Debian Codegraph @

rfc2098.html

raw
1 <pre>Network Working Group Y. Katsube
2 Request for Comments: 2098 K. Nagami
3 Category: Informational H. Esaki
4 Toshiba R&amp;D Center
5 February 1997
6
7
8 <span class="h1">Toshiba's Router Architecture Extensions for ATM : Overview</span>
9
10 Status of this Memo
11
12 This memo provides information for the Internet community. This memo
13 does not specify an Internet standard of any kind. Distribution of
14 this memo is unlimited.
15
16 Abstract
17
18 This memo describes a new internetworking architecture which makes
19 better use of the property of ATM. IP datagrams are transferred
20 along hop-by-hop path via routers, but datagram assembly/disassembly
21 and IP header processing are not necessarily carried out at
22 individual routers in the proposed architecture. A concept of "Cell
23 Switch Router (CSR)" is introduced as a new internetworking
24 equipment, which has ATM cell switching capabilities in addition to
25 conventional IP datagram forwarding. Proposed architecture can
26 provide applications with high-throughput and low-latency ATM pipes
27 while retaining current router-based internetworking concept. It
28 also provides applications with specific QoS/bandwidth by cooperating
29 with internetworking level resource reservation protocols such as
30 RSVP.
31
32 <span class="h2"><a class="selflink" id="section-1" href="#section-1">1</a>. Introduction</span>
33
34 The Internet is growing both in its size and its traffic volume. In
35 addition, recent applications often require guaranteed bandwidth and
36 QoS rather than best effort. Such changes make the current hop-by-
37 hop datagram forwarding paradigm inadequate, then accelerate
38 investigations on new internetworking architectures.
39
40 Roughly two distinct approaches can be seen as possible solutions;
41 the use of ATM to convey IP datagrams, and the revision of IP to
42 support flow concept and resource reservation. Integration or
43 interworking of these approaches will be necessary to provide end
44 hosts with high throughput and QoS guaranteed internetworking
45 services over any datalink platforms as well as ATM.
46
47 New internetworking architecture proposed in this draft is based on
48 "Cell Switch Router (CSR)" which has the following properties.
49
50
51
52 <span class="grey">Katsube, et. al. Informational [Page 1]</span></pre>
53 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-2" ></span>
54 <span class="grey"><a href="./rfc2098">RFC 2098</a> Toshiba's Router Extension for ATM February 1997</span>
55
56
57 - It makes the best use of ATM's property while retaining current
58 router-based internetworking and routing architecture.
59
60 - It takes into account interoperability with future IP that
61 supports flow concept and resource reservations.
62
63 <a href="#section-2">Section 2</a> of this draft explains background and motivations of our
64 proposal. <a href="#section-3">Section 3</a> describes an overview of the proposed
65 internetworking architecture and its several remarkable features.
66 <a href="#section-4">Section 4</a> discusses control architectures for CSR, which will need to
67 be further investigated.
68
69 <span class="h2"><a class="selflink" id="section-2" href="#section-2">2</a>. Background and Motivation</span>
70
71 It is considered that the current hop-by-hop best effort datagram
72 forwarding paradigm will not be adequate to support future large
73 scale Internet which accommodates huge amount of traffic with certain
74 QoS requirements. Two major schools of investigations can be seen in
75 IETF whose main purpose is to improve ability of the Internet with
76 regard to its throughput and QoS. One is to utilize ATM technology
77 as much as possible, and the other is to introduce the concept of
78 resource reservation and flow into IP.
79
80 1) Utilization of ATM
81
82 Although basic properties of ATM; necessity of connection setup,
83 necessity of traffic contract, etc.; is not necessarily suited to
84 conventional IP datagram transmission, its excellent throughput and
85 delay characteristics let us to investigate the realization of IP
86 datagram transmission over ATM.
87
88 A typical internetworking architecture is the "Classical IP Model"
89 [<a href="./rfc1577" title="&quot;Classical IP and ARP over ATM&quot;">RFC1577</a>]. This model allows direct ATM connectivities only between
90 nodes that share the same IP address prefix. IP datagrams should
91 traverse routers whenever they go beyond IP subnet boundaries even
92 though their source and destination are accommodated in the same ATM
93 cloud. Although an ATMARP is introduced which is not based on legacy
94 datalink broadcast but on centralized ATMARP servers, this model does
95 not require drastic changes to the legacy internetworking
96 architectures with regard to the IP datagram forwarding process.
97 This model still has problems of limited throughput and large
98 latency, compared with the ability of ATM, due to IP header
99 processing at every router. It will become more critical when
100 multimedia applications that require much larger bandwidth and lower
101 latency become dominant in the near future.
102
103
104
105
106
107
108 <span class="grey">Katsube, et. al. Informational [Page 2]</span></pre>
109 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-3" ></span>
110 <span class="grey"><a href="./rfc2098">RFC 2098</a> Toshiba's Router Extension for ATM February 1997</span>
111
112
113 Another internetworking architecture is "NHRP (Next Hop Resolution
114 Protocol) Model" [<a href="#ref-NHRP09" title="&quot;NBMA Next Hop Resolution Protocol (NHRP)&quot;">NHRP09</a>]. This model aims at resolving throughput
115 and latency problems in the Classical IP Model and making the best
116 use of ATM. ATM connections can be directly established from an
117 ingress point to an egress point of an ATM cloud even when they do
118 not share the same IP address prefix. In order to enable it, the
119 Next Hop Server [KAT95] is introduced which can find an egress point
120 of the ATM cloud nearest to the given destination and resolves its
121 ATM address. A sort of query/response protocols between the
122 server(s) and clients and possibly server and server are specified.
