Debian Codegraph @

rfc2678.html

raw
1 <pre>Network Working Group J. Mahdavi
2 Request for Comments: 2678 Pittsburgh Supercomputing Center
3 Obsoletes: <a href="./rfc2498">2498</a> V. Paxson
4 Category: Standards Track Lawrence Berkeley National Laboratory
5 September 1999
6
7
8 <span class="h1">IPPM Metrics for Measuring Connectivity</span>
9
10 Status of this Memo
11
12 This document specifies an Internet standards track protocol for the
13 Internet community, and requests discussion and suggestions for
14 improvements. Please refer to the current edition of the "Internet
15 Official Protocol Standards" (STD 1) for the standardization state
16 and status of this protocol. Distribution of this memo is unlimited.
17
18 Copyright Notice
19
20 Copyright (C) The Internet Society (1999). All Rights Reserved.
21
22 <span class="h2"><a class="selflink" id="section-1" href="#section-1">1</a>. Introduction</span>
23
24 Connectivity is the basic stuff from which the Internet is made.
25 Therefore, metrics determining whether pairs of hosts (IP addresses)
26 can reach each other must form the base of a measurement suite. We
27 define several such metrics, some of which serve mainly as building
28 blocks for the others.
29
30 This memo defines a series of metrics for connectivity between a pair
31 of Internet hosts. It builds on notions introduced and discussed in
32 <a href="./rfc2330">RFC 2330</a>, the IPPM framework document. The reader is assumed to be
33 familiar with that document.
34
35 The structure of the memo is as follows:
36
37 + An analytic metric, called Type-P-Instantaneous-Unidirectional-
38 Connectivity, will be introduced to define one-way connectivity at
39 one moment in time.
40 + Using this metric, another analytic metric, called Type-P-
41 Instantaneous-Bidirectional-Connectivity, will be introduced to
42 define two-way connectivity at one moment in time.
43 + Using these metrics, corresponding one- and two-way analytic
44 metrics are defined for connectivity over an interval of time.
45
46
47
48
49
50
51
52 <span class="grey">Mahdavi &amp; Paxson 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="./rfc2678">RFC 2678</a> IPPM Metrics for Measuring Connectivity September 1999</span>
55
56
57 + Using these metrics, an analytic metric, called Type-P1-P2-
58 Interval-Temporal-Connectivity, will be introduced to define a
59 useful notion of two-way connectivity between two hosts over an
60 interval of time.
61 + Methodologies are then presented and discussed for estimating
62 Type-P1-P2-Interval-Temporal-Connectivity in a variety of
63 settings.
64
65 Careful definition of Type-P1-P2-Interval-Temporal-Connectivity and
66 the discussion of the metric and the methodologies for estimating it
67 are the two chief contributions of the memo.
68
69 <span class="h2"><a class="selflink" id="section-2" href="#section-2">2</a>. Instantaneous One-way Connectivity</span>
70
71 <span class="h3"><a class="selflink" id="section-2.1" href="#section-2.1">2.1</a>. Metric Name:</span>
72
73 Type-P-Instantaneous-Unidirectional-Connectivity
74
75 <span class="h3"><a class="selflink" id="section-2.2" href="#section-2.2">2.2</a>. Metric Parameters:</span>
76
77 + Src, the IP address of a host
78 + Dst, the IP address of a host
79 + T, a time
80
81 <span class="h3"><a class="selflink" id="section-2.3" href="#section-2.3">2.3</a>. Metric Units:</span>
82
83 Boolean.
84
85 <span class="h3"><a class="selflink" id="section-2.4" href="#section-2.4">2.4</a>. Definition:</span>
86
87 Src has *Type-P-Instantaneous-Unidirectional-Connectivity* to Dst at
88 time T if a type-P packet transmitted from Src to Dst at time T will
89 arrive at Dst.
90
91 <span class="h3"><a class="selflink" id="section-2.5" href="#section-2.5">2.5</a>. Discussion:</span>
92
93 For most applications (e.g., any TCP connection) bidirectional
94 connectivity is considerably more germane than unidirectional
95 connectivity, although unidirectional connectivity can be of interest
96 for some security applications (e.g., testing whether a firewall
97 correctly filters out a "ping of death"). Most applications also
98 require connectivity over an interval, while this metric is
99 instantaneous, though, again, for some security applications
100 instantaneous connectivity remains of interest. Finally, one might
101 not have instantaneous connectivity due to a transient event such as
102 a full queue at a router, even if at nearby instants in time one does
103 have connectivity. These points are addressed below, with this
104 metric serving as a building block.
