <span class="grey">Yavatkar, et al. Standards Track [Page 24]</span></pre>
1341 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-25" ></span>
1342 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
1343
1344
1345 ("map_address") to find out the MAC address for the next hop
1346 L3 device, and then inserts a LAN_NHOP_L2ADDR object (that
1347 carries the MAC address) in the message.
1348
1349 * Since R1 is not the DSBM for seg A, it sends the PATH message
1350 to the DSBMLogicalAddress.
1351
1352 Upon receipt of the PATH message, S1 composes and forwards a PATH
1353 message as follows:
1354
1355 RSVP Contents
1356 RSVP session IP address IP address of H5
1357 Sender Template IP address of H1
1358 PHOP IP addr of S1 (seed the return path for RESV
1359 messages)
1360 RSVP_HOP_L2 L2 address of S1
1361 LAN_NHOP LAN_NHOP_L3 (IP) and LAN_NHOP_L2
1362 address of R2
1363 (layer 2 devices do not modify the LAN_NHOP)
1364 LAN_LOOPBACK IP addr of S1
1365
1366 IP Header
1367 Source address IP address of H1
1368 Destn address AllSBMIPaddr (224.0.0.17, since S1 is the
1369 DSBM for seg B).
1370
1371 MAC Header
1372 Destn address All SBM MAC address (since S1 is the DSBM
1373 for seg B).
1374
1375 * S1 looks at the LAN_NHOP address information to determine the
1376 L2 address towards which it should forward the PATH message.
1377
1378 * From the bridge forwarding tables, it determines that the L2
1379 address is reachable via seg B.
1380
1381 * S1 inserts the RSVP_HOP_L2 object and overwrites the RSVP HOP
1382 object (PHOP) with its own addresses.
1383
1384 * Since S1 is the DSBM for seg B, it addresses the PATH message
1385 to the AllSBMAddress.
1386
1387 Upon receipt of the PATH message, S3 composes and forwards a PATH
1388 message as follows:
1389
1390
1391
1392
1393
1394
1395
1396 <span class="grey">Yavatkar, et al. Standards Track [Page 25]</span></pre>
1397 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-26" ></span>
1398 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
1399
1400
1401 RSVP Contents
1402 RSVP session IP addr IP address of H5
1403 Sender Template IP address of H1
1404 PHOP IP addr of S3 (seed the return
1405 path for RESV messages)
1406 RSVP_HOP_L2 L2 address of S3
1407 LAN_NHOP LAN_NHOP_L3 (IP) and
1408 LAN_NHOP_L2 (MAC) address of R2
1409 (L2 devices don't modify LAN_NHOP)
1410 LAN_LOOPBACK IP address of S3
1411
1412 IP Header
1413 Source address IP address of H1
1414 Destn address DSBMLogical IP addr (since S3 is
1415 not the DSBM for seg F)
1416
1417 MAC Header
1418 Destn address DSBMLogical MAC address
1419
1420 * S3 looks at the LAN_NHOP address information to determine the
1421 L2 address towards which it should forward the PATH message.
1422
1423 * From the bridge forwarding tables, it determines that the L2
1424 address is reachable via segment F.
1425
1426 * It has discovered that R2 is the DSBM for segment F. It
1427 therefore sends the PATH message to the DSBMLogicalAddress.
1428
1429 * Note that S3 may or may not choose to overwrite the PHOP
1430 objects with its own IP and L2 addresses. If it does so, it
1431 will receive RESV messages. In this case, it must also store
1432 the PHOP info received in the incident PATH message so that
1433 it is able to forward the RESV messages on the correct path.
1434
1435 Upon receipt of the PATH message, R2 composes and forwards a PATH
1436 message as follows:
1437
1438 RSVP Contents
1439 RSVP session IP addr IP address of H5
1440 Sender Template IP address of H1
1441 PHOP IP addr of R2 (seed the return path for RESV
1442 messages)
1443 RSVP_HOP_L2 Removed by R2 (R2 is not sending onto a
1444 managed segment)
1445 LAN_NHOP Removed by R2 (R2 is not sending onto a
1446 managed segment)
1447
1448
1449
1450
1451
1452 <span class="grey">Yavatkar, et al. Standards Track [Page 26]</span></pre>
1453 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-27" ></span>
1454 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
1455
1456
1457 IP Header
1458 Source address IP address of H1
1459 Destn address IP address of H5, the RSVP session address
1460
1461 MAC Header
1462 Destn address L2 addr corresponding to H5, the next
1463 layer 3 hop
1464
1465 * R2 does a routing lookup on the RSVP session address, to
1466 determine the IP address of the next layer 3 hop, H5.
1467
1468 * It determines that H5 is accessible via a segment for which
1469 there is no DSBM (not a managed segment).
1470
1471 * Therefore, it removes the LAN_NHOP and RSVP_HOP_L2 objects
1472 and places the RSVP session address in the destination
1473 address of the IP header. It places the L2 address of the
1474 next layer 3 hop, into the destination address of the MAC
1475 header and forwards the PATH message to H5.
1476
1477 <span class="h3"><a class="selflink" id="section-5.7" href="#section-5.7">5.7</a>. Applying the Rules - Multicast Session</span>
1478
1479 The rules described above also apply to multicast (m/c) sessions.
1480 For the purpose of this discussion, it is assumed that layer 2
1481 devices track multicast group membership on each port individually.
1482 Layer 2 devices which do not do so, will merely generate extra
1483 multicast traffic. This is the case for L2 devices which do not
1484 implement multicast filtering or GARP/GMRP capability.
1485
1486 Assume that H1 is sending a PATH for an m/c session for which H3 and
1487 H5 are the receivers. The rules are applied as they are in the
1488 unicast case described previously, until the PATH message reaches R2,
1489 with the following exception. The RSVP session address and the
1490 LAN_NHOP carry the destination m/c addresses rather than the unicast
1491 addresses carried in the unicast example.
1492
1493 Now let's look at the processing applied by R2 upon receipt of the
1494 PATH message. Recall that R2 is the DSBM for segment F. Therefore, S3
1495 will have forwarded its PATH message to the DSBMLogicalAddress, to be
1496 picked up by R2. The PATH message will not have been seen by H3 (one
1497 of the m/c receivers), since it monitors only the AllSBMAddress, not
1498 the DSBMLogicalAddress for incoming PATH messages. We rely on R2 to
1499 reflect the PATH message back onto seg f, and to forward it to H5. R2
1500 forwards the following PATH message onto seg f:
1501
1502 RSVP Contents
1503 RSVP session addr m/c session address
1504 Sender Template IP address of H1
1505
1506
1507
1508 <span class="grey">Yavatkar, et al. Standards Track [Page 27]</span></pre>
1509 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-28" ></span>
1510 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
1511
1512
1513 PHOP IP addr of R2 (seed the return path for
1514 RESV messages)
1515 RSVP_HOP_L2 L2 addr of R2
1516 LAN_NHOP m/c session address and corresponding L2 address
1517 LAN_LOOPBACK IP addr of S3 (DSBMs reflecting a PATH
1518 message don't modify this object)
1519
1520 IP Header
1521 Source address IP address of H1
1522
1523 Destn address AllSBMIP address (since R2 is the DSBM for seg F)
1524
1525 MAC Header
1526 Destn address AllSBMMAC address (since R2 is the
1527 DSBM for seg F)
1528
1529 Since H3 is monitoring the All SBM Address, it will receive the PATH
1530 message reflected by R2. Note that R2 violated the standard
1531 forwarding rules here by sending an incoming message back onto the
1532 interface from which it was received. It protected against loops by
1533 leaving S3's address in the LAN_LOOPBACK object unchanged.
1534
1535 R2 forwards the following PATH message on to H5:
1536
1537 RSVP Contents
1538 RSVP session addr m/c session address
1539 Sender Template IP address of H1
1540 PHOP IP addr of R2 (seed the return path for RESV
1541 messages)
1542 RSVP_HOP_L2 Removed by R2 (R2 is not sending onto a
1543 managed segment)
1544 LAN_NHOP Removed by R2 (R2 is not sending onto a
1545 managed segment)
1546 LAN_LOOPBACK Removed by R2 (R2 is not sending onto a
1547 managed segment)
1548
1549 IP Header
1550 Source address IP address of H1
1551 Destn address m/c session address
1552
1553 MAC Header
1554 Destn address MAC addr corresponding to the m/c
1555 session address
1556
1557 * R2 determines that there is an m/c receiver accessible via a
1558 segment for which there is no DSBM. Therefore, it removes the
1559 LAN_NHOP and RSVP_HOP_L2 objects and places the RSVP session
1560 address in the destination address of the IP header. It
1561
1562
1563
1564 <span class="grey">Yavatkar, et al. Standards Track [Page 28]</span></pre>
1565 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-29" ></span>
1566 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
1567
1568
1569 places the corresponding L2 address into the destination
1570 address of the MAC header and multicasts the message towards
1571 H5.
1572
1573 <span class="h3"><a class="selflink" id="section-5.8" href="#section-5.8">5.8</a>. Merging Traffic Class objects</span>
1574
1575 When a DSBM client receives TCLASS objects from different senders
1576 (different PATH messages) in the same RSVP session and needs to
1577 combine them for sending back a single RESV message (as in a wild-
1578 card style reservation), the DSBM client must choose an appropriate
1579 value that corresponds to the desired-delay traffic class. An
1580 accompanying document discusses the guidelines for traffic class
1581 selection based on desired service and the TSpec information [RFC-
1582 MAP].
1583
1584 In addition, when a SBM or DSBM needs to merge RESVs from different
1585 next hops at a merge point, it must decide how to handle the TCLASS
1586 values in the incoming RESVs if they do not match. Consider the case
1587 when a reservation is in place for a flow at a DSBM (or SBM) with a
1588 successful admission control done for the TCLASS requested in the
1589 first RESV for the flow. If another RESV (not the refresh of the
1590 previously admitted RESV) for the same flow arrives at the DSBM, the
1591 DSBM must first check the TCLASS value in the new RESV against the
1592 TCLASS value in the already installed RESV. If the two values are
1593 same, the RESV requests are merged and the new, merged RESV installed
1594 and forwarded using the normal rules of message processing. However,
1595 if the two values are not identical, the DSBM must generate and send
1596 a RESV_ERR message towards the sender (NHOP) of the newer, RESV
1597 message. The RESV_ERR must specify the error code corresponding to
1598 the RSVP "traffic control error" (RESV_ERR code 21) that indicates
1599 failure to merge two incompatible service requests (sub-code 01 for
1600 the RSVP traffic control error) [<a href="./rfc2205">RFC-2205</a>]. The RESV_ERR message may
1601 include additional objects to assist downstream nodes in recovering
1602 from this condition. The definition and usage of such objects is
1603 beyond the scope of this memo.