123 After the ATM address of a desired egress point is resolved, the
124 client establishes a direct ATM connection to that point through ATM
125 signaling procedures [<a href="#ref-ATM3.1" title="&quot;ATM User-Network Interface Specification, v.3.1&quot;">ATM3.1</a>]. Once a direct ATM connection has been
126 set up through this procedure, IP datagrams do not have to experience
127 hop-by-hop IP processing but can be transmitted over the direct ATM
128 connection. Therefore, high throughput and low latency
129 communications become possible even if they go beyond IP subnet
130 boundaries. It should be noted that the provision of such direct ATM
131 connections does not mean disappearance of legacy routers which
132 interconnect distinct ATM-based IP subnets. For example, hop-by-hop
133 IP datagram forwarding function would still be required in the
134 following cases:
135
136 - When you want to transmit IP datagrams before direct ATM connection
137 from an ingress point to an egress point of the ATM cloud is
138 established
139
140 - When you neither require a certain QoS nor transmit large amount of
141 IP datagrams for some communication
142
143 - When the direct ATM connection is not allowed by security or policy
144 reasons
145
146 2) IP level resource reservation and flow support
147
148 Apart from investigation on specific datalink technology such as ATM,
149 resource reservation technologies for desired IP level flows have
150 been studied and are still under discussion. Their typical examples
151 are RSVP [<a href="#ref-RSVP13" title="&quot;Resource ReSerVation Protocol (RSVP), Version 1 Functional Specification&quot;">RSVP13</a>] and STII [<a href="./rfc1819" title="&quot;Internet STream Protocol Version 2 (STII) Protocol Specification Version ST2+&quot;">RFC1819</a>].
152
153 RSVP itself is not a connection oriented technology since datagrams
154 can be transmitted regardless of the result of the resource
155 reservation process. After a resource reservation process from a
156 receiver (or receivers) to a sender (or senders) is successfully
157 completed, RSVP-capable routers along the path of the flow reserve
158 their resources for datagram forwarding according to the requested
159 flow spec.
160
161
162
163
164 <span class="grey">Katsube, et. al. Informational [Page 3]</span></pre>
165 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-4" ></span>
166 <span class="grey"><a href="./rfc2098">RFC 2098</a> Toshiba's Router Extension for ATM February 1997</span>
167
168
169 STII is regarded as a connection oriented IP which requires
170 connection setup process from a sender to a receiver (or receivers)
171 before transmitting datagrams. STII-capable routers along the path
172 of the requested connection reserve their resources for datagram
173 forwarding according to the flow spec.
174
175 Neither RSVP nor STII restrict underlying datalink networks since
176 their primary purpose is to let routers provide each IP flow with
177 desired forwarding quality (by controlling their datagram scheduling
178 rules). Since various datalink networks will coexist as well as ATM
179 in the future, these IP level resource reservation technologies would
180 be necessary in order to provide end-to-end IP flow with desired
181 bandwidth and QoS.
182
183 aking this background into consideration, we should be aware of
184 several issues which motivate our proposal.
185
186 - As of the time of writing, the ATM specific internetworking
187 architecture proposed does not take into account interoperability
188 with IP level resource reservation or connection setup protocols.
189 In particular, operating RSVP in the NHRP-based ATM cloud seems to
190 require much effort since RSVP is a soft-state receiver-oriented
191 protocol with multicast capability as a default, while ATM with
192 NHRP is a hard-state sender-oriented protocol which does not
193 support multicast yet.
194
195 - Although RSVP or STII-based routers will provide each IP flow with
196 a desired bandwidth and QoS, they have some native throughput
197 limitations due to the processor-based IP forwarding mechanism
198 compared with the hardware switching mechanism of ATM.
199
200 The main objective of our proposal is to resolve the above issues.
201
202 The proposed internetworking architecture makes the best use of the
203 property of ATM by extending legacy routers to handle future IP
204 features such as flow support and resource reservation with the help
205 of ATM's cell switching capabilities.
206
207 <span class="h2"><a class="selflink" id="section-3" href="#section-3">3</a>. Internetworking Architecture Based On the Cell Switch Router (CSR)</span>
208
209 <span class="h3"><a class="selflink" id="section-3.1" href="#section-3.1">3.1</a> Overview</span>
210
211 The Cell Switch Router (CSR) is a key network element of the proposed
212 internetworking architecture. The CSR provides cell switching
213 functionality in addition to conventional IP datagram forwarding.
214 Communications with high throughput and low latency, that are native
215 properties of ATM, become possible by using this cell switching
216 functionality even when the communications pass through IP subnetwork
217
218
219
220 <span class="grey">Katsube, et. al. Informational [Page 4]</span></pre>
221 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-5" ></span>
222 <span class="grey"><a href="./rfc2098">RFC 2098</a> Toshiba's Router Extension for ATM February 1997</span>
223
224
225 boundaries. In an ATM internet composed of CSRs, VPI/VCI-based cell
226 switching which bypasses datagram assembly/disassembly and IP header
227 processing is possible at every CSR for communications which lend
228 themselves to such (e.g., communications which require certain amount
229 of bandwidth and QoS), while conventional hop-by-hop datagram
230 forwarding based on the IP header is also possible at every CSR for
231 other conventional communications.
232
233 By using such cell-level switching capabilities, the CSR is able to
234 concatenate incoming and outgoing ATM VCs, although the concatenation
235 in this case is controlled outside the ATM cloud (ATM's control/
236 management-plane) unlike conventional ATM switch nodes. That is, the
237 CSR is attached to ATM networks via an ATM-UNI instead of NNI. By
238 carrying out such VPI/VCI concatenations at multiple CSRs
239 consecutively, ATM level connectivity composed of multiple ATM VCs,
240 each of which connects adjacent CSRs (or CSR and hosts/routers), can
241 be provided. We call such an ATM pipe "ATM Bypass-pipe" to
242 differentiate it from "ATM VCC (VC connection)" provided by a single
243 ATM datalink cloud through ATM signaling.
244
245 Example network configurations based on CSRs are shown in figure 1.
246 An ATM datalink network may be a large cloud which accommodates
247 multiple IP subnets X, Y and Z. Or several distinct ATM datalinks
248 may accommodate single IP subnet X, Y and Z respectively. The latter
249 configuration would be straightforward in discussing the CSR, but the
250 CSR is also applicable to the former configuration as well. In
251 addition, the CSR would be applicable as a router which interconnects
252 multiple NHRP-based ATM clouds.