105
106
107
108 <span class="grey">Mahdavi &amp; Paxson 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="./rfc2678">RFC 2678</a> IPPM Metrics for Measuring Connectivity September 1999</span>
111
112
113 Note also that we have not explicitly defined *when* the packet
114 arrives at Dst. The TTL field in IP packets is meant to limit IP
115 packet lifetimes to 255 seconds (<a href="./rfc791">RFC 791</a>). In practice the TTL field
116 can be strictly a hop count (<a href="./rfc1812">RFC 1812</a>), with most Internet hops being
117 much shorter than one second. This means that most packets will have
118 nowhere near the 255 second lifetime. In principle, however, it is
119 also possible that packets might survive longer than 255 seconds.
120 Consideration of packet lifetimes must be taken into account in
121 attempts to measure the value of this metric.
122
123 Finally, one might assume that unidirectional connectivity is
124 difficult to measure in the absence of connectivity in the reverse
125 direction. Consider, however, the possibility that a process on
126 Dst's host notes when it receives packets from Src and reports this
127 fact either using an external channel, or later in time when Dst does
128 have connectivity to Src. Such a methodology could reliably measure
129 the unidirectional connectivity defined in this metric.
130
131 <span class="h2"><a class="selflink" id="section-3" href="#section-3">3</a>. Instantaneous Two-way Connectivity</span>
132
133 <span class="h3"><a class="selflink" id="section-3.1" href="#section-3.1">3.1</a>. Metric Name:</span>
134
135 Type-P-Instantaneous-Bidirectional-Connectivity
136
137 <span class="h3"><a class="selflink" id="section-3.2" href="#section-3.2">3.2</a>. Metric Parameters:</span>
138
139 + A1, the IP address of a host
140 + A2, the IP address of a host
141 + T, a time
142
143 <span class="h3"><a class="selflink" id="section-3.3" href="#section-3.3">3.3</a>. Metric Units:</span>
144
145 Boolean.
146
147 <span class="h3"><a class="selflink" id="section-3.4" href="#section-3.4">3.4</a>. Definition:</span>
148
149 Addresses A1 and A2 have *Type-P-Instantaneous-Bidirectional-
150 Connectivity* at time T if address A1 has Type-P-Instantaneous-
151 Unidirectional-Connectivity to address A2 and address A2 has Type-P-
152 Instantaneous-Unidirectional-Connectivity to address A1.
153
154 <span class="h3"><a class="selflink" id="section-3.5" href="#section-3.5">3.5</a>. Discussion:</span>
155
156 An alternative definition would be that A1 and A2 are fully connected
157 if at time T address A1 has instantaneous connectivity to address A2,
158 and at time T+dT address A2 has instantaneous connectivity to A1,
159 where T+dT is when the packet sent from A1 arrives at A2. This
160 definition is more useful for measurement, because the measurement
161
162
163
164 <span class="grey">Mahdavi &amp; Paxson 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="./rfc2678">RFC 2678</a> IPPM Metrics for Measuring Connectivity September 1999</span>
167
168
169 can use a reply from A2 to A1 in order to assess full connectivity.
170 It is a more complex definition, however, because it breaks the
171 symmetry between A1 and A2, and requires a notion of quantifying how
172 long a particular packet from A1 takes to reach A2. We postpone
173 discussion of this distinction until the development of interval-
174 connectivity metrics below.
175
176 <span class="h2"><a class="selflink" id="section-4" href="#section-4">4</a>. One-way Connectivity</span>
177
178 <span class="h3"><a class="selflink" id="section-4.1" href="#section-4.1">4.1</a>. Metric Name:</span>
179
180 Type-P-Interval-Unidirectional-Connectivity
181
182 <span class="h3"><a class="selflink" id="section-4.2" href="#section-4.2">4.2</a>. Metric Parameters:</span>
183
184 + Src, the IP address of a host
185 + Dst, the IP address of a host
186 + T, a time
187 + dT, a duration
188 {Comment: Thus, the closed interval [T, T+dT] denotes a time
189 interval.}
190
191 <span class="h3"><a class="selflink" id="section-4.3" href="#section-4.3">4.3</a>. Metric Units:</span>
192
193 Boolean.
194
195 <span class="h3"><a class="selflink" id="section-4.4" href="#section-4.4">4.4</a>. Definition:</span>