1604
1605 <span class="h3"><a class="selflink" id="section-5.9" href="#section-5.9">5.9</a>. Operation of SBM Transparent Devices</span>
1606
1607 SBM transparent devices are unaware of the entire SBM/DSBM protocol.
1608 They do not intercept messages addressed to either of the SBM related
1609 local group addresses (the DSBMLogicalAddrss and the ALLSBMAddress),
1610 but instead, pass them through. As a result, they do not divide the
1611 DSBM election scope, they do not explicitly participate in routing of
1612 PATH or RESV messages, and they do not participate in admission
1613 control. They are entirely transparent with respect to SBM operation.
1614
1615
1616
1617
1618
1619
1620 <span class="grey">Yavatkar, et al. Standards Track [Page 29]</span></pre>
1621 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-30" ></span>
1622 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
1623
1624
1625 According to the definitions provided, physical segments
1626 interconnected by SBM transparent devices are considered a single
1627 managed segment. Therefore, DSBMs must perform admission control on
1628 such managed segments, with limited knowledge of the segment's
1629 topology. In this case, the network administrator should configure
1630 the DSBM for each managed segment, with some reasonable approximation
1631 of the segment's capacity. A conservative policy would configure the
1632 DSBM for the lowest capacity route through the managed segment. A
1633 liberal policy would configure the DSBM for the highest capacity
1634 route through the managed segment. A network administrator will
1635 likely choose some value between the two, based on the level of
1636 guarantee required and some knowledge of likely traffic patterns.
1637
1638 This document does not specify the configuration mechanism or the
1639 choice of a policy.
1640
1641 <span class="h3"><a class="selflink" id="section-5.10" href="#section-5.10">5.10</a>. Operation of SBMs Which are NOT DSBMs</span>
1642
1643 In the example illustrated, S3 hosts a SBM, but the SBM on S3 did not
1644 win the election to act as DSBM on any segment. One might ask what
1645 purpose such a SBM protocol entity serves. Such SBMs actually provide
1646 two useful functions. First, the additional SBMs remain passive in
1647 the background for fault tolerance. They listen to the periodic
1648 announcements from the current DSBM for the managed segment (Appendix
1649 A describes this in more detail) and step in to elect a new DSBM when
1650 the current DSBM fails or ceases to be operational for some reason.
1651 Second, such SBMs also provide the important service of dividing the
1652 election scope and reducing the size and complexity of managed
1653 segments. For example, consider the sample topology in Figure 3
1654 again. the device S3 contains an SBM that is not a DSBM for any f the
1655 segments, B, E, or F, attached to it. However, if the SBM protocol
1656 entity on S3 was not present, segments B and F would not be separate
1657 segments from the point of view of the SBM protocol. Instead, they
1658 would constitute a single managed segment, managed by a single DSBM.
1659 Because the SBM entity on S3 divides the election scope, seg B and
1660 seg F are each managed by separate DSBMs. Each of these segments have
1661 a trivial topology and a well defined capacity. As a result, the
1662 DSBMs for these segments do not need to perform admission control
1663 based on approximations (as would be the case if S3 were SBM
1664 transparent).
1665
1666 Note that, SBM protocol entities which are not DSBMs, are not
1667 required to overwrite the PHOP in incident PATH messages with their
1668 own address. This is because it is not necessary for RESV messages to
1669 be routed through these devices. RESV messages are only required to
1670 be routed through the correct sequence of DSBMs. SBMs may not
1671 process RESV messages that do pass through them, other than to
1672 forward them towards their destination address, using standard
1673
1674
1675
1676 <span class="grey">Yavatkar, et al. Standards Track [Page 30]</span></pre>
1677 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-31" ></span>
1678 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
1679
1680
1681 forwarding rules.
1682
1683 SBM protocol entities which are not DSBMs are required to overwrite
1684 the address in the LAN_LOOPBACK object with their own address, in
1685 order to avoid looping multicast messages. However, no state need be
1686 stored.
1687
1688 <span class="h2"><a class="selflink" id="section-6" href="#section-6">6</a>. Inter-Operability Considerations</span>
1689
1690 There are a few interesting inter-operability issues related to the
1691 deployment of a DSBM-based admission control method in an environment
1692 consisting of network nodes with and without RSVP capability. In the
1693 following, we list some of these scenarios and explain how SBM-aware
1694 clients and nodes can operate in those scenarios:
1695
1696 <span class="h3"><a class="selflink" id="section-6.1" href="#section-6.1">6.1</a>. An L2 domain with no RSVP capability.</span>
1697
1698 It is possible to envisage L2 domains that do not use RSVP signaling
1699 for requesting resource reservations, but, instead, use some other
1700 (e.g., SNMP or static configuration) mechanism to reserve bandwidth
1701 at a particular network device such as a router. In that case, the
1702 question is how does a DSBM-based admission control method work and
1703 interoperate with the non-RSVP mechanism. The SBM-based method does
1704 not attempt to provide an admission control solution for such an
1705 environment. The SBM-based approach is part of an end to end
1706 signaling approach to establish resource reservations and does not
1707 attempt to provide a solution for SNMP-based configuration scenario.
1708
1709 As stated earlier, the SBM-based approach can, however, co-exist with
1710 any other, non-RSVP bandwidth allocation mechanism as long as
1711 resources being reserved are either partitioned statically between
1712 the different mechanisms or are resolved dynamically through a common
1713 bandwidth allocator so that there is no over-commitment of the same
1714 resource.
1715
1716 <span class="h3"><a class="selflink" id="section-6.2" href="#section-6.2">6.2</a>. An L2 domain with SBM-transparent L2 Devices.</span>
1717
1718 This scenario has been addressed earlier in the document. The SBM-
1719 based method is designed to operate in such an environment. When
1720 SBM-transparent L2 devices interconnect SBM-aware devices, the
1721 resulting managed segment is a combination of one or more physical
1722 segments and the DSBM for the managed segment may not be as efficient
1723 in allocating resources as it would if all L2 devices were SBM-aware.
1724
1725
1726
1727
1728
1729
1730
1731
1732 <span class="grey">Yavatkar, et al. Standards Track [Page 31]</span></pre>
1733 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-32" ></span>
1734 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
1735
1736
1737 <span class="h3"><a class="selflink" id="section-6.3" href="#section-6.3">6.3</a>. An L2 domain on which some RSVP-based senders are not DSBM clients.</span>
1738
1739 All senders that are sourcing RSVP-based traffic flows onto a managed
1740 segment MUST be SBM-aware and participate in the SBM protocol. Use
1741 of the standard, non-SBM version of RSVP may result in over-
1742 allocation of resources, as such use bypasses the resource management
1743 function of the DSBM. All other senders (i.e., senders that are not
1744 sending streams subject to RSVP admission control) should be elastic
1745 applications that send traffic of lower priority than the RSVP
1746 traffic, and use TCP-like congestion avoidance mechanisms.
1747
1748 All DSBMs, SBMs, or DSBM clients on a managed segment (a segment with
1749 a currently active DSBM) must not accept PATH messages from senders
1750 that are not SBM-aware. PATH messages from such devices can be easily
1751 detected by SBMs and DSBM clients as they would not be multicast to
1752 the ALLSBMAddress (in case of SBMs and DSBM clients) or the
1753 DSBMLogicalAddress (in case of DSBMs).
1754
1755 <span class="h3"><a class="selflink" id="section-6.4" href="#section-6.4">6.4</a>. A non-SBM router that interconnects two DSBM-managed L2 domains.</span>
1756
1757 Multicast SBM messages (e.g., election and PATH messages) have local
1758 scope and are not intended to pass between the two domains. A
1759 correctly configured non-SBM router will not pass such messages
1760 between the domains. A broken router implementation that does so may
1761 cause incorrect operation of the SBM protocol and consequent over- or
1762 under-allocation of resources.
1763
1764 <span class="h3"><a class="selflink" id="section-6.5" href="#section-6.5">6.5</a>. Interoperability with RSVP clients that use UDP encapsulation and</span>
1765 <span class="h3"> are not capable of receiving/sending RSVP messages using RAW_IP</span>
1766
1767 This document stipulates that DSBMs, DSBM clients, and SBMs use only
1768 raw IP for encapsulating RSVP messages that are forwarded onto a L2
1769 domain. <a href="./rfc2205">RFC-2205</a> (the RSVP Proposed Standard) includes support for
1770 both raw IP and UDP encapsulation. Thus, a RSVP node using only the
1771 UDP encapsulation will not be able to interoperate with the DSBM
1772 unless DSBM accepts and supports UDP encapsulated RSVP messages.
1773
1774 <span class="h2"><a class="selflink" id="section-7" href="#section-7">7</a>. Guidelines for Implementers</span>
1775
1776 In the following, we provide guidelines for implementers on different
1777 aspects of the implementation of the SBM-based admission control
1778 procedure including suggestions for DSBM initialization, etc.
1779
1780 <span class="h3"><a class="selflink" id="section-7.1" href="#section-7.1">7.1</a>. DSBM Initialization</span>
1781
1782 As stated earlier, DSBM initialization includes configuration of
1783 maximum bandwidth that can be reserved on a managed segment under its
1784 control. We suggest the following guideline.
1785
1786
1787
1788 <span class="grey">Yavatkar, et al. Standards Track [Page 32]</span></pre>
1789 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-33" ></span>
1790 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
1791
1792
1793 In the case of a managed segment consisting of L2 devices
1794 interconnected by a single shared segment, DSBM entities on such
1795 devices should assume the bandwidth of the interface as the total
1796 link bandwidth. In the case of a DSBM located in a L2 switch, it
1797 might additionally need to be configured with an estimate of the
1798 device's switching capacity if that is less than the link bandwidth,
1799 and possibly with some estimate of the buffering resources of the
1800 switch (see [<a href="#ref-RFC-FRAME" title="&quot;A Framework for Providing Integrated Services Over Shared and Switched LAN Technologies&quot;">RFC-FRAME</a>] for the architectural model assumed for L2
1801 switches). Given the total link bandwidth, the DSBM may be further
1802 configured to limit the maximum amount of bandwidth for RSVP-enabled
1803 flows to ensure spare capacity for best-effort traffic.