253
254 Two different kinds of ATM VCs are defined between adjacent CSRs or
255 between CSR and ATM-attached hosts/routers.
256
257 1) Default-VC
258
259 It is a general purpose VC used by any communications which select
260 conventional hop-by-hop IP routed paths. All incoming cells received
261 from this VC are assembled to IP datagrams and handled based on their
262 IP headers. VCs set up in the Classical IP Model are classified into
263 this category.
264
265 2) Dedicated-VC
266
267 It is used by specific communications (IP flows) which are specified
268 by, for example, any combination of the destination IP address/port,
269 the source IP address/port or IPv6 flow label. It can be
270 concatenated with other Dedicated-VCs which accommodate the same IP
271 flow as it, and can constitute an ATM Bypass-pipe for those IP flows.
272
273
274
275
276 <span class="grey">Katsube, et. al. Informational [Page 5]</span></pre>
277 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-6" ></span>
278 <span class="grey"><a href="./rfc2098">RFC 2098</a> Toshiba's Router Extension for ATM February 1997</span>
279
280
281 Ingress/egress nodes of the Bypass-pipe can be either CSRs or ATM-
282 attached routers/hosts both of which speak a Bypass-pipe control
283 protocol. (we call that "Bypass-capable nodes") On the other hand,
284 intermediate nodes of the Bypass-pipe should be CSRs since they need
285 to have cell switching capabilities as well as to speak the Bypass-
286 pipe control protocol.
287
288 The route for a Bypass-pipe follows IP routing information in each
289 CSR. In figure 1, IP datagrams from a source host or router X.1 to a
290 destination host or router Z.1 are transferred over the route X.1 -&gt;
291 CSR1 -&gt; CSR2 -&gt; Z.1 regardless of whether the communication is on a
292 hop-by-hop basis or Bypass-pipe basis. Routes for individual
293 Dedicated-VCs which constitutes the Bypass-pipe X.1 --&gt; Z.1 (X.1 -&gt;
294 CSR1, CSR1 -&gt; CSR2, CSR2 -&gt; Z.1) would be determined based on ATM
295 routing protocols such as PNNI [<a href="#ref-PNNI1.0" title="&quot;P-NNI Specification Version 1.0&quot;">PNNI1.0</a>], and would be independent of
296 IP level routing.
297
298 An example of IP datagram transmission mechanism is as follows.
299
300 o The host/router X.1 checks an identifier of each IP datagram,
301 which may be the "destination IP address (prefix)",
302 "source/destination IP address (prefix) pair", "destination IP
303 address and port", "source IP address and Flow label (in IPv6)",
304 and so on. Based on either of those identifiers, it determines
305 over which VC the datagram should be transmitted.
306
307 o The CSR1/2 checks the VPI/VCI value of each incoming cell. When
308 the mapping from the incoming interface/VPI/VCI to outgoing
309 interface/VPI/VCI is found in an ATM routing table, it is directly
310 forwarded to the specified interface through an ATM switch module.
311 When the mapping in not found in the ATM routing table (or the
312 table shows an IP module as an output interface), the cell is
313 assembled to an IP datagram and then forwarded to an appropriate
314 outgoing interface/VPI/VCI based on an identifier of the datagram.
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332 <span class="grey">Katsube, et. al. Informational [Page 6]</span></pre>
333 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-7" ></span>
334 <span class="grey"><a href="./rfc2098">RFC 2098</a> Toshiba's Router Extension for ATM February 1997</span>
335
336
337 IP subnet X IP subnet Y IP subnet Z
338 &lt;---------------------&gt; &lt;-----------------&gt; &lt;---------------------&gt;
339
340 +-------+ Default +-------+ Default +-------+ Default +-------+
341 | | -VC | CSR 1 | -VC | CSR 2 | -VC | |
342 | Host +=============+ +===============+ +=============+ Host |
343 | X.1 +-------------+++++---------------+++++-------------+ Z.1 |
344 | +-------------+++++---------------+++++-------------+ |
345 | +-------------+++++---------------+++++-------------+ |
346 | |Dedicated | | Dedicated | |Dedicated | |
347 +-------+ -VCs +-------+ -VCs +-------+ -VCs +-------+
348 &lt;---------------------------------------------------&gt;
349 Bypass-pipe
350
351
352 Figure 1 Internetworking Architecture based on CSR
353
354 <span class="h3"><a class="selflink" id="section-3.2" href="#section-3.2">3.2</a> Features</span>
355
356 The main feature of the CSR-based internetworking architecture is the
357 same as that of the NHRP-based architecture in the sense that they
358 both provide direct ATM level connectivity beyond IP subnet
359 boundaries. There are, however, several notable differences in the
360 CSR-based architecture compared with the NHRP-based one as follows.
361
362 1) Relationship between IP routing and ATM routing
363
364 In the NHRP model, an egress point of the ATM network is first
365 determined in the next hop resolution phase based on IP level routing
366 information. Then the actual route for an ATM-VC to the obtained
367 egress point is determined in the ATM connection setup phase based on
368 ATM level routing information. Both kinds of routing information
369 would be calculated according to factors such as network topology and
370 available bandwidth for the large ATM cloud. The ATM routing will be
371 based on PNNI phase1 [<a href="#ref-PNNI1.0" title="&quot;P-NNI Specification Version 1.0&quot;">PNNI1.0</a>] while the IP routing will be based on
372 OSPF, BGP, IS-IS, etc. We need to manage two different routing
373 protocols over the large ATM cloud until Integtrated-PNNI [<a href="#ref-IPNNI96" title="&quot;Issues and Approaches for Integrated PNNI&quot;">IPNNI96</a>]
374 which takes both ATM level metric and IP level metric into account
375 will be phased in in the future.