196
197 Address Src has *Type-P-Interval-Unidirectional-Connectivity* to
198 address Dst during the interval [T, T+dT] if for some T' within [T,
199 T+dT] it has Type-P-instantaneous-connectivity to Dst.
200
201 <span class="h2"><a class="selflink" id="section-5" href="#section-5">5</a>. Two-way Connectivity</span>
202
203 <span class="h3"><a class="selflink" id="section-5.1" href="#section-5.1">5.1</a>. Metric Name:</span>
204
205 Type-P-Interval-Bidirectional-Connectivity
206
207 <span class="h3"><a class="selflink" id="section-5.2" href="#section-5.2">5.2</a>. Metric Parameters:</span>
208
209 + A1, the IP address of a host
210 + A2, the IP address of a host
211 + T, a time
212 + dT, a duration
213 {Comment: Thus, the closed interval [T, T+dT] denotes a time
214 interval.}
215
216
217
218
219
220 <span class="grey">Mahdavi &amp; Paxson 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="./rfc2678">RFC 2678</a> IPPM Metrics for Measuring Connectivity September 1999</span>
223
224
225 <span class="h3"><a class="selflink" id="section-5.3" href="#section-5.3">5.3</a>. Metric Units:</span>
226
227 Boolean.
228
229 <span class="h3"><a class="selflink" id="section-5.4" href="#section-5.4">5.4</a>. Definition:</span>
230
231 Addresses A1 and A2 have *Type-P-Interval-Bidirectional-Connectivity*
232 between them during the interval [T, T+dT] if address A1 has Type-P-
233 Interval-Unidirectional-Connectivity to address A2 during the
234 interval and address A2 has Type-P-Interval-Unidirectional-
235 Connectivity to address A1 during the interval.
236
237 <span class="h3"><a class="selflink" id="section-5.5" href="#section-5.5">5.5</a>. Discussion:</span>
238
239 This metric is not quite what's needed for defining "generally
240 useful" connectivity - that requires the notion that a packet sent
241 from A1 to A2 can elicit a response from A2 that will reach A1. With
242 this definition, it could be that A1 and A2 have full-connectivity
243 but only, for example, at time T1 early enough in the interval [T,
244 T+dT] that A1 and A2 cannot reply to packets sent by the other. This
245 deficiency motivates the next metric.
246
247 <span class="h2"><a class="selflink" id="section-6" href="#section-6">6</a>. Two-way Temporal Connectivity</span>
248
249 <span class="h3"><a class="selflink" id="section-6.1" href="#section-6.1">6.1</a>. Metric Name:</span>
250
251 Type-P1-P2-Interval-Temporal-Connectivity
252
253 <span class="h3"><a class="selflink" id="section-6.2" href="#section-6.2">6.2</a>. Metric Parameters:</span>
254
255 + Src, the IP address of a host
256 + Dst, the IP address of a host
257 + T, a time
258 + dT, a duration
259 {Comment: Thus, the closed interval [T, T+dT] denotes a time
260 interval.}
261
262 <span class="h3"><a class="selflink" id="section-6.3" href="#section-6.3">6.3</a>. Metric Units:</span>
263
264 Boolean.
265
266
267
268
269
270
271
272
273
274
275
276 <span class="grey">Mahdavi &amp; Paxson 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="./rfc2678">RFC 2678</a> IPPM Metrics for Measuring Connectivity September 1999</span>
279
280
281 <span class="h3"><a class="selflink" id="section-6.4" href="#section-6.4">6.4</a>. Definition:</span>
282
283 Address Src has *Type-P1-P2-Interval-Temporal-Connectivity* to
284 address Dst during the interval [T, T+dT] if there exist times T1 and
285 T2, and time intervals dT1 and dT2, such that:
286
287 + T1, T1+dT1, T2, T2+dT2 are all in [T, T+dT].
288 + T1+dT1 &lt;= T2.
289 + At time T1, Src has Type-P1 instantanous connectivity to Dst.
290 + At time T2, Dst has Type-P2 instantanous connectivity to Src.
291 + dT1 is the time taken for a Type-P1 packet sent by Src at time T1
292 to arrive at Dst.
293 + dT2 is the time taken for a Type-P2 packet sent by Dst at time T2
294 to arrive at Src.
295
296 <span class="h3"><a class="selflink" id="section-6.5" href="#section-6.5">6.5</a>. Discussion:</span>
297
298 This metric defines "generally useful" connectivity -- Src can send a
299 packet to Dst that elicits a response. Because many applications
300 utilize different types of packets for forward and reverse traffic,
301 it is possible (and likely) that the desired responses to a Type-P1
302 packet will be of a different type Type-P2. Therefore, in this
303 metric we allow for different types of packets in the forward and
304 reverse directions.