1804
1805 <span class="h3"><a class="selflink" id="section-7.2" href="#section-7.2">7.2</a>. Operation of DSBMs in Different L2 Topologies</span>
1806
1807 Depending on a L2 topology, a DSBM may be called upon to manage
1808 resources for one or more segments and the implementers must bear in
1809 mind efficiency implications of the use of DSBM in different L2
1810 topologies. Trivial L2 topologies consist of a single "physical
1811 segment". In this case, the 'managed segment' is equivalent to a
1812 single segment. Complex L2 topologies may consist of a number of
1813 Admission control on such an L2 extended segment can be performed
1814 from a single pool of resources, similar to a single shared segment,
1815 from the point of view of a single DSBM.
1816
1817 This configuration compromises the efficiency with which the DSBM can
1818 allocate resources. This is because the single DSBM is required to
1819 make admission control decisions for all reservation requests within
1820 the L2 topology, with no knowledge of the actual physical segments
1821 affected by the reservation.
1822
1823 We can realize improvements in the efficiency of resource allocation
1824 by subdividing the complex segment into a number of managed segments,
1825 each managed by their own DSBM. In this case, each DSBM manages a
1826 managed segment having a relatively simple topology. Since managed
1827 segments are simpler, the DSBM can be configured with a more accurate
1828 estimate of the resources available for all reservations in the
1829 managed segment. In the ultimate configuration, each physical segment
1830 is a managed segment and is managed by its own DSBM. We make no
1831 assumption about the number of managed segments but state, simply,
1832 that in complex L2 topologies, the efficiency of resource allocation
1833 improves as the granularity of managed segments increases.
1834
1835 <span class="h2"><a class="selflink" id="section-8" href="#section-8">8</a>. Security Considerations</span>
1836
1837 The message formatting and usage rules described in this note raise
1838 security issues, identical to those raised by the use of RSVP and
1839 Integrated Services. It is necessary to control and authenticate
1840
1841
1842
1843
1844 <span class="grey">Yavatkar, et al. Standards Track [Page 33]</span></pre>
1845 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-34" ></span>
1846 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
1847
1848
1849 access to enhanced qualities of service enabled by the technology
1850 described in this RFC. This requirement is discussed further in
1851 [<a href="./rfc2205">RFC-2205</a>], [<a href="./rfc2211">RFC-2211</a>], and [<a href="./rfc2212">RFC-2212</a>].
1852
1853 [<a id="ref-RFC-RSVPMD5">RFC-RSVPMD5</a>] describes the mechanism used to protect the integrity
1854 of RSVP messages carrying the information described here. A SBM
1855 implementation should satisfy the requirements of that RFC and
1856 provide the suggested mechanisms just as though it were a
1857 conventional RSVP implementation. It should further use the same
1858 mechanisms to protect the additional, SBM-specific objects in a
1859 message.
1860
1861 Finally, it is also necessary to authenticate DSBM candidates during
1862 the election process, and a mechanism based on a shared secret among
1863 the DSBM candidates may be used. The mechanism defined in [RFC-
1864 RSVPMD5] should be used.
1865
1866 <span class="h2"><a class="selflink" id="section-9" href="#section-9">9</a>. References</span>
1867
1868 [<a id="ref-RFC 2205">RFC 2205</a>] Braden, R., Zhang, L., Berson, S., Herzog, S. and S.
1869 Jamin, "Resource ReSerVation Protocol (RSVP) -- Version
1870 1 Functional Specification", <a href="./rfc2205">RFC 2205</a>, September 1997.
1871
1872 [<a id="ref-RFC-RSVPMD5">RFC-RSVPMD5</a>] Baker, F., Lindell, B. and M. Talwar, "RSVP
1873 Cryptographic Authentication", <a href="./rfc2747">RFC 2747</a>, January 2000.
1874
1875 [<a id="ref-RFC 2206">RFC 2206</a>] Baker, F. and J. Krawczyk, "RSVP Management Information
1876 Base", <a href="./rfc2206">RFC 2206</a>, September 1997.
1877
1878 [<a id="ref-RFC 2211">RFC 2211</a>] Wroclawski, J., "Specification of the Controlled-Load
1879 Network Element Service", <a href="./rfc2211">RFC 2211</a>, September 1997.
1880
1881 [<a id="ref-RFC 2212">RFC 2212</a>] Shenker, S., Partridge, C. and R. Guerin,
1882 "Specification of Guaranteed Quality of Service", <a href="./rfc2212">RFC</a>
1883 <a href="./rfc2212">2212</a>, September 1997.
1884
1885 [<a id="ref-RFC 2215">RFC 2215</a>] Shenker, S. and J. Wroclawski, "General
1886 Characterization Parameters for Integrated Service
1887 Network Elements", <a href="./rfc2215">RFC 2215</a>, September 1997.
1888
1889 [<a id="ref-RFC 2210">RFC 2210</a>] Wroclawski, J., "The Use of RSVP with IETF Integrated
1890 Services", <a href="./rfc2210">RFC 2210</a>, September 1997.
1891
1892 [<a id="ref-RFC 2213">RFC 2213</a>] Baker, F. and J. Krawczyk, "Integrated Services
1893 Management Information Base", <a href="./rfc2213">RFC 2213</a>, September 1997.
1894
1895
1896
1897
1898
1899
1900 <span class="grey">Yavatkar, et al. Standards Track [Page 34]</span></pre>
1901 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-35" ></span>
1902 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
1903
1904
1905 [<a id="ref-RFC-FRAME">RFC-FRAME</a>] Ghanwani, A., Pace, W., Srinivasan, V., Smith, A. and
1906 M.Seaman, "A Framework for Providing Integrated
1907 Services Over Shared and Switched LAN Technologies",
1908 <a href="./rfc2816">RFC 2816</a>, May 2000.
1909
1910 [<a id="ref-RFC-MAP">RFC-MAP</a>] Seaman, M., Smith, A. and E. Crawley, "Integrated
1911 Service Mappings on IEEE 802 Networks", <a href="./rfc2815">RFC 2815</a>, May
1912 2000.
1913
1914 [<a id="ref-IEEE802Q">IEEE802Q</a>] "IEEE Standards for Local and Metropolitan Area
1915 Networks: Virtual Bridged Local Area Networks", Draft
1916 Standard P802.1Q/D9, February 20, 1998.
1917
1918 [<a id="ref-IEEEP8021p">IEEEP8021p</a>] "Information technology - Telecommunications and
1919 information exchange between systems - Local and
1920 metropolitan area networks - Common specifications -
1921 Part 3: Media Access Control (MAC) Bridges: Revision
1922 (Incorporating IEEE P802.1p: Traffic Class Expediting
1923 and Dynamic Multicast Filtering)", ISO/IEC Final CD
1924 15802-3 IEEE P802.1D/D15, November 24, 1997.
1925
1926 [<a id="ref-IEEE8021D">IEEE8021D</a>] "MAC Bridges", ISO/IEC 10038, ANSI/IEEE Std 802.1D-
1927 1993.
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956 <span class="grey">Yavatkar, et al. Standards Track [Page 35]</span></pre>
1957 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-36" ></span>
1958 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
1959
1960
1961 <span class="h3"><a class="selflink" id="appendix-A.1" href="#appendix-A.1">A.1</a>. Introduction</span>
1962
1963 To simplify the rest of this discussion, we will assume that there is
1964 a single DSBM for the entire L2 domain (i.e., assume a shared L2
1965 segment for the entire L2 domain). Later, we will discuss how a DSBM
1966 is elected for a half-duplex or full-duplex switched segment.
1967
1968 To allow for quick recovery from the failure of a DSBM, we assume
1969 that additional SBMs may be active in a L2 domain for fault
1970 tolerance. When more than one SBM is active in a L2 domain, the SBMs
1971 use an election algorithm to elect a DSBM for the L2 domain. After
1972 the DSBM is elected and is operational, other SBMs remain passive in
1973 the background to step in to elect a new DSBM when necessary. The
1974 protocol for electing and discovering DSBM is called the "DSBM
1975 election protocol" and is described in the rest of this Appendix.
1976
1977 <span class="h4"><a class="selflink" id="appendix-A.1.1" href="#appendix-A.1.1">A.1.1</a>. How a DSBM Client Detects a Managed Segment</span>
1978
1979 Once elected, a DSBM periodically multicasts an I_AM_DSBM message on
1980 the AllSBMAddress to indicate its presence. The message is sent every
1981 period (e.g., every 5 seconds) according to the RefreshInterval timer
1982 value (a configuration parameter). Absence of such a message over a
1983 certain time interval (called "DSBMDeadInterval"; another
1984 configuration parameter typically set to a multiple of
1985 RefreshInterval) indicates that the DSBM has failed or terminated and
1986 triggers another round of the DSBM election. The DSBM clients always
1987 listen for periodic DSBM advertisements. The advertisement includes
1988 the unicast IP address of the DSBM (DSBMAddress) and DSBM clients
1989 send their PATH/RESV (or other) messages to the DSBM. When a DSBM
1990 client detects the failure of a DSBM, it waits for a subsequent
1991 I_AM_DSBM advertisement before resuming any communication with the
1992 DSBM. During the period when a DSBM is not present, a DSBM client may
1993 forward outgoing PATH messages using the standard RSVP forwarding
1994 rules.
1995
1996 The exact message formats and addresses used for communication with
1997 (and among) SBM(s) are described in <a href="#appendix-B">Appendix B</a>.
1998
1999 <span class="h3"><a class="selflink" id="appendix-A.2" href="#appendix-A.2">A.2</a>. Overview of the DSBM Election Procedure</span>
2000
2001 When a SBM first starts up, it listens for incoming DSBM
2002 advertisements for some period to check whether a DSBM already exists
2003 in its L2 domain. If one already exists (and no new election is in
2004 progress), the new SBM stays quiet in the background until an
2005 election of DSBM is necessary. All messages related to the DSBM
2006 election and DSBM advertisements are always sent to the
2007 AllSBMAddress.
2008
2009
2010
2011
2012 <span class="grey">Yavatkar, et al. Standards Track [Page 36]</span></pre>
2013 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-37" ></span>
2014 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
2015
2016
2017 If no DSBM exists, the SBM initiates the election of a DSBM by
2018 sending out a DSBM_WILLING message that lists its IP address as a
2019 candidate DSBM and its "SBM priority". Each SBM is assigned a
2020 priority to determine its relative precedence. When more than one
2021 SBM candidate exists, the SBM priority determines who gets to be the
2022 DSBM based on the relative priority of candidates. If there is a tie
2023 based on the priority value, the tie is broken using the IP
2024 addresses of tied candidates (one with the higher IP address in the
2025 lexicographic order wins). The details of the election protocol start
2026 in Section A.4.