376
377 In the CSR model, IP level routing determines an egress point of the
378 ATM cloud as well as determines inter-subnet level path to the point
379 that shows which CSRs it should pass through. ATM level routing
380 determines an intra-subnet level path for ATM-VCs (both Dedicated-VC
381 and Default-VC) only between adjacent nodes (CSRs or ATM-attached
382 hosts/routers). Since the roles of routing are hierarchically
383 subdivided into inter-subnet level (router level) and intra-subnet
384 level (ATM SW level), ATM routing does not have to operate all over
385
386
387
388 <span class="grey">Katsube, et. al. Informational [Page 7]</span></pre>
389 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-8" ></span>
390 <span class="grey"><a href="./rfc2098">RFC 2098</a> Toshiba's Router Extension for ATM February 1997</span>
391
392
393 the ATM cloud but only in individual IP subnets independent from each
394 other. This will decrease the amount of information for ATM routing
395 protocol handling. But an end-to-end ATM path may not be optimal
396 compared with the NHRP model since the path should go through routers
397 at subnet boundaries in the CSR model.
398
399 2) Dynamic routing and redundancy support
400
401 A CSR-based network can dynamically change routes for Bypass-pipes
402 when related IP level routing information changes. Bypass-pipes
403 related to the routing changes do not have to be torn down nor
404 established from scratch since intermediate CSRs related to IP
405 routing changes can follow them and change routes for related
406 Bypass-pipes by themselves.
407
408 The same things apply when some error or outage happens in any ATM
409 nodes/links/routers on the route of a Bypass-pipe. CSRs that have
410 noticed such errors or outages would change routes for related
411 Bypass-pipes by themselves.
412
413 3) Interoperability with IP level resource reservation protocols in
414 multicast environments
415
416 As current NHRP specification assumes application of NHRP to unicast
417 environments only, multicast IP flows should still be carried based
418 on a hop-by-hop manner with multicast routers. In addition,
419 realization of IP level resource reservation protocols such as RSVP
420 over NHRP environments requires further investigation.
421
422 The CSR-based internetworking architecture which keeps subnet-by-
423 subnet internetworking with regard to any control protocol sequence
424 can provide multicast Bypass-pipes without requiring any
425 modifications in IP multicast over ATM [<a href="#ref-IPMC96" title="&quot;Support for Multicast over UNI 3.0/3.1 based ATM Networks&quot;">IPMC96</a>] or multicast routing
426 techniques. In addition, since the CSR can handle RSVP messages
427 which are transmitted in a hop-by-hop manner, it can provide Bypass-
428 pipes which satisfy QoS requirements by the cooperation of the RSVP
429 and the Bypass-pipe control protocol.
430
431 <span class="h2"><a class="selflink" id="section-4" href="#section-4">4</a>. Control Architecture for CSR</span>
432
433 Several issues with regard to a control architecture for the CSR are
434 discussed in this section.
435
436 <span class="h3"><a class="selflink" id="section-4.1" href="#section-4.1">4.1</a> Network Reference Model</span>
437
438 In order to help understanding discussions in this section, the
439 following network reference model is assumed. Source hosts S1, S2,
440 and destination hosts D1, D2 are attached to Ethernets, while S3 and
441
442
443
444 <span class="grey">Katsube, et. al. Informational [Page 8]</span></pre>
445 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-9" ></span>
446 <span class="grey"><a href="./rfc2098">RFC 2098</a> Toshiba's Router Extension for ATM February 1997</span>
447
448
449 D3 are attached to the ATM. Routers R1 and R5 are attached to
450 Ethernets only, while R2, R3 and R4 are attached to the ATM. The ATM
451 datalink for subnet #3 and subnet #4 can either be physically
452 separated datalinks or be the same datalink. In other words, R3 can
453 be either one-port or multi-port router.
454
455 Ether Ether ATM ATM Ether Ether
456 | | +-----+ +-----+ | |
457 | | | | | | | |
458 S1--| S2---| S3---| | | |---D3 |---D2 |--D1
459 | | | | | | | |
460 |---R1---|---R2---| |--R3--| |---R4---|---R5---|
461 | | | | | | | |
462 | | +-----+ +-----+ | |
463 subnet subnet subnet subnet subnet subnet
464 #1 #2 #3 #4 #5 #6
465
466
467 Figure 2 Network Reference Model
468
469 Bypass-pipes can be configured [S3 or R2]--&gt;R3--&gt;[D3 or R4]. That
470 means that S3, D3, R2, R3 and R4 need to speak Bypass-pipe control
471 protocol, and means that R3 needs to be the CSR. We use term
472 "Bypass-capable nodes" for hosts/routers which can speak Bypass-pipe
473 control protocol but are not necessarily CSRs.
474
475 As shown in this reference model, Bypass-pipe can be configured from
476 host to host (S3--&gt;R3--&gt;D3), router to host (R2--&gt;R3--&gt;D3), host to
477 router (S3--&gt;R3--&gt;R4), and router to router (R2--&gt;R3--&gt;R4).
478
479 <span class="h3"><a class="selflink" id="section-4.2" href="#section-4.2">4.2</a> Possible Use of Bypass-pipe</span>
480
481 Possible use (or purposes) of Bypass-pipe provided by CSRs, in other
482 words, possible triggers that initiate Bypass-pipe setup procedure,
483 is discussed in this subsection.