305
306 <span class="h3"><a class="selflink" id="section-6.6" href="#section-6.6">6.6</a>. Methodologies:</span>
307
308 Here we sketch a class of methodologies for estimating Type-P1-P2-
309 Interval-Temporal-Connectivity. It is a class rather than a single
310 methodology because the particulars will depend on the types P1 and
311 P2.
312
313 <span class="h4"><a class="selflink" id="section-6.6.1" href="#section-6.6.1">6.6.1</a>. Inputs:</span>
314
315 + Types P1 and P2, addresses A1 and A2, interval [T, T+dT].
316 + N, the number of packets to send as probes for determining
317 connectivity.
318 + W, the "waiting time", which bounds for how long it is useful to
319 wait for a reply to a packet.
320 Required: W &lt;= 255, dT &gt; W.
321
322 <span class="h4"><a class="selflink" id="section-6.6.2" href="#section-6.6.2">6.6.2</a>. Recommended values:</span>
323
324 dT = 60 seconds.
325 W = 10 seconds.
326 N = 20 packets.
327
328
329
330
331
332 <span class="grey">Mahdavi &amp; Paxson 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="./rfc2678">RFC 2678</a> IPPM Metrics for Measuring Connectivity September 1999</span>
335
336
337 <span class="h4"><a class="selflink" id="section-6.6.3" href="#section-6.6.3">6.6.3</a>. Algorithm:</span>
338
339 + Compute N *sending-times* that are randomly, uniformly distributed
340 over [T, T+dT-W].
341 + At each sending time, transmit from A1 a well-formed packet of
342 type P1 to A2.
343 + Inspect incoming network traffic to A1 to determine if a
344 successful reply is received. The particulars of doing so are
345 dependent on types P1 &amp; P2, discussed below. If any successful
346 reply is received, the value of the measurement is "true". At
347 this point, the measurement can terminate.
348 + If no successful replies are received by time T+dT, the value of
349 the measurement is "false".
350
351 <span class="h4"><a class="selflink" id="section-6.6.4" href="#section-6.6.4">6.6.4</a>. Discussion:</span>
352
353 The algorithm is inexact because it does not (and cannot) probe
354 temporal connectivity at every instant in time between [T, T+dT].
355 The value of N trades off measurement precision against network
356 measurement load. The state-of-the-art in Internet research does not
357 yet offer solid guidance for picking N. The values given above are
358 just guidelines.
359
360 <span class="h4"><a class="selflink" id="section-6.6.5" href="#section-6.6.5">6.6.5</a>. Specific methodology for TCP:</span>
361
362 A TCP-port-N1-port-N2 methodology sends TCP SYN packets with source
363 port N1 and dest port N2 at address A2. Network traffic incoming to
364 A1 is interpreted as follows:
365
366 + A SYN-ack packet from A2 to A1 with the proper acknowledgement
367 fields and ports indicates temporal connectivity. The measurement
368 terminates immediately with a value of "true". {Comment: if, as a
369 side effect of the methodology, a full TCP connection has been
370 established between A1 and A2 -- that is, if A1's TCP stack
371 acknowledges A2's SYN-ack packet, completing the three-way
372 handshake -- then the connection now established between A1 and A2
373 is best torn down using the usual FIN handshake, and not using a
374 RST packet, because RST packets are not reliably delivered. If
375 the three-way handshake is not completed, however, which will
376 occur if the measurement tool on A1 synthesizes its own initial
377 SYN packet rather than going through A1's TCP stack, then A1's TCP
378 stack will automatically terminate the connection in a reliable
379 fashion as A2 continues transmitting the SYN-ack in an attempt to
380 establish the connection. Finally, we note that using A1's TCP
381 stack to conduct the measurement complicates the methodology in
382 that the stack may retransmit the initial SYN packet, altering the
383 number of probe packets sent.}
384
385
386
387
388 <span class="grey">Mahdavi &amp; Paxson 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="./rfc2678">RFC 2678</a> IPPM Metrics for Measuring Connectivity September 1999</span>
391
392
393 + A RST packet from A2 to A1 with the proper ports indicates
394 temporal connectivity between the addresses (and a *lack* of
395 service connectivity for TCP-port-N1-port-N2 - something that
396 probably should be addressed with another metric).