2027
2028 <span class="h4"><a class="selflink" id="appendix-A.2.1" href="#appendix-A.2.1">A.2.1</a> Summary of the Election Algorithm</span>
2029
2030 For the purpose of the algorithm, a SBM is in one of the four states
2031 (Idle, DetectDSBM, ElectDSBM, IAMDSBM).
2032
2033 A SBM (call it X) starts up in the DetectDSBM state and waits for a
2034 ListenInterval for incoming I_AM_DSBM (DSBM advertisement) or
2035 DSBM_WILLING messages. If an I_AM_DSBM advertisement is received
2036 during this state, the SBM notes the current DSBM (its IP address and
2037 priority) and enters the Idle state. If a DSBM_WILLING message is
2038 received from another SBM (call it Y) during this state, then X
2039 enters the ElectDSBM state. Before entering the new state, X first
2040 checks to see whether it itself is a better candidate than Y and, if
2041 so, sends out a DSBM_WILLING message and then enters the ElectDSBM
2042 state.
2043
2044 When a SBM (call it X) enters the ElectDSBM state, it sets a timer
2045 (called ElectionIntervalTimer, and typically set to a value at least
2046 equal to the DSBMDeadInterval value) to wait for the election to
2047 finish and to discover who is the best candidate. In this state, X
2048 keeps track of the best (or better) candidate seen so far (including
2049 itself). Whenever it receives another DSBM_WILLING message it updates
2050 its notion of the best (or better) candidate based on the priority
2051 (and tie-breaking) criterion. During the ElectionInterval, X sends
2052 out a DSBM_WILLING message every RefreshInterval to (re)assert its
2053 candidacy.
2054
2055 At the end of the ElectionInterval, X checks whether it is the best
2056 candidate so far. If so, it declares itself to be the DSBM (by
2057 sending out the I_AM_DSBM advertisement) and enters the IAMDSBM
2058 state; otherwise, it decides to wait for the best candidate to
2059 declare itself the winner. To wait, X re-initializes its ElectDSBM
2060 state and continues to wait for another round of election (each round
2061 lasts for an ElectionTimerInterval duration).
2062
2063
2064
2065
2066
2067
2068 <span class="grey">Yavatkar, et al. Standards Track [Page 37]</span></pre>
2069 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-38" ></span>
2070 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
2071
2072
2073 A SBM is in Idle state when no election is in progress and the DSBM
2074 is already elected (and happens to be someone else). In this state,
2075 it listens for incoming I_AM_DSBM advertisements and uses a
2076 DSBMDeadIntervalTimer to detect the failure of DSBM. Every time the
2077 advertisement is received, the timer is restarted. If the timer
2078 fires, the SBM goes into the DetectDSBM state to prepare to elect the
2079 new DSBM. If a SBM receives a DSBM_WILLING message from the current
2080 DSBM in this state, the SBM enters the ElectDSBM state after sending
2081 out a DSBM_WILLING message (to announce its own candidacy).
2082
2083 In the IAMDSBM state, the DSBM sends out I_AM_DSBM advertisements
2084 every refresh interval. If the DSBM wishes to shut down (gracefully
2085 terminate), it sends out a DSBM_WILLING message (with SBM priority
2086 value set to zero) to initiate the election procedure. The priority
2087 value zero effectively removes the outgoing DSBM from the election
2088 procedure and makes way for the election of a different DSBM.
2089
2090 <span class="h3"><a class="selflink" id="appendix-A.3" href="#appendix-A.3">A.3</a>. Recovering from DSBM Failure</span>
2091
2092 When a DSBM fails (DSBMDeadIntervalTimer fires), all the SBMs enter
2093 the ElectDSBM state and start the election process.
2094
2095 At the end of the ElectionInterval, the elected DSBM sends out an
2096 I_AM_DSBM advertisement and the DSBM is then operational.
2097
2098 <span class="h3"><a class="selflink" id="appendix-A.4" href="#appendix-A.4">A.4</a>. DSBM Advertisements</span>
2099
2100 The I_AM_DSBM advertisement contains the following information:
2101
2102 1. DSBM address information -- contains the IP and L2 addresses of
2103 the DSBM and its SBM priority (a configuration parameter --
2104 priority specified by a network administrator). The priority
2105 value is used to choose among candidate SBMs during the election
2106 algorithm. Higher integer values indicate higher priority and the
2107 value is in the range 0..255. The value zero indicates that the
2108 SBM is not eligible to be the DSBM. The IP address is required
2109 and used for breaking ties. The L2 address is for the interface
2110 of the managed segment.
2111
2112 2. RegreshInterval -- contains the value of RefreshInterval in
2113 seconds. Value zero indicates the parameter has been omitted in
2114 the message. Receivers may substitute their own default value in
2115 this case.
2116
2117 3. DSBMDeadInterval -- contains the value of DSBMDeadInterval in
2118 seconds. If the value is omitted (or value zero is specified), a
2119 default value (from initial configuration) should be used.
2120
2121
2122
2123
2124 <span class="grey">Yavatkar, et al. Standards Track [Page 38]</span></pre>
2125 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-39" ></span>
2126 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
2127
2128
2129 4. Miscellaneous configuration information to be advertised to
2130 senders on the managed segment. See <a href="#appendix-C">Appendix C</a> for further
2131 details.
2132
2133 <span class="h3"><a class="selflink" id="appendix-A.5" href="#appendix-A.5">A.5</a>. DSBM_WILLING Messages</span>
2134
2135 When a SBM wishes to declare its candidacy to be the DSBM during an
2136 election phase, it sends out a DSBM_WILLING message. The DSBM_WILLING
2137 message contains the following information:
2138
2139 1. DSBM address information -- Contains the SBM's own addresses (IP
2140 and L2 address), if it wishes to be the DSBM. The IP address is
2141 required and used for breaking ties. The L2 address is the
2142 address of the interface for the managed segment in question.
2143 Also, the DSBM address information includes the corresponding
2144 priority of the SBM whose address is given above.
2145
2146 <span class="h3"><a class="selflink" id="appendix-A.6" href="#appendix-A.6">A.6</a>. SBM State Variables</span>
2147
2148 For each network interface, a SBM maintains the following state
2149 variables related to the election of the DSBM for the L2 domain on
2150 that interface:
2151
2152 a) LocalDSBMAddrInfo -- current DSBM's IP address (initially,
2153 0.0.0.0) and priority. All IP addresses are assumed to be in
2154 network byte order. In addition, current DSBM's L2 address is
2155 also stored as part of this state information.
2156
2157 b) OwnAddrInfo -- SBM's own IP address and L2 address for the
2158 interface and its own priority (a configuration parameter).
2159
2160 c) RefreshInterval in seconds. When the DSBM is not yet elected,
2161 it is set to a default value specified as a configuration
2162 parameter.
2163
2164 d) DSBMDeadInterval in seconds. When the DSBM is not yet elected,
2165 it is initially set to a default value specified as a
2166 configuration parameter.
2167
2168 f) ListenInterval in seconds -- a configuration parameter that
2169 decides how long a SBM spends in the DetectDSBM state (see
2170 below).
2171
2172 g) ElectionInterval in seconds -- a configuration parameter that
2173 decides how long a SBM spends in the ElectDSBM state when it has
2174 declared its candidacy.
2175
2176
2177
2178
2179
2180 <span class="grey">Yavatkar, et al. Standards Track [Page 39]</span></pre>
2181 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-40" ></span>
2182 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
2183
2184
2185 Figure 3 shows the state transition diagram for the election protocol
2186 and the various states are described below. A complete description of
2187 the state machine is provided in Section A.10.
2188
2189 <span class="h3"><a class="selflink" id="appendix-A.7" href="#appendix-A.7">A.7</a>. DSBM Election States</span>
2190
2191 DOWN -- SBM is not operational.
2192
2193 DetectDSBM -- typically, the initial state of a SBM when it
2194 starts up. In this state, it checks to see whether a DSBM already
2195 exists in its domain.
2196
2197 Idle -- SBM is in this state when no election is in progress and
2198 it is not the DSBM. In this state, SBM passively monitors the
2199 state of the DSBM.
2200
2201 ElectDSBM -- SBM is in this state when a DSBM election is in
2202 progress.
2203
2204 IAMDSBM -- SBM is in this state when it is the DSBM for the L2
2205 domain.
2206
2207 <span class="h3"><a class="selflink" id="appendix-A.8" href="#appendix-A.8">A.8</a>. Events that cause state changes</span>
2208
2209 StartUp -- SBM starts operation.
2210
2211 ListenInterval Timeout -- The ListenInterval timer has fired.
2212 This means that the SBM has monitored its domain to check for an
2213 existing DSBM or to check whether there are candidates (other
2214 than itself) willing to be the DSBM.
2215
2216 DSBM_WILLING message received -- This means that the SBM received
2217 a DSBM_WILLING message from some other SBM. Such a message is
2218 sent when a SBM wishes to declare its candidacy to be the DSBM.
2219
2220 I_AM_DSBM message received -- SBM received a DSBM advertisement
2221 from the DSBM in its L2 domain.
2222
2223 DSBMDeadInterval Timeout -- The DSBMDeadIntervalTimer has fired.
2224 This means that the SBM did not receive even one DSBM
2225 advertisement during this period and indicates possible failure
2226 of the DSBM.
2227
2228 RefreshInterval Timeout -- The RefreshIntervalTimer has fired. In
2229 the IAMDSBM state, this means it is the time for sending out the
2230 next DSBM advertisement. In the ElectDSBM state, the event means
2231 that it is the time to send out another DSBM_WILLING message.
2232
2233
2234
2235
2236 <span class="grey">Yavatkar, et al. Standards Track [Page 40]</span></pre>
2237 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-41" ></span>
2238 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
2239
2240
2241 ElectionInterval Timeout -- The ElectionIntervalTimer has fired.
2242 This means that the SBM has waited long enough after declaring
2243 its candidacy to determine whether or not it succeeded.