484
485 Following two purposes for Bypass-pipe setup are assumed at present;
486
487 a) Provision of low latency path
488
489 This indicates cases in which end hosts or routers initiate a
490 Bypass-pipe setup procedure when they will transmit large amount of
491 datagrams toward a specific destination. For instance,
492
493 - End hosts or routers initiate Bypass-pipe setup procedures based
494 on the measurement of IP datagrams transmitted toward a certain
495 destination.
496
497
498
499
500 <span class="grey">Katsube, et. al. Informational [Page 9]</span></pre>
501 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-10" ></span>
502 <span class="grey"><a href="./rfc2098">RFC 2098</a> Toshiba's Router Extension for ATM February 1997</span>
503
504
505 - End hosts or routers initiate Bypass-pipe setup procedures when
506 it detects datagrams with certain higher layer protocols such as
507 ftp, nntp, http, etc.
508
509 Other triggers may be possible depending on the policy in each
510 network. In any case, the purpose of Bypass-pipe setup in each of
511 these cases is to reduce IP processing burden at intermediate routers
512 as well as to provide a communication path with low latency for burst
513 data transfer, rather than to provide end host applications with
514 specific bandwidth/QoS.
515
516 There would be no rule for determining bandwidth for such kinds of
517 Bypass-pipes since no explicit information about bandwidth/QoS
518 requirement by end hosts is available without IP-level resource
519 reservation protocols such as RSVP. Using UBR VCs as components of
520 the Bypass-pipe would be the easiest choice although there is no
521 guarantees for cell loss quality, while using other services such as
522 CBR/VBR/ABR with an adequate parameter tuning would be possible.
523
524 b) Provision of specific bandwidth/QoS requested by hosts
525
526 This indicates cases in which routers or end hosts initiate a
527 Bypass-pipe setup procedure by triggers related to IP-level
528 bandwidth/QoS request from end hosts. The "resource management
529 entity" in the host or router, which has received bandwidth/QoS
530 requests from applications or adjacent nodes may choose to
531 accommodate the requested IP flow to an existing VC or choose to
532 allocate a new Dedicated-VC for the requested IP flow. Selecting the
533 latter choice at each router can correspond to the trigger for
534 constituting a Bypass-pipe. When both an incoming VC and an outgoing
535 VC (or VCs) are dedicated to the same IP flow(s), those VCs can be
536 concatenated at the CSR (ATM cut-through) to constitute a Bypass-
537 pipe. Bandwidth for the Bypass-pipe (namely, individual VCs
538 constituting the Bypass-pipe) in this case would be determined based
539 on the bandwidth/QoS requirements by the end host which is conveyed
540 by, e.g., RSVP messages. The ATM service classes; e.g., CBR/VBR/ABR;
541 that would be selected depends on the IP-level service classes
542 requested by the end hosts.
543
544 Bypass-pipe provision for the purpose of b) will surely be beneficial
545 in the near future when related IP-level resource reservation
546 protocol will become available as well as when definitions of
547 individual service classes and flow specs offered to applications
548 become clear. On the other hand, Bypass-pipe setup for the purpose
549 of a) may be beneficial right now since it does not require
550 availability of IP-level resource reservation protocols. In that
551 sense, a) can be regarded as a kind of short-term use while b) is a
552 long-term use.
553
554
555
556 <span class="grey">Katsube, et. al. Informational [Page 10]</span></pre>
557 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-11" ></span>
558 <span class="grey"><a href="./rfc2098">RFC 2098</a> Toshiba's Router Extension for ATM February 1997</span>
559
560
561 <span class="h3"><a class="selflink" id="section-4.3" href="#section-4.3">4.3</a> Variations of Bypass-pipe Control Architecture</span>
562
563 A number of variations regarding Bypass-pipe control architecture are
564 introduced. Items which are related to architectural variations are;
565
566 o Ways of providing Dedicated-VCs
567
568 o Channels for Bypass-pipe control message transfer
569
570 o Bypass-pipe control procedures
571
572 Each of these items are discussed below.
573
574 <span class="h4"><a class="selflink" id="section-4.3.1" href="#section-4.3.1">4.3.1</a> Ways of Providing Dedicated-VCs</span>
575
576 There are roughly three alternatives regarding the way of providing
577 Dedicated-VCs in individual IP subnets as components of a Bypass-
578 pipe.
579
580 a) On-demand SVC setup
581
582 Dedicated-VCs are set up in individual IP subnets each time you want
583 to set up a Bypass-pipe through the ATM signaling procedure.
584
585 b) Picking up one from a bunch of (semi-)PVCs
586
587 Several VCs are set up beforehand between CSR and CSR, or CSR and
588 other ATM-attached nodes (hosts/router) in each IP subnet. Unused VC
589 is picked up as a Dedicated-VC from these PVCs in each IP subnet when
590 a Bypass-pipe is set up.
591
592 c) Picking up one VCI in PVP/SVP
593
594 PVPs or SVPs are set up between CSR and CSR, or CSR and other ATM-
595 attached nodes (hosts/routers) in each IP subnet. PVPs would be set
596 up as a router/host initialization procedure, while SVPs, on the
597 other hand, would be set up through ATM signaling when the first VC
598 (either Default- or Dedicated-) setup request is initiated by either
599 of some peer nodes. Then, Unused VCI value is picked up as a
600 Dedicated-VC in the PVP/SVP in each IP subnet when a Bypass-pipe is
601 set up. The SVP can be released through ATM signaling when no VCI
602 value is in active state.
