idnits 2.17.1 draft-ietf-ippm-loss-episode-metrics-04.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- No issues found here. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year == The document seems to lack the recommended RFC 2119 boilerplate, even if it appears to use RFC 2119 keywords -- however, there's a paragraph with a matching beginning. Boilerplate error? (The document does seem to have the reference to RFC 2119 which the ID-Checklist requires). == The document seems to contain a disclaimer for pre-RFC5378 work, but was first submitted on or after 10 November 2008. The disclaimer is usually necessary only for documents that revise or obsolete older RFCs, and that take significant amounts of text from those RFCs. If you can contact all authors of the source material and they are willing to grant the BCP78 rights to the IETF Trust, you can and should remove the disclaimer. Otherwise, the disclaimer is needed and you can ignore this comment. (See the Legal Provisions document at https://trustee.ietf.org/license-info for more information.) -- The document date (January 17, 2012) is 4482 days in the past. Is this intentional? Checking references for intended status: Proposed Standard ---------------------------------------------------------------------------- (See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) -- Looks like a reference, but probably isn't: '0' on line 676 -- Looks like a reference, but probably isn't: '1' on line 676 ** Obsolete normative reference: RFC 2680 (Obsoleted by RFC 7680) Summary: 1 error (**), 0 flaws (~~), 3 warnings (==), 3 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network Working Group N. Duffield 3 Internet-Draft AT&T Labs-Research 4 Intended status: Standards Track A. Morton 5 Expires: July 20, 2012 AT&T Labs 6 J. Sommers 7 Colgate University 8 January 17, 2012 10 Loss Episode Metrics for IPPM 11 draft-ietf-ippm-loss-episode-metrics-04 13 Abstract 15 The IETF has developed a one way packet loss metric that measures the 16 loss rate on a Poisson probe stream between two hosts. However, the 17 impact of packet loss on applications is in general sensitive not 18 just to the average loss rate, but also to the way in which packet 19 losses are distributed in loss episodes (i.e., maximal sets of 20 consecutively lost probe packets). This document defines one-way 21 packet loss episode metrics, specifically the frequency and average 22 duration of loss episodes, and a probing methodology under which the 23 loss episode metrics are to be measured. 25 Requirements Language 27 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 28 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 29 document are to be interpreted as described in RFC 2119 [RFC2119] 31 Status of this Memo 33 This Internet-Draft is submitted in full conformance with the 34 provisions of BCP 78 and BCP 79. 36 Internet-Drafts are working documents of the Internet Engineering 37 Task Force (IETF). Note that other groups may also distribute 38 working documents as Internet-Drafts. The list of current Internet- 39 Drafts is at http://datatracker.ietf.org/drafts/current/. 41 Internet-Drafts are draft documents valid for a maximum of six months 42 and may be updated, replaced, or obsoleted by other documents at any 43 time. It is inappropriate to use Internet-Drafts as reference 44 material or to cite them other than as "work in progress." 46 This Internet-Draft will expire on July 20, 2012. 48 Copyright Notice 49 Copyright (c) 2012 IETF Trust and the persons identified as the 50 document authors. All rights reserved. 52 This document is subject to BCP 78 and the IETF Trust's Legal 53 Provisions Relating to IETF Documents 54 (http://trustee.ietf.org/license-info) in effect on the date of 55 publication of this document. Please review these documents 56 carefully, as they describe your rights and restrictions with respect 57 to this document. Code Components extracted from this document must 58 include Simplified BSD License text as described in Section 4.e of 59 the Trust Legal Provisions and are provided without warranty as 60 described in the Simplified BSD License. 62 This document may contain material from IETF Documents or IETF 63 Contributions published or made publicly available before November 64 10, 2008. The person(s) controlling the copyright in some of this 65 material may not have granted the IETF Trust the right to allow 66 modifications of such material outside the IETF Standards Process. 