2244
2245 <span class="h3"><a class="selflink" id="appendix-A.9" href="#appendix-A.9">A.9</a>. State Transition Diagram (Figure 3)</span>
2246
2247 +-----------+
2248 +--&lt;--------------&lt;-|DetectDSBM |----&gt;------+
2249 | +-----------+ |
2250 | |
2251 | |
2252 | |
2253 | +-------------+ +---------+ |
2254 +-&gt;---| Idle |--&lt;&gt;---|ElectDSBM|--&lt;--+
2255 +-------------+ +---------+
2256 | |
2257 | |
2258 | |
2259 | +-----------+ |
2260 +&lt;&lt;- +---| IAMDSBM |-&lt;-+
2261 | +-----------+
2262 |
2263 | +-----------+
2264 +&gt;&gt;-| SHUTDOWN |
2265 +-----------+
2266
2267 <span class="h3"><a class="selflink" id="appendix-A.10" href="#appendix-A.10">A.10</a>. Election State Machine</span>
2268
2269 Based on the events and states described above, the state changes at
2270 a SBM are described below. Each state change is triggered by an event
2271 and is typically accompanied by a sequence of actions. The state
2272 machine is described assuming a single threaded implementation (to
2273 avoid race conditions between state changes and timer events) with no
2274 timer events occurring during the execution of the state machine.
2275
2276 The following routines will be frequently used in the description of
2277 the state machine:
2278
2279 ComparePrio(FirstAddrInfo, SecondAddrInfo)
2280 -- determines whether the entity represented by the first parameter
2281 is better than the second entity using the priority information
2282 and the IP address information in the two parameters. If any
2283 address is zero, that entity automatically loses; then first
2284 priorities are compared; higher priority candidate wins. If there
2285 is a tie based on the priority value, the tie is broken using the
2286 IP addresses of tied candidates (one with the higher IP address
2287 in the lexicographic order wins). Returns TRUE if first entity
2288 is a better choice. FALSE otherwise.
2289
2290
2291
2292 <span class="grey">Yavatkar, et al. Standards Track [Page 41]</span></pre>
2293 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-42" ></span>
2294 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
2295
2296
2297 SendDSBMWilling Message()
2298 Begin
2299 Send out DSBM_WILLING message listing myself as a candidate for
2300 DSBM (copy OwnAddr and priority into appropriate fields)
2301 start RefreshIntervalTimer
2302 goto ElectDSBM state
2303 End
2304
2305 AmIBetterDSBM(OtherAddrInfo)
2306 Begin
2307 if (ComparePrio(OwnAddrInfo, OtherAddrInfo))
2308 return TRUE
2309
2310 change LocalDSBMInfo = OtherDSBMAddrInfo
2311 return FALSE
2312 End
2313
2314 UpdateDSBMInfo()
2315 /* invoked in an assignment such as LocalDSBMInfo = OtherAddrInfo */
2316 Begin
2317 update LocalDSBMInfo such as IP addr, DSBM L2 address,
2318 DSBM priority, RefreshIntervalTimer, DSBMDeadIntervalTimer
2319 End
2320
2321 <span class="h4"><a class="selflink" id="appendix-A.10.1" href="#appendix-A.10.1">A.10.1</a> State Changes</span>
2322
2323 In the following, the action "continue" or "continue in current
2324 state" means an "exit" from the current action sequence without a
2325 state transition.
2326
2327 State: DOWN
2328 Event: StartUp
2329 New State: DetectDSBM
2330 Action: Initialize the local state variables (LocalDSBMADDR and
2331 LocalDSBMAddrInfo set to 0). Start the ListenIntervalTimer.
2332
2333 State: DetectDSBM
2334 New State: Idle
2335 Event: I_AM_DSBM message received
2336 Action: set LocalDSBMAddrInfo = IncomingDSBMAddrInfo
2337 start DeadDSBMInterval timer
2338 goto Idle State
2339
2340 State: DetectDSBM
2341 Event: ListenIntervalTimer fired
2342 New State: ElectDSBM
2343 Action: Start ElectionIntervalTimer
2344 SendDSBMWillingMessage();
2345
2346
2347
2348 <span class="grey">Yavatkar, et al. Standards Track [Page 42]</span></pre>
2349 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-43" ></span>
2350 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
2351
2352
2353 State: DetectDSBM
2354 Event: DSBM_WILLING message received
2355 New State: ElectDSBM
2356 Action: Cancel any active timers
2357
2358 Start ElectionIntervalTimer
2359 /* am I a better choice than this dude? */
2360 If (ComparePrio(OwnAddrInfo, IncomingDSBMInfo)) {
2361 /* I am better */
2362 SendDSBMWillingMessage()
2363 } else {
2364 Change LocalDSBMAddrInfo = IncomingDSBMAddrInfo
2365 goto ElectDSBM state
2366 }
2367
2368 State: Idle
2369 Event: DSBMDeadIntervalTimer fired.
2370 New State: ElectDSBM
2371 Action: start ElectionIntervalTimer
2372 set LocalDSBMAddrInfo = OwnAddrInfo
2373 SendDSBMWiliingMessage()
2374
2375 State: Idle
2376 Event: I_AM_DSBM message received.
2377 New State: Idle
2378 Action: /* first check whether anything has changed */
2379 if (!ComparePrio(LocalDSBMAddrInfo, IncomingDSBMAddrInfo))
2380 change LocalDSBMAddrInfo to reflect new info
2381 endif
2382 restart DSBMDeadIntervalTimer;
2383 continue in current state;
2384
2385 State: Idle
2386 Event: DSBM_WILLING Message is received
2387 New State: Depends on action (ElectDSBM or Idle)
2388 Action: /* check whether it is from the DSBM itself (shutdown) */
2389 if (IncomingDSBMAddr == LocalDSBMAddr) {
2390 cancel active timers
2391 Set LocalDSBMAddrInfo = OwnAddrInfo
2392 Start ElectionIntervalTimer
2393 SendDSBMWillingMessage() /* goto ElectDSBM state */
2394 }
2395
2396 /* else, ignore it */
2397 continue in current state
2398
2399 State: ElectDSBM
2400 Event: ElectionIntervalTimer Fired
2401
2402
2403
2404 <span class="grey">Yavatkar, et al. Standards Track [Page 43]</span></pre>
2405 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-44" ></span>
2406 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
2407
2408
2409 New State: depends on action (IAMDSBM or Current State)
2410 Action: If (LocalDSBMAddrInfo == OwnAddrInfo) {
2411 /* I won */
2412 send I_AM_DSBM message
2413 start RefreshIntervalTimer
2414 goto IAMDSBM state
2415 } else { /* someone else won, so wait for it to declare
2416 itself to be the DSBM */
2417 set LocalDSBMAddressInfo = OwnAddrInfo
2418 start ElectionIntervalTimer
2419 SendDSBMWillingMessage()
2420 continue in current state
2421 }
2422
2423 State: ElectDSBM
2424 Event: I_AM_DSBM message received
2425 New State: Idle
2426 Action: set LocalDSBMAddrInfo = IncomingDSBMAddrInfo
2427 Cancel any active timers
2428 start DeadDSBMInterval timer
2429 goto Idle State
2430
2431 State: ElectDSBM
2432 Event: DSBM_WILLING message received
2433 New State: ElectDSBM
2434 Action: Check whether it's a loopback and if so, discard, continue;
2435 if (!AmIBetterDSBM(IncomingDSBMAddrInfo)) {
2436 Change LocalDSBMAddrInfo = IncomingDSBMAddrInfo
2437 Cancel RefreshIntervalTimer
2438 } else if (LocalDSBMAddrInfo == OwnAddrInfo) {
2439 SendDSBMWillingMessage()
2440 }
2441 continue in current state
2442
2443 State: ElectDSBM
2444 Event: RefreshIntervalTimer fired
2445 New State: ElectDSBM
2446 Action: /* continue to send DSBMWilling messages until
2447 election interval ends */
2448 SendDSBMWillingMessage()
2449
2450 State: IAMDSBM
2451 Event: DSBM_WILLING message received
2452 New State: depends on action (IAMDSBM or SteadyState)
2453 Action: /* check whether other guy is better */
2454 If (ComparePrio(OwnAddrInfo, IncomingAddrInfo)) {
2455 /* I am better */
2456 send I_AM_DSBM message
2457
2458
2459
2460 <span class="grey">Yavatkar, et al. Standards Track [Page 44]</span></pre>
2461 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-45" ></span>
2462 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
2463
2464
2465 restart RefreshIntervalTimer
2466 continue in current state
2467 } else {
2468 Set LocalDSBMAddrInfo = IncomingAddrInfo
2469 cancel active timers
2470 start DSBMDeadIntervalTimer
2471 goto SteadyState
2472 }
2473
2474 State: IAMDSBM
2475 Event: RefreshIntervalTimer fired
2476 New State: IAMDSBM
2477 Action: send I_AM_DSBM message
2478 restart RefreshIntervalTimer
2479
2480 State: IAMDSBM
2481 Event: I_AM_DSBM message received
2482 New State: depends on action (IAMDSBM or Idle)
2483 Action: /* check whether other guy is better */
2484 If (ComparePrio(OwnAddrInfo, IncomingAddrInfo)) {
2485 /* I am better */
2486 send I_AM_DSBM message
2487 restart RefreshIntervalTimer
2488 continue in current state
2489 } else {
2490 Set LocalDSBMAddrInfo = IncomingAddrInfo
2491 cancel active timers
2492 start DSBMDeadIntervalTimer
2493 goto Idle State
2494 }
2495
2496 State: IAMDSBM
2497 Event: Want to shut myself down
2498 New State: DOWN
2499 Action: send DSBM_WILLING message with My address filled in, but
2500 priority set to zero
2501 goto Down State
2502
2503 <span class="h4"><a class="selflink" id="appendix-A.10.2" href="#appendix-A.10.2">A.10.2</a> Suggested Values of Interval Timers</span>
2504
2505 To avoid DSBM outages for long period, to ensure quick recovery from
2506 DSBM failures, and to avoid timeout of PATH and RESV state at the
2507 edge devices, we suggest the following values for various timers.
2508
2509 Assuming that the RSVP implementations use a 30 second timeout for
2510 PATH and RESV refreshes, we suggest that the RefreshIntervalTimer
2511 should be set to about 5 seconds with DSBMDeadIntervalTimer set to 15
2512 seconds (K=3, K*RefreshInterval). The DetectDSBMTimer should be set
2513
2514
2515
2516 <span class="grey">Yavatkar, et al. Standards Track [Page 45]</span></pre>
2517 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-46" ></span>
2518 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
2519
2520
2521 to a random value between (DSBMDeadIntervalTimer,
2522 2*DSBMDeadIntervalTimer). The ElectionIntervalTimer should be set at
2523 least to the value of DSBMDeadIntervalTimer to ensure that each SBM
2524 has a chance to have its DSBM_WILLING message (sent every
2525 RefreshInterval in ElectDSBM state) delivered to others.
2526
2527 <span class="h4"><a class="selflink" id="appendix-A.10.3" href="#appendix-A.10.3">A.10.3</a>. Guidelines for Choice of Values for SBM_PRIORITY</span>
2528
2529 Network administrators should configure SBM protocol entity at each
2530 SBM-capable device with the device's "SBM priority" for each of the
2531 interfaces attached to a managed segment. SBM_PRIORITY is an 8-bit,
2532 unsigned integer value (in the range 0-255) with higher integer
2533 values denoting higher priority. The value zero for an interface
2534 indicates that the SBM protocol entity on the device is not eligible
2535 to be a DSBM for the segment attached to the interface.