603
604 The best choice will be a) with regard to efficient network resource
605 usage. However, you may go through three steps, ATMARP (for unicast
606 [<a href="./rfc1577" title="&quot;Classical IP and ARP over ATM&quot;">RFC1577</a>] or multicast [<a href="#ref-IPMC96" title="&quot;Support for Multicast over UNI 3.0/3.1 based ATM Networks&quot;">IPMC96</a>] in each IP subnet), SVC setup (in
607 each IP subnet) and exchange of Bypass-pipe control message in this
608 case. Whether a) is practical choice or not will depend on whether
609
610
611
612 <span class="grey">Katsube, et. al. Informational [Page 11]</span></pre>
613 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-12" ></span>
614 <span class="grey"><a href="./rfc2098">RFC 2098</a> Toshiba's Router Extension for ATM February 1997</span>
615
616
617 you can allow larger Bypass-pipe setup time due to three-step
618 procedure mentioned above, or whether you can send datagrams over
619 Default-VCs in a hop-by-hop manner while waiting for the Bypass-pipe
620 set up.
621
622 In the case of b) or c), the issue of Bypass-pipe setup time will be
623 improved since SVC setup step can be skipped. In b), each node (CSR
624 or ATM-attached host/router) should specify some traffic descriptors
625 even for unused VCs, and the ATM datalink should reserve its desired
626 resource (such as VCI value and bandwidth) for them. In addition,
627 the ATM datalink may have to carry out UPC functions for those unused
628 VCs. Such burden would be reduced when you use UBR-PVCs and set peak
629 cell rate for each of them equal to link rate, but bandwidth/QoS for
630 the Bypass-pipe is not provided in this case. In c), on the other
631 hand, traffic descriptors which should be specified by each node for
632 the ATM datalink is not each VC's but VP's only. Resource
633 reservations for individual VCs will be carried out not as a
634 functionality of the ATM datalink but of each CSR or ATM-attached
635 host/router if necessary. A functionality which need to be provided
636 by the ATM datalink is control of VPs' bandwidth only such as UPC and
637 dynamic bandwidth negotiation if it would be widely available.
638
639 <span class="h4"><a class="selflink" id="section-4.3.2" href="#section-4.3.2">4.3.2</a> Channels for Bypass-pipe Control Message Transfer</span>
640
641 There are several alternatives regarding the channels for managing
642 (setting up, releasing, and possibly changing the route of) a
643 Bypass-pipe. This subsection explains these alternatives and
644 discusses their properties.
645
646 Three alternatives are discussed, Inband control message, Outband
647 control message, and use of ATM signaling.
648
649 i) Inband Control Message
650
651 When setting up a Bypass-pipe, control messages are transmitted over
652 a Dedicated-VC which will eventually be used as a component of the
653 Bypass-pipe. These messages are handled at each CSR, and similar
654 messages are transmitted to the next-hop node over a Dedicated-VC
655 along the selected route (based on IP routing table). Unlike outband
656 message protocol described in ii), each message does not have to
657 indicate a Dedicated-VC which will be used since the message itself
658 is carried over "that" VC.
659
660 The inband control message can be either "datagram dedicated for
661 Bypass-pipe control" or "actual IP datagram" sent by user
662 application. Actual IP datagrams can be transmitted over Bypass-pipe
663 after it has been set up in the former case. In the latter case, on
664 the other hand, the first (or several) IP datagram(s) received from
665
666
667
668 <span class="grey">Katsube, et. al. Informational [Page 12]</span></pre>
669 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-13" ></span>
670 <span class="grey"><a href="./rfc2098">RFC 2098</a> Toshiba's Router Extension for ATM February 1997</span>
671
672
673 an unused Dedicated-VC are analyzed at IP level and transmitted
674 toward adequate next hop over an unused Dedicated-VC. Then incoming
675 Dedicated-VC and outgoing Dedicated-VC are concatenated to construct
676 a Bypass-pipe.
677
678 In inband control, Bypass-pipe control messages transmitted after a
679 Bypass-pipe has been set up cannot be identified at intermediate CSRs
680 since those messages are forwarded at cell level there. As a
681 possible solution for this issue, intermediate CSRs can identify
682 Bypass-pipe control messages by marking cell headers, e.g., PTI bit
683 which indicates F5 OAM cell. With regard to Bypass-pipe release,
684 explicit release message may not be necessary if individual CSRs
685 administer the amount of traffic over each Dedicated-VC and deletes
686 concatenation information for an inactive Bypass-pipe with their own
687 decision.
688
689 ii) Outband Control Message
690
691 When a Bypass-pipe is set up or released, control messages are
692 transmitted over VCs which are different from Dedicated-VCs used as
693 components of the Bypass-pipe. Unlike inband message protocol
694 described in i), each message has to indicate which Dedicated-VCs the
695 message would like to control. Therefore, an identifier that
696 uniquely discriminates a VC, which is not a VPI/VCI that is not
697 identical at both endpoints of the VC, need to be defined and be
698 given at VC initiation phase. However, an issue of control message
699 transmission after a Bypass-pipe has been set up in inband case does
700 not exist.
701
702 Four alternatives are possible regarding how to convey Bypass-pipe
703 control messages hop-by-hop over ATM datalink networks.
704
705 1) Defines VC for Bypass-pipe control messages only.
706
707 2) Uses Default-VC and discriminates Bypass-pipe control messages
708 from user datagrams by an LLC/SANP value in <a href="./rfc1483">RFC1483</a> encapsulation.
709
710 3) Uses Default-VC and discriminates Bypass-pipe control messages
711 from user datagrams by a protocol field value in IP header.
712
713 4) Uses Default-VC and discriminates Bypass-pipe control messages
714 from user datagrams by a port ID in the UDP frame.
715
716 When we take into account interoperability with Bypass-incapable
717 routers, 1) will not be a good choice. Whether we select 2) or 3) 4)
718 depends on whether we should consider multiprotocol rather than IP
719 only.