397 + An ICMP port-unreachable from A2 to A1 indicates temporal
398 connectivity between the addresses (and again a *lack* of service
399 connectivity for TCP-port-N1-port-N2). {Comment: TCP
400 implementations generally do not need to send ICMP port-
401 unreachable messages because a separate mechanism is available
402 (sending a RST). However, <a href="./rfc1122">RFC 1122</a> states that a TCP receiving an
403 ICMP port-unreachable MUST treat it the same as the equivalent
404 transport-level mechanism (for TCP, a RST).}
405 + An ICMP host-unreachable or network-unreachable to A1 (not
406 necessarily from A2) with an enclosed IP header matching that sent
407 from A1 to A2 *suggests* a lack of temporal connectivity. If by
408 time T+dT no evidence of temporal connectivity has been gathered,
409 then the receipt of the ICMP can be used as additional information
410 to the measurement value of "false".
411
412 {Comment: Similar methodologies are needed for ICMP Echo, UDP, etc.}
413
414 <span class="h2"><a class="selflink" id="section-7" href="#section-7">7</a>. Acknowledgments</span>
415
416 The comments of Guy Almes, Martin Horneffer, Jeff Sedayao, and Sean
417 Shapira are appreciated.
418
419 <span class="h2"><a class="selflink" id="section-8" href="#section-8">8</a>. Security Considerations</span>
420
421 As noted in <a href="./rfc2330">RFC 2330</a>, active measurement techniques, such as those
422 defined in this document, can be abused for denial-of-service attacks
423 disguised as legitimate measurement activity. Furthermore, testing
424 for connectivity can be used to probe firewalls and other security
425 mechnisms for weak spots.
426
427 <span class="h2"><a class="selflink" id="section-9" href="#section-9">9</a>. References</span>
428
429 [<a id="ref-RFC1812">RFC1812</a>] Baker, F., "Requirements for IP Version 4 Routers", <a href="./rfc1812">RFC</a>
430 <a href="./rfc1812">1812</a>, June 1995.
431
432 [<a id="ref-RFC1122">RFC1122</a>] Braden, R., Editor, "Requirements for Internet Hosts --
433 Communication Layers", STD, 3, <a href="./rfc1122">RFC 1122</a>, October 1989.
434
435 [<a id="ref-RFC2330">RFC2330</a>] Paxson, V., Almes, G., Mahdavi, J. and M. Mathis,
436 "Framework for IP Performance Metrics", <a href="./rfc2330">RFC 2330</a>, May
437 1998.
438
439 [<a id="ref-RFC791">RFC791</a>] Postel, J., "Internet Protocol", STD 5, <a href="./rfc791">RFC 791</a>, September
440 1981.
441
442
443
444 <span class="grey">Mahdavi &amp; Paxson 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="./rfc2678">RFC 2678</a> IPPM Metrics for Measuring Connectivity September 1999</span>
447
448
449 <span class="h2"><a class="selflink" id="section-10" href="#section-10">10</a>. Authors' Addresses</span>
450
451 Jamshid Mahdavi
452 Pittsburgh Supercomputing Center
453 4400 5th Avenue
454 Pittsburgh, PA 15213
455 USA
456
457 EMail: [email protected]
458
459
460 Vern Paxson
461 MS 50A-3111
462 Lawrence Berkeley National Laboratory
463 University of California
464 Berkeley, CA 94720
465 USA
466
467 Phone: +1 510/486-7504
468 EMail: [email protected]
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500 <span class="grey">Mahdavi &amp; Paxson 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="./rfc2678">RFC 2678</a> IPPM Metrics for Measuring Connectivity September 1999</span>
503
504
505 <span class="h2"><a class="selflink" id="section-11" href="#section-11">11</a>. Full Copyright Statement</span>
506
507 Copyright (C) The Internet Society (1999). All Rights Reserved.
508
509 This document and translations of it may be copied and furnished to
510 others, and derivative works that comment on or otherwise explain it
511 or assist in its implementation may be prepared, copied, published
512 and distributed, in whole or in part, without restriction of any
513 kind, provided that the above copyright notice and this paragraph are
514 included on all such copies and derivative works. However, this
515 document itself may not be modified in any way, such as by removing
516 the copyright notice or references to the Internet Society or other
517 Internet organizations, except as needed for the purpose of
518 developing Internet standards in which case the procedures for
519 copyrights defined in the Internet Standards process must be
520 followed, or as required to translate it into languages other than
521 English.
522
523 The limited permissions granted above are perpetual and will not be
524 revoked by the Internet Society or its successors or assigns.
525
526 This document and the information contained herein is provided on an
527 "AS IS" basis and THE INTERNET SOCIETY AND THE INTERNET ENGINEERING
528 TASK FORCE DISCLAIMS ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING
529 BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION
530 HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF
531 MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.
532
533 Acknowledgement
534
535 Funding for the RFC Editor function is currently provided by the
536 Internet Society.
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556 Mahdavi &amp; Paxson Standards Track [Page 10]
557 </pre>