2536
2537 A separate range of values is reserved for each type of SBM-capable
2538 device to reflect the relative priority among different classes of
2539 L2/L3 devices. L2 devices get higher priority followed by routers
2540 followed by hosts. The priority values in the range of 128..255 are
2541 reserved for L2 devices, the values in the range of 64..127 are
2542 reserved for routers, and values in the range of 1..63 are reserved
2543 for hosts.
2544
2545 <span class="h3"><a class="selflink" id="appendix-A.11" href="#appendix-A.11">A.11</a>. DSBM Election over switched links</span>
2546
2547 The election algorithm works as described before in this case except
2548 each SBM-capable L2 device restricts the scope of the election to its
2549 local segment. As described in Section B.1 below, all messages
2550 related to the DSBM election are sent to a special multicast address
2551 (AllSBMAddress). AllSBMAddress (its corresponding MAC multicast
2552 address) is configured in the permanent database of SBM-capable,
2553 layer 2 devices so that all frames with AllSBMAddress as the
2554 destination address are not forwarded and instead directed to the SBM
2555 management entity in those devices. Thus, a DSBM can be elected
2556 separately on each point-to-point segment in a switched topology. For
2557 example, in Figure 2, DSBM for "segment A" will be elected using the
2558 election algorithm between R1 and S1 and none of the election-related
2559 messages on this segment will be forwarded by S1 beyond "segment A".
2560 Similarly, a separate election will take place on each segment in
2561 this topology.
2562
2563 When a switched segment is a half-duplex segment, two senders (one
2564 sender at each end of the link) share the link. In this case, one of
2565 the two senders will win the DSBM election and will be responsible
2566 for managing the segment.
2567
2568
2569
2570
2571
2572 <span class="grey">Yavatkar, et al. Standards Track [Page 46]</span></pre>
2573 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-47" ></span>
2574 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
2575
2576
2577 If a switched segment is full-duplex, exactly one sender sends on the
2578 link in each direction. In this case, either one or two DSBMs can
2579 exist on such a managed segment. If a sender at each end wishes to
2580 serve as a DSBM for that end, it can declare itself to be the DSBM by
2581 sending out an I_AM_DSBM advertisement and start managing the
2582 resources for the outgoing traffic over the segment. If one of the
2583 two senders does not wish itself to be the DSBM, then the other DSBM
2584 will not receive any DSBM advertisement from its peer and assume
2585 itself to be the DSBM for traffic traversing in both directions over
2586 the managed segment.
2587
2588
2589
2590
2591
2592
2593
2594
2595
2596
2597
2598
2599
2600
2601
2602
2603
2604
2605
2606
2607
2608
2609
2610
2611
2612
2613
2614
2615
2616
2617
2618
2619
2620
2621
2622
2623
2624
2625
2626
2627
2628 <span class="grey">Yavatkar, et al. Standards Track [Page 47]</span></pre>
2629 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-48" ></span>
2630 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
2631
2632
2633 Appendix B Message Encapsulation and Formats
2634
2635 To minimize changes to the existing RSVP implementations and to
2636 ensure quick deployment of a SBM in conjunction with RSVP, all
2637 communication to and from a DSBM will be performed using messages
2638 constructed using the current rules for RSVP message formats and raw
2639 IP encapsulation. For more details on the RSVP message formats, refer
2640 to the RSVP specification (<a href="./rfc2205">RFC 2205</a>). No changes to the RSVP message
2641 formats are proposed, but new message types and new L2-specific
2642 objects are added to the RSVP message formats to accommodate DSBM-
2643 related messages. These additions are described below.
2644
2645 <span class="h3"><a class="selflink" id="appendix-B.1" href="#appendix-B.1">B.1</a> Message Addressing</span>
2646
2647 For the purpose of DSBM election and detection, AllSBMAddress is used
2648 as the destination address while sending out both DSBM_WILLING and
2649 I_AM_DSBM messages. A DSBM client first detects a managed segment by
2650 listening to I_AM_DSBM advertisements and records the DSBMAddress
2651 (unicast IP address of the DSBM).
2652
2653 <span class="h3"><a class="selflink" id="appendix-B.2" href="#appendix-B.2">B.2</a>. Message Sizes</span>
2654
2655 Each message must occupy exactly one IP datagram. If it exceeds the
2656 MTU, such a datagram will be fragmented by IP and reassembled at the
2657 recipient node. This has a consequence that a single message may not
2658 exceed the maximum IP datagram size, approximately 64K bytes.
2659
2660 <span class="h3"><a class="selflink" id="appendix-B.3" href="#appendix-B.3">B.3</a>. RSVP-related Message Formats</span>
2661
2662 All RSVP messages directed to and from a DSBM may contain various
2663 RSVP objects defined in the RSVP specification and messages continue
2664 to follow the formatting rules specified in the RSVP specification.
2665 In addition, an RSVP implementation must also recognize new object
2666 classes that are described below.
2667
2668 <span class="h4"><a class="selflink" id="appendix-B.3.1" href="#appendix-B.3.1">B.3.1</a>. Object Formats</span>
2669
2670 All objects are defined using the format specified in the RSVP
2671 specification. Each object has a 32-bit header that contains length
2672 (of the object in bytes including the object header), the object
2673 class number, and a C-Type. All unused fields should be set to zero
2674 and ignored on receipt.
2675
2676
2677
2678
2679
2680
2681
2682
2683
2684 <span class="grey">Yavatkar, et al. Standards Track [Page 48]</span></pre>
2685 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-49" ></span>
2686 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
2687
2688
2689 <span class="h4"><a class="selflink" id="appendix-B.3.2" href="#appendix-B.3.2">B.3.2</a>. SBM Specific Objects</span>
2690
2691 Note that the Class-Num values for the SBM specific objects
2692 (LAN_NHOP, LAN_LOOPBACK, and RSVP_HOP_L2) are chosen from the
2693 codespace 10XXXXXX. This coding assures that non-SBM aware RSVP nodes
2694 will ignore the objects without forwarding them or generating an
2695 error message.
2696
2697 Within the SBM specific codespace, note the following interpretation
2698 of the third most significant bit of the Class-Num:
2699
2700 a) Objects of the form 100XXXXX are to be silently
2701 discarded by SBM nodes that do not recognize them.
2702
2703 b) Objects of the form 101XXXXX are to be silently
2704 forwarded by SBM nodes that do not recognize them.
2705
2706 <span class="h4"><a class="selflink" id="appendix-B.3.3" href="#appendix-B.3.3">B.3.3</a>. IEEE 802 Canonical Address Format</span>
2707
2708 The 48-bit MAC Addresses used by IEEE 802 were originally defined in
2709 terms of wire order transmission of bits in the source and
2710 destination MAC address fields. The same wire order applied to both
2711 Ethernet and Token Ring. Since the bit transmission order of Ethernet
2712 and Token Ring data differ - Ethernet octets are transmitted least
2713 significant bit first, Token Ring most significant first - the
2714 numeric values naturally associated with the same address on
2715 different 802 media differ. To facilitate the communication of
2716 address values in higher layer protocols which might span both token
2717 ring and Ethernet attached systems connected by bridges, it was
2718 necessary to define one reference format - the so called canonical
2719 format for these addresses. Formally the canonical format defines the
2720 value of the address, separate from the encoding rules used for
2721 transmission. It comprises a sequence of octets derived from the
2722 original wire order transmission bit order as follows. The least
2723 significant bit of the first octet is the first bit transmitted, the
2724 next least significant bit the second bit, and so on to the most
2725 significant bit of the first octet being the 8th bit transmitted; the
2726 least significant bit of the second octet is the 9th bit transmitted,
2727 and so on to the most significant bit of the sixth octet of the
2728 canonical format being the last bit of the address transmitted.
2729
2730 This canonical format corresponds to the natural value of the address
2731 octets for Ethernet. The actual transmission order or formal encoding
2732 rules for addresses on media which do not transmit bit serially are
2733 derived from the canonical format octet values.
2734
2735
2736
2737
2738
2739
2740 <span class="grey">Yavatkar, et al. Standards Track [Page 49]</span></pre>
2741 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-50" ></span>
2742 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
2743
2744
2745 This document requires that all L2 addresses used in conjunction with
2746 the SBM protocol be encoded in the canonical format as a sequence of
2747 6 octets. In the following, we define the object formats for objects
2748 that contain L2 addresses that are based on the canonical
2749 representation.
2750
2751 <span class="h4"><a class="selflink" id="appendix-B.3.4" href="#appendix-B.3.4">B.3.4</a>. RSVP_HOP_L2 object</span>
2752
2753 RSVP_HOP_L2 object uses object class = 161; it contains the L2
2754 address of the previous hop L3 device in the IEEE Canonical address
2755 format discussed above.
2756
2757 RSVP_HOP_L2 object: class = 161, C-Type represents the addressing
2758 format used. In our case, C-Type=1 represents the IEEE Canonical
2759 Address format.
2760
2761 0 1 2 3
2762 +---------------+---------------+---------------+----------------+
2763 | Length | 161 |C-Type(addrtype)|
2764 +---------------+---------------+---------------+----------------+
2765 | Variable length Opaque data |
2766 +---------------+---------------+---------------+----------------+
2767
2768 C-Type = 1 (IEEE Canonical Address format)
2769
2770 When C-Type=1, the object format is:
2771
2772 0 1 2 3
2773 +---------------+---------------+---------------+---------------+
2774 | 12 | 161 | 1 |
2775 +---------------+---------------+---------------+---------------+
2776 | Octets 0-3 of the MAC address |
2777 +---------------+---------------+---------------+---------------+
2778 | Octets 4-5 of the MAC addr. | /// | /// |
2779 +---------------+---------------+---------------+---------------+
2780
2781 /// -- unused (set to zero)
2782
2783 <span class="h4"><a class="selflink" id="appendix-B.3.5" href="#appendix-B.3.5">B.3.5</a>. LAN_NHOP object</span>
2784
2785 LAN_NHOP object represents two objects, namely, LAN_NHOP_L3 address
2786 object and LAN_NHOP_L2 address object.