720
721
722
723
724 <span class="grey">Katsube, et. al. Informational [Page 13]</span></pre>
725 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-14" ></span>
726 <span class="grey"><a href="./rfc2098">RFC 2098</a> Toshiba's Router Extension for ATM February 1997</span>
727
728
729 In the case of IP multicast, point-to-multipoint VCs in individual
730 subnets are concatenated at CSRs consecutively in order to constitute
731 end-to-end multicast tree. Above four alternatives may require the
732 same number of point-to-multipoint Defalut-VCs as the number of
733 requested point-to-multipoint Dedicated-VCs in multicast case. The
734 fifth alternative which can reduce the necessary number of VCs to
735 convey control messages in a multicast environment is;
736
737 5) Defines point-to-multipoint VC whose leaves are members of
738 multicast group 224.0.0.1. All nodes which are members of at
739 least one of active multicast group would become leaves of this
740 point-to-multipoint VC.
741
742 Each upstream node may become a root of the point-to-multipoint VC,
743 or a sort of multicast server to which each upstream node transmits
744 cells over a point-to-point VC may become a root of that. In any
745 case, Bypass-pipe control messages for every multicast group are
746 transmitted to all nodes which are members of either of the group.
747 When a downstream node has received control messages which are not
748 related to a multicast group it belongs, it should discard them by
749 referring to a destination group address on their IP header.
750 Donwstream node would still need to use point-to-point VC to send
751 control messages toward upstream.
752
753 iii) Use of ATM Signaling Message
754
755 Supposing that ATM signaling messages can convey IP addresses (and
756 possibly port IDs) of source and destination, it may be possible that
757 ATM signaling messages be used as Bypass-pipe control messages also.
758 In that case, an ATM connection setup message indicates a setup of a
759 Dedicated-VC to an ATM address of a desirable next-hop IP node, and
760 also indicates a setup of a Bypass-pipe to an IP address (and
761 possibly port ID) of a target destination node. Information elements
762 for the Dedicated-VC setup (ATM address of a next-hop node,
763 bandwidth, QoS, etc.) are handled at ATM nodes, while information
764 elements for the Bypass-pipe setup (source and destination IP
765 addresses, possibly their port IDs, or flow label for IPv6, etc.) are
766 transparently transferred to the next-hop IP node. The next-hop IP
767 node accepts Dedicated-VC setup and handles such IP level information
768 elements.
769
770 ATM signaling messages can be transferred from receiver to sender as
771 well as sender to receiver when you set zero Forward Cell Rate and
772 non-zero Backward Cell Rate as an ATM traffic descriptor information
773 element in unicast case, or when Leaf Initiated Join capabilities
774 will become available in multicast case.
775
776
777
778
779
780 <span class="grey">Katsube, et. al. Informational [Page 14]</span></pre>
781 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-15" ></span>
782 <span class="grey"><a href="./rfc2098">RFC 2098</a> Toshiba's Router Extension for ATM February 1997</span>
783
784
785 Issues in this method are,
786
787 - Information elements which specify IP level (and port level)
788 information need to be defined, e.g., B-HLI or B-UUI, as an ATM
789 signaling specification.
790
791 - It would be difficult to support soft-state Bypass-pipe control
792 which transmits control messages periodically since ATM signaling
793 is a hard-state protocol.
794
795 <span class="h4"><a class="selflink" id="section-4.3.3" href="#section-4.3.3">4.3.3</a> Bypass-pipe Control Procedures</span>
796
797 This subsection discusses several items with regard to actual
798 procedures for Bypass-pipe control.
799
800 a) Distributed trigger vs. Centralized (restricted) trigger
801
802 The first item to be discussed is whether the functionality of
803 detecting a trigger of Dedicated-VC/Bypass-pipe control is
804 distributed to all the nodes (including CSRs and hosts/edge devices)
805 or restricted to specific nodes.
806
807 In the case of the distributed trigger, every node is regarded as
808 having a capability of detecting a trigger of Bypass-pipe setup or
809 termination. For example, every node detects datagrams for ftp, and
810 sets up (or fetches) a Dedicated-VC individually to construct a
811 Bypass-pipe. After setting up or fetching the Dedicated-VCs,
812 messages which informs (or requests) the transmission of the IP flow
813 over the Dedicated-VC are exchanged between adjacent nodes. That
814 enables peer nodes to share the same knowledge about the mapping
815 relationship between the IP flow and the Dedicated-VC. There is no
816 end-to-end message transmission in the Bypass-pipe control procedure
817 itself, but transmission between adjacent nodes only.
818
819 In the case of the centralized (or restricted) trigger, capability of
820 detecting a trigger of Bypass-pipe setup or termination is restricted
821 to nodes which are located at "the boundary of the CSR-cloud". The
822 boundary of the CSR-cloud signifies, for individual IP flows, the
823 node which is the first-hop or the last-hop CSR-capable node. For
824 example, a node which detects datagrams for ftp can initiate Bypass-
825 pipe setup procedure only when its previous hop is non-ATM or CSR-
826 incapable. In this case, Bypass-pipe control messages are originated
827 at the boundary of the CSR-cloud, and forwarded hop-by-hop toward
828 another side of the boundary, which is similar to ATM signaling
829 messages. The semantics of the messages may be the request of end-
830 to-end Bypass-pipe setup as well as notification or request of
831 mapping relationship between the IP flow and the Dedicated-VC.
832
833
834
835
836 <span class="grey">Katsube, et. al. Informational [Page 15]</span></pre>
837 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-16" ></span>
838 <span class="grey"><a href="./rfc2098">RFC 2098</a> Toshiba's Router Extension for ATM February 1997</span>
839
840
841 b) Upstream-initiated control vs. Downstream-initiated control
842
843 The second item to be discussed is whether the setup of a Dedicated-
844 VC and the control procedure for constructing a Bypass-pipe are
845 initiated by upstream side or downstream side.