2787 &lt;LAN_NHOP object&gt; ::= &lt;LAN_NHOP_L2 object&gt; &lt;LAN_NHOP_L3 object&gt;
2788
2789 LAN_NHOP_L2 address object uses object class = 162 and uses the same
2790 format (but different class number) as the RSVP_HOP_L2 object. It
2791 provides the L2 or MAC address of the next hop L3 device.
2792
2793
2794
2795
2796 <span class="grey">Yavatkar, et al. Standards Track [Page 50]</span></pre>
2797 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-51" ></span>
2798 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
2799
2800
2801 0 1 2 3
2802 +---------------+---------------+---------------+----------------+
2803 | Length | 162 |C-Type(addrtype)|
2804 +---------------+---------------+---------------+----------------+
2805 | Variable length Opaque data |
2806 +---------------+---------------+---------------+----------------+
2807
2808 C-Type = 1 (IEEE 802 Canonical Address Format as defined below) See
2809 the RSVP_HOP_L2 address object for more details.
2810
2811 LAN_NHOP_L3 object uses object class = 163 and gives the L3 or IP
2812 address of the next hop L3 device.
2813
2814 LAN_NHOP_L3 object: class = 163, C-Type specifies IPv4 or IPv6
2815 address family used.
2816
2817 IPv4 LAN_NHOP_L3 object: class =163, C-Type = 1
2818 +---------------+---------------+---------------+---------------+
2819 | Length = 8 | 163 | 1 |
2820 +---------------+---------------+---------------+---------------+
2821 | IPv4 NHOP address |
2822 +---------------------------------------------------------------+
2823
2824 IPv6 LAN_NHOP_L3 object: class =163, C-Type = 2
2825 +---------------+---------------+---------------+---------------+
2826 | Length = 20 | 163 | 2 |
2827 +---------------+---------------+---------------+---------------+
2828 | IPv6 NHOP address (16 bytes) |
2829 +---------------------------------------------------------------+
2830
2831 <span class="h4"><a class="selflink" id="appendix-B.3.6" href="#appendix-B.3.6">B.3.6</a>. LAN_LOOPBACK Object</span>
2832
2833 The LAN_LOOPBACK object gives the IP address of the outgoing
2834 interface for a PATH message and uses object class=164; both IPv4 and
2835 IPv6 formats are specified.
2836
2837 IPv4 LAN_LOOPBACK object: class = 164, C-Type = 1
2838
2839 0 1 2 3
2840 +---------------+---------------+---------------+---------------+
2841 | Length | 164 | 1 |
2842 +---------------+---------------+---------------+---------------+
2843 | IPV4 address of an interface |
2844 +---------------+---------------+---------------+---------------+
2845
2846
2847
2848
2849
2850
2851
2852 <span class="grey">Yavatkar, et al. Standards Track [Page 51]</span></pre>
2853 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-52" ></span>
2854 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
2855
2856
2857 IPv6 LAN_LOOPBACK object: class = 164, C-Type = 2
2858
2859 +---------------+---------------+---------------+---------------+
2860 | Length | 164 | 2 |
2861 +---------------+---------------+---------------+---------------+
2862 | |
2863 + +
2864 | |
2865 + IPV6 address of an interface +
2866 | |
2867 + +
2868 | |
2869 +---------------+---------------+---------------+---------------+
2870
2871 <span class="h4"><a class="selflink" id="appendix-B.3.7" href="#appendix-B.3.7">B.3.7</a>. TCLASS Object</span>
2872
2873 TCLASS object (traffic class based on IEEE 802.1p) uses object
2874 class = 165.
2875
2876 0 1 2 3
2877 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
2878 | Length | 165 | 1 |
2879 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
2880 | /// | /// | /// | /// | PV |
2881 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
2882
2883 Only 3 bits in data contain the user_priority value (PV).
2884
2885 <span class="h3"><a class="selflink" id="appendix-B.4" href="#appendix-B.4">B.4</a>. RSVP PATH and PATH_TEAR Message Formats</span>
2886
2887 As specified in the RSVP specification, a PATH and PATH_TEAR messages
2888 contain the RSVP Common Header and the relevant RSVP objects.
2889
2890 For the RSVP Common Header, refer to the RSVP specification (<a href="./rfc2205">RFC</a>
2891 <a href="./rfc2205">2205</a>). Enhancements to an RSVP_PATH message include additional
2892 objects as specified below.
2893
2894 &lt;PATH Message&gt; ::= &lt;RSVP Common Header&gt; [&lt;INTEGRITY&gt;]
2895 &lt;RSVP_HOP_L2&gt; &lt;LAN_NHOP&gt;
2896 &lt;LAN_LOOPBACK&gt; [&lt;TCLASS&gt;] &lt;SESSION&gt;&lt;RSVP_HOP&gt;
2897 &lt;TIME_VALUES&gt; [&lt;POLICY DATA&gt;] &lt;sender descriptor&gt;
2898
2899 &lt;PATH_TEAR Message&gt; ::= &lt;RSVP Common Header&gt; [&lt;INTEGRITY&gt;]
2900 &lt;LAN_LOOPBACK&gt; &lt;LAN_NHOP&gt; &lt;SESSION&gt; &lt;RSVP_HOP&gt;
2901 [&lt;sender descriptor&gt;]
2902
2903
2904
2905
2906
2907
2908 <span class="grey">Yavatkar, et al. Standards Track [Page 52]</span></pre>
2909 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-53" ></span>
2910 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
2911
2912
2913 If the INTEGRITY object is present, it must immediately follow the
2914 RSVP common header. L2-specific objects must always precede the
2915 SESSION object.
2916
2917 <span class="h3"><a class="selflink" id="appendix-B.5" href="#appendix-B.5">B.5</a>. RSVP RESV Message Format</span>
2918
2919 As specified in the RSVP specification, an RSVP_RESV message contains
2920 the RSVP Common Header and relevant RSVP objects. In addition, it may
2921 contain an optional TCLASS object as described earlier.
2922
2923 <span class="h3"><a class="selflink" id="appendix-B.6" href="#appendix-B.6">B.6</a>. Additional RSVP message types to handle SBM interactions</span>
2924
2925 New RSVP message types are introduced to allow interactions between a
2926 DSBM and an RSVP node (host/router) for the purpose of discovering
2927 and binding to a DSBM. New RSVP message types needed are as follows:
2928
2929 RSVP Msg Type (8 bits) Value
2930 DSBM_WILLING 66
2931 I_AM_DSBM 67
2932
2933 All SBM-specific messages are formatted as RSVP messages with an RSVP
2934 common header followed by SBM-specific objects.
2935
2936 &lt;SBMP_MESSAGE&gt; ::= &lt;SBMP common header&gt; &lt;SBM-specific objects&gt;
2937
2938 where &lt;SBMP common header&gt; ::= &lt;RSVP common Header&gt; [&lt;INTEGRITY&gt;]
2939
2940 For each SBM message type, there is a set of rules for the
2941 permissible choice of object types. These rules are specified using
2942
2943 Backus-Naur Form (BNF) augmented with square brackets surrounding
2944 optional sub-sequences. The BNF implies an order for the objects in a
2945 message. However, in many (but not all) cases, object order makes no
2946 logical difference. An implementation should create messages with the
2947 objects in the order shown here, but accept the objects in any
2948 permissible order. Any exceptions to this rule will be pointed out in
2949 the specific message formats.
2950
2951 DSBM_WILLING Message
2952
2953 &lt;DSBM_WILLING message&gt; ::= &lt;SBM Common Header&gt; &lt;DSBM IP ADDRESS&gt;
2954 &lt;DSBM L2 address&gt; &lt;SBM PRIORITY&gt;
2955
2956 I_AM_DSBM Message
2957
2958 &lt;I_AM_DSBM&gt; ::= &lt;SBM Common Header&gt; &lt;DSBM IP ADDRESS&gt; &lt;DSBM L2 address&gt;
2959 &lt;SBM PRIORITY&gt; &lt;DSBM Timer Intervals&gt;
2960 [&lt;NON_RESV_SEND_LIMIT&gt;]
2961
2962
2963
2964 <span class="grey">Yavatkar, et al. Standards Track [Page 53]</span></pre>
2965 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-54" ></span>
2966 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
2967
2968
2969 For compatibility reasons, receivers of the I_AM_DSBM message must be
2970 prepared to receive additional objects of the Unknown Class type
2971 [<a href="./rfc2205">RFC-2205</a>].
2972
2973 All I_AM_DSBM messages are multicast to the well known AllSBMAddress.
2974 The default priority of a SBM is 1 and higher priority values
2975 represent higher precedence. The priority value zero indicates that
2976 the SBM is not eligible to be the DSBM.
2977
2978 Relevant Objects
2979
2980 DSBM IP ADDRESS objects use object class = 42; IPv4 DSBM IP ADDRESS
2981 object uses &lt;Class=42, C-Type=1&gt; and IPv6 DSBM IP ADDRESS object uses
2982 &lt;Class=42, C-Type=2&gt;.
2983
2984 IPv4 DSBM IP ADDRESS object: class = 42, C-Type =1
2985 0 1 2 3
2986 +---------------+---------------+---------------+---------------+
2987 | IPv4 DSBM IP Address |
2988 +---------------+---------------+---------------+---------------+
2989
2990 IPv6 DSBM IP ADDRESS object: Class = 42, C-Type = 2
2991
2992 +---------------+---------------+---------------+---------------+
2993 | |
2994 + +
2995 | |
2996 + IPv6 DSBM IP Address +
2997 | |
2998 + +
2999 | |
3000 +---------------+---------------+---------------+---------------+
3001
3002 &lt;DSBM L2 address&gt; Object is the same as &lt;RSVP_HOP_L2&gt; object with C-
3003 Type = 1 for IEEE Canonical Address format.
3004
3005 &lt;DSBM L2 address&gt; ::= &lt;RSVP_HOP_L2&gt;
3006
3007 A SBM may omit this object by including a NULL L2 address object.
3008 For C-Type=1 (IEEE Canonical address format), such a version of the
3009 L2 address object contains value zero in the six octets corresponding
3010 to the MAC address (see section B.3.4 for the exact format).
3011
3012
3013
3014
3015
3016
3017
3018
3019
3020 <span class="grey">Yavatkar, et al. Standards Track [Page 54]</span></pre>
3021 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-55" ></span>
3022 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
3023
3024
3025 SBM_PRIORITY Object: class = 43, C-Type =1
3026
3027 0 1 2 3
3028 +---------------+---------------+---------------+---------------+
3029 | /// | /// | /// | SBM priority |
3030 +---------------+---------------+---------------+---------------+
3031
3032 TIMER INTERVAL VALUES.
3033
3034 The two timer intervals, namely, DSBM Dead Interval and DSBM Refresh
3035 Interval, are specified as integer values each in the range of 0..255
3036 seconds. Both values are included in a single "DSBM Timer Intervals"
3037 object described below.
3038
3039 DSBM Timer Intervals Object: class = 44, C-Type =1
3040
3041 +---------------+---------------+---------------+----------------+
3042 | /// | /// | DeadInterval | RefreshInterval|
3043 +---------------+---------------+---------------+----------------+
3044
3045 NON_RESV_SEND_LIMIT Object: class = 45, C-Type = 1
3046
3047 0 1 2 3
3048 +---------------+---------------+---------------+----------------+
3049 | NonResvSendLimit(limit on traffic allowed to send without RESV)|
3050 | |
3051 +---------------+---------------+---------------+----------------+
3052
3053 &lt;NonResvSendLimit&gt; ::= &lt;Intserv Sender_TSPEC object&gt;
3054 (class=12, C-Type =2)
3055
3056 The NON_RESV_SEND_LIMIT object specifies a per-flow limit on the
3057 profile of traffic which a sending host is allowed to send onto a
3058 managed segment without a valid RSVP reservation (see <a href="#appendix-C">Appendix C</a> for
3059 further details on the usage of this object). The object contains the
3060 NonResvSendLimit parameter. This parameter is equivalent to the
3061 Intserv SENDER_TSPEC (see <a href="./rfc2210">RFC 2210</a> for contents and encoding rules).
3062 The SENDER_TSPEC includes five parameters which describe a traffic
3063 profile (r, b, p, m and M). Sending hosts compare the SENDER_TSPEC
3064 describing a sender traffic flow to the SENDER_TSPEC advertised by
3065 the DSBM. If the SENDER_TSPEC of the traffic flow in question is less
3066 than or equal to the SENDER_TSPEC advertised by the DSBM, it is
3067 allowable to send traffic on the corresponding flow without a valid
3068 RSVP reservation in place. Otherwise it is not.
3069
3070 The network administrator may configure the DSBM to disallow any sent
3071 traffic in the absence of an RSVP reservation by configuring a
3072 NonResvSendLimit in which r = 0, b = 0, p = 0, m = infinity and M =
3073
3074
3075
3076 <span class="grey">Yavatkar, et al. Standards Track [Page 55]</span></pre>
3077 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-56" ></span>
3078 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
3079
3080
3081 0. Similarly the network administrator may allow any traffic to be
3082 sent in the absence of an RSVP reservation by configuring a
3083 NonResvSendLimit in which r = infinity, b = infinity, p = infinity, m
3084 = 0 and M = infinity. Of course, any of these parameters may be set
3085 to values between zero and infinity to advertise finite per-flow
3086 limits.
3087
3088 The NON_RESV_SEND_LIMIT object is optional. Senders on a managed
3089 segment should interpret the absence of the NON_RESV_SEND_LIMIT
3090 object as equivalent to an infinitely large SENDER_TSPEC (it is
3091 permissible to send any traffic profile in the absence of an RSVP
3092 reservation).
3093
3094
3095
3096
3097
3098
3099
3100
3101
3102
3103
3104
3105
3106
3107
3108
3109
3110
3111
3112
3113
3114
3115
3116
3117
3118
3119
3120
3121
3122
3123
3124
3125
3126
3127
3128
3129
3130
3131
3132 <span class="grey">Yavatkar, et al. Standards Track [Page 56]</span></pre>
3133 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-57" ></span>
3134 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
3135
3136
3137 Appendix C The DSBM as a Source of Centralized Configuration Information
3138
3139 There are certain configuration parameters which it may be useful to
3140 distribute to layer-3 senders on a managed segment. The DSBM may
3141 serve as a centralized management point from which such parameters
3142 can easily be distributed. In particular, it is possible for the
3143 network administrator configuring a DSBM to cause certain
3144 configuration parameters to be distributed as objects appended to the
3145 I_AM_DSBM messages. The following configuration object is defined at
3146 this time. Others may be defined in the future. See <a href="#appendix-B">Appendix B</a> for
3147 further details regarding the NON_RESV_SEND_LIMIT object.
3148
3149 <span class="h3"><a class="selflink" id="appendix-C.1" href="#appendix-C.1">C.1</a>. NON_RESV_SEND_LIMIT</span>
3150
3151 As we QoS enable layer 2 segments, we expect an evolution from
3152 subnets comprised of traditional shared segments (with no means of
3153 traffic separation and no DSBM), to subnets comprised of dedicated
3154 segments switched by sophisticated switches (with both DSBM and
3155 802.1p traffic separation capability).
3156
3157 A set of intermediate configurations consists of a group of QoS
3158 enabled hosts sending onto a traditional shared segment. A layer-3
3159 device (or a layer-2 device) acts as a DSBM for the shared segment,
3160 but cannot enforce traffic separation. In such a configuration, the
3161 DSBM can be configured to limit the number of reservations approved
3162 for senders on the segment, but cannot prevent them from sending. As
3163 a result, senders may congest the segment even though a network
3164 administrator has configured an appropriate limit for admission
3165 control in the DSBM.
3166
3167 One solution to this problem which would give the network
3168 administrator control over the segment, is to require applications
3169 (or operating systems on behalf of applications) not to send until
3170 they have obtained a reservation. This is problematic as most
3171 applications are used to sending as soon as they wish to and expect
3172 to get whatever service quality the network is able to grant at that
3173 time. Furthermore, it may often be acceptable to allow certain
3174 applications to send before a reservation is received. For example,
3175 on a segment comprised of a single 10 Mbps ethernet and 10 hosts, it
3176 may be acceptable to allow a 16 Kbps telephony stream to be
3177 transmitted but not a 3 Mbps video stream.
3178
3179 A more pragmatic solution then, is to allow the network administrator
3180 to set a per-flow limit on the amount of non-adaptive traffic which a
3181 sender is allowed to generate on a managed segment in the absence of
3182 a valid reservation. This limit is advertised by the DSBM and
3183 received by sending hosts. An API on the sending host can then
3184 approve or deny an application's QoS request based on the resources
3185
3186
3187
3188 <span class="grey">Yavatkar, et al. Standards Track [Page 57]</span></pre>
3189 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-58" ></span>
3190 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
3191
3192
3193 requested.
3194
3195 The NON_RESV_SEND_LIMIT object can be used to advertise a Flowspec
3196 which describes the shape of traffic that a sender is allowed to
3197 generate on a managed segment when its RSVP reservation requests have
3198 either not yet completed or have been rejected.
3199
3200 ACKNOWLEDGEMENTS
3201
3202 Authors are grateful to Eric Crawley (Argon), Russ Fenger (Intel),
3203 David Melman (Siemens), Ramesh Pabbati (Microsoft), Mick Seaman
3204 (3COM), Andrew Smith (Extreme Networks) for their constructive
3205 comments on the SBM design and the earlier versions of this document.
3206
3207 <span class="h2"><a class="selflink" id="section-6" href="#section-6">6</a>. Authors' Addresses</span>
3208
3209 Raj Yavatkar
3210 Intel Corporation
3211 2111 N.E. 25th Avenue,
3212 Hillsboro, OR 97124
3213 USA
3214
3215 Phone: +1 503-264-9077
3216 EMail: [email protected]
3217
3218
3219 Don Hoffman
3220 Teledesic Corporation
3221 2300 Carillon Point
3222 Kirkland, WA 98033
3223 USA
3224
3225 Phone: +1 425-602-0000
3226
3227
3228 Yoram Bernet
3229 Microsoft
3230 1 Microsoft Way
3231 Redmond, WA 98052
3232 USA
3233
3234 Phone: +1 206 936 9568
3235 EMail: [email protected]
3236
3237
3238
3239
3240
3241
3242
3243
3244 <span class="grey">Yavatkar, et al. Standards Track [Page 58]</span></pre>
3245 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-59" ></span>
3246 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
3247
3248
3249 Fred Baker
3250 Cisco Systems
3251 519 Lado Drive
3252 Santa Barbara, California 93111
3253 USA
3254
3255 Phone: +1 408 526 4257
3256 EMail: [email protected]
3257
3258
3259 Michael Speer
3260 Sun Microsystems, Inc
3261 901 San Antonio Road UMPK15-215
3262 Palo Alto, CA 94303
3263
3264 Phone: +1 650-786-6368
3265 EMail: [email protected]
3266
3267
3268
3269
3270
3271
3272
3273
3274
3275
3276
3277
3278
3279
3280
3281
3282
3283
3284
3285
3286
3287
3288
3289
3290
3291
3292
3293
3294
3295
3296
3297
3298
3299
3300 <span class="grey">Yavatkar, et al. Standards Track [Page 59]</span></pre>
3301 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-60" ></span>
3302 <span class="grey"><a href="./rfc2814">RFC 2814</a> SBM (Subnet Bandwidth Manager) May 2000</span>
3303
3304
3305 Full Copyright Statement
3306
3307 Copyright (C) The Internet Society (2000). All Rights Reserved.
3308
3309 This document and translations of it may be copied and furnished to
3310 others, and derivative works that comment on or otherwise explain it
3311 or assist in its implementation may be prepared, copied, published
3312 and distributed, in whole or in part, without restriction of any
3313 kind, provided that the above copyright notice and this paragraph are
3314 included on all such copies and derivative works. However, this
3315 document itself may not be modified in any way, such as by removing
3316 the copyright notice or references to the Internet Society or other
3317 Internet organizations, except as needed for the purpose of
3318 developing Internet standards in which case the procedures for
3319 copyrights defined in the Internet Standards process must be
3320 followed, or as required to translate it into languages other than
3321 English.
3322
3323 The limited permissions granted above are perpetual and will not be
3324 revoked by the Internet Society or its successors or assigns.
3325
3326 This document and the information contained herein is provided on an
3327 "AS IS" basis and THE INTERNET SOCIETY AND THE INTERNET ENGINEERING
3328 TASK FORCE DISCLAIMS ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING
3329 BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION
3330 HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF
3331 MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.
3332
3333 Acknowledgement
3334
3335 Funding for the RFC Editor function is currently provided by the
3336 Internet Society.
3337
3338
3339
3340
3341
3342
3343
3344
3345
3346
3347
3348
3349
3350
3351
3352
3353
3354
3355
3356 Yavatkar, et al. Standards Track [Page 60]
3357 </pre>