846
847 In the case of the upstream-initiated control, the upstream node
848 takes the initiative when setting up a Dedicated-VC for a specific IP
849 flow and creating the mapping relationship between the IP flow and
850 the Dedicated-VC. For example, a CSR which detects datagrams for ftp
851 sets up (or fetches) a Dedicated-VC toward its downstream neighbor
852 and notifies its downstream neighbor that it will transmit a specific
853 IP flow over the Dedicated-VC. This means that the downstream node
854 is requested to receive datagrams from the Dedicated-VC.
855
856 In the case of the downstream-initiated control, the downstream node
857 takes the initiative when setting up a Dedicated-VC for a specific IP
858 flow and creating the mapping relationship between the IP flow and
859 the Dedicated-VC. For example, a CSR which detects datagrams for ftp
860 sets up (or fetches) a Dedicated-VC toward its upstream neighbor and
861 requests its upstream neighbor to transmit a specific IP flow over
862 the Dedicated-VC. This means that the upstream node is requested to
863 transmit the IP flow over the Dedicated-VC.
864
865 c) Hard-state management vs. Soft-state management
866
867 The third item to be discussed is whether the control (setup,
868 maintain, and release) of the Bypass-pipe is based on hard-state or
869 soft-state.
870
871 In hard-state management, individual nodes transmit Bypass-pipe
872 control messages only when they want to notify or request any change
873 in their neighbors' state. They should wait for an acknowledgement
874 of the message before they change their internal state. For example,
875 after setting up a Bypass-pipe, it is maintained until either of a
876 peer nodes transmits a message to release the Bypass-pipe.
877
878 In soft-state management, individual nodes periodically transmit
879 Bypass-pipe control messages in order to maintain their neighbors'
880 state. They do not have to wait for an acknowledgement of the
881 message before they changes its internal state. For example, even
882 after setting up a Bypass-pipe, either of a peer nodes is required to
883 periodically transmit refresh messages to its neighbor in order to
884 maintain the Bypass-pipe.
885
886 <span class="h2"><a class="selflink" id="section-5" href="#section-5">5</a>. Security Considerations</span>
887
888 Security issues are not discussed in this memo.
889
890
891
892 <span class="grey">Katsube, et. al. Informational [Page 16]</span></pre>
893 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-17" ></span>
894 <span class="grey"><a href="./rfc2098">RFC 2098</a> Toshiba's Router Extension for ATM February 1997</span>
895
896
897 <span class="h2"><a class="selflink" id="section-6" href="#section-6">6</a>. Summary</span>
898
899 Basic concept of Cell Switch Router (CSR) are clarified and control
900 architecture for CSR is discussed. A number of methods to control
901 Bypass-pipe will be possible each of which has its own advantages and
902 disadvantages. Further investigation and discussion will be
903 necessary to design control protocol which may depend on the
904 requirements by users.
905
906 <span class="h2"><a class="selflink" id="section-7" href="#section-7">7</a>. References</span>
907
908 [<a id="ref-IPMC96">IPMC96</a>] Armitage, G., "Support for Multicast over UNI 3.0/3.1 based
909 ATM Networks", <a href="./rfc2022">RFC 2022</a>, November 1996.
910
911 [<a id="ref-ATM3.1">ATM3.1</a>] The ATM-Forum, "ATM User-Network Interface Specification,
912 v.3.1", Sept. 1994.
913
914 [<a id="ref-RSVP13">RSVP13</a>] Braden, R., et al., "Resource ReSerVation Protocol (RSVP),
915 Version 1 Functional Specification", Work in Progress.
916
917 [<a id="ref-IPNNI96">IPNNI96</a>] R. Callon, et al., "Issues and Approaches for Integrated
918 PNNI", The ATM Forum Contribution No. 96-0355, April 1996.
919
920 [<a id="ref-NHRP09">NHRP09</a>] Luciani, J., et al., "NBMA Next Hop Resolution Protocol
921 (NHRP)", Work in Progress.
922
923 [<a id="ref-PNNI1.0">PNNI1.0</a>] The ATM-Forum, "P-NNI Specification Version 1.0", March
924 1996.
925
926 [<a id="ref-RFC1483">RFC1483</a>] Heinanen, J., "Multiprotocol Encapsulation over ATM
927 Adaptation Layer 5", <a href="./rfc1483">RFC 1483</a>, July 1993.
928
929 [<a id="ref-RFC1577">RFC1577</a>] Laubach, M., "Classical IP and ARP over ATM", <a href="./rfc1577">RFC 1577</a>,
930 October 1993.
931
932 [<a id="ref-RFC1819">RFC1819</a>] Delgrossi, L, and L. Berger, "Internet STream Protocol
933 Version 2 (STII) Protocol Specification Version ST2+", <a href="./rfc1819">RFC 1819</a>,
934 August 1995.
935
936
937
938
939
940
941
942
943
944
945
946
947
948 <span class="grey">Katsube, et. al. Informational [Page 17]</span></pre>
949 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-18" ></span>
950 <span class="grey"><a href="./rfc2098">RFC 2098</a> Toshiba's Router Extension for ATM February 1997</span>
951
952
953 <span class="h2"><a class="selflink" id="section-8" href="#section-8">8</a>. Authors' Addresses</span>
954
955 Yasuhiro Katsube
956 R&amp;D Center, Toshiba
957 1 Komukai Toshiba-cho, Saiwai-ku, Kawasaki 210
958 Japan
959 Phone : +81-44-549-2238
960 EMail : [email protected]
961
962 Ken-ichi Nagami
963 R&amp;D Center, Toshiba
964 1 Komukai Toshiba-cho, Saiwai-ku, Kawasaki 210
965 Japan
966 Phone : +81-44-549-2238
967 EMail : [email protected]
968
969 Hiroshi Esaki
970 R&amp;D Center, Toshiba
971 1 Komukai Toshiba-cho, Saiwai-ku, Kawasaki 210
972 Japan
973 Phone : +81-44-549-2238
974 EMail : [email protected]
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004 Katsube, et. al. Informational [Page 18]
1005 </pre>