67 Without obtaining an adequate license from the person(s) controlling 68 the copyright in such materials, this document may not be modified 69 outside the IETF Standards Process, and derivative works of it may 70 not be created outside the IETF Standards Process, except to format 71 it for publication as an RFC or to translate it into languages other 72 than English. 74 Table of Contents 76 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 5 77 1.1. Background and Motivation . . . . . . . . . . . . . . . . 5 78 1.2. Loss Episode Metrics and Bi-Packet Probes . . . . . . . . 6 79 1.3. Outline and Contents . . . . . . . . . . . . . . . . . . . 7 80 2. Singleton Definition for Type-P-One-way Bi-Packet Loss . . . . 8 81 2.1. Metric Name . . . . . . . . . . . . . . . . . . . . . . . 8 82 2.2. Metric Parameters . . . . . . . . . . . . . . . . . . . . 8 83 2.3. Metric Units . . . . . . . . . . . . . . . . . . . . . . . 8 84 2.4. Metric Definition . . . . . . . . . . . . . . . . . . . . 8 85 2.5. Discussion . . . . . . . . . . . . . . . . . . . . . . . . 9 86 2.6. Methodologies . . . . . . . . . . . . . . . . . . . . . . 9 87 2.7. Errors and Uncertainties . . . . . . . . . . . . . . . . . 9 88 2.8. Reporting the Metric . . . . . . . . . . . . . . . . . . . 9 89 3. General Definition of samples for 90 Type-P-One-way-Bi-Packet-Loss . . . . . . . . . . . . . . . . 9 91 3.1. Metric Name . . . . . . . . . . . . . . . . . . . . . . . 10 92 3.2. Metric Parameters . . . . . . . . . . . . . . . . . . . . 10 93 3.3. Metric Units . . . . . . . . . . . . . . . . . . . . . . . 10 94 3.4. Metric Definition . . . . . . . . . . . . . . . . . . . . 10 95 3.5. Discussion . . . . . . . . . . . . . . . . . . . . . . . . 10 96 3.6. Methodologies . . . . . . . . . . . . . . . . . . . . . . 10 97 3.7. Errors and Uncertainties . . . . . . . . . . . . . . . . . 11 98 3.8. Reporting the Metric . . . . . . . . . . . . . . . . . . . 11 99 4. An active probing methodology for Bi-Packet Loss . . . . . . . 11 100 4.1. Metric Name . . . . . . . . . . . . . . . . . . . . . . . 11 101 4.2. Metric Parameters . . . . . . . . . . . . . . . . . . . . 11 102 4.3. Metric Units . . . . . . . . . . . . . . . . . . . . . . . 12 103 4.4. Metric Definition . . . . . . . . . . . . . . . . . . . . 12 104 4.5. Discussion . . . . . . . . . . . . . . . . . . . . . . . . 12 105 4.6. Methodologies . . . . . . . . . . . . . . . . . . . . . . 12 106 4.7. Errors and Uncertainties . . . . . . . . . . . . . . . . . 13 107 4.8. Reporting the Metric . . . . . . . . . . . . . . . . . . . 13 108 5. Loss Episode Proto-Metrics . . . . . . . . . . . . . . . . . . 13 109 5.1. Loss-Pair-Counts . . . . . . . . . . . . . . . . . . . . . 13 110 5.2. Bi-Packet-Loss-Ratio . . . . . . . . . . . . . . . . . . . 14 111 5.3. Bi-Packet-Loss-Episode-Duration-Number . . . . . . . . . . 14 112 5.4. Bi-Packet-Loss-Episode-Frequency-Number . . . . . . . . . 14 113 6. Loss Episode Metrics derived from Bi-Packet Loss Probing . . . 14 114 6.1. Geometric Stream: Loss Ratio . . . . . . . . . . . . . . . 15 115 6.1.1. Metric Name . . . . . . . . . . . . . . . . . . . . . 15 116 6.1.2. Metric Parameters . . . . . . . . . . . . . . . . . . 15 117 6.1.3. Metric Units . . . . . . . . . . . . . . . . . . . . . 16 118 6.1.4. Metric Definition . . . . . . . . . . . . . . . . . . 16 119 6.1.5. Discussion . . . . . . . . . . . . . . . . . . . . . . 16 120 6.1.6. Methodologies . . . . . . . . . . . . . . . . . . . . 16 121 6.1.7. Errors and Uncertainties . . . . . . . . . . . . . . . 16 122 6.1.8. Reporting the Metric . . . . . . . . . . . . . . . . . 16 123 6.2. Geometric Stream: Loss Episode Duration . . . . . . . . . 16 124 6.2.1. Metric Name . . . . . . . . . . . . . . . . . . . . . 16 125 6.2.2. Metric Parameters . . . . . . . . . . . . . . . . . . 16 126 6.2.3. Metric Units . . . . . . . . . . . . . . . . . . . . . 17 127 6.2.4. Metric Definition . . . . . . . . . . . . . . . . . . 17 128 6.2.5. Discussion . . . . . . . . . . . . . . . . . . . . . . 17 129 6.2.6. Methodologies . . . . . . . . . . . . . . . . . . . . 17 130 6.2.7. Errors and Uncertainties . . . . . . . . . . . . . . . 17 131 6.2.8. Reporting the Metric . . . . . . . . . . . . . . . . . 18 132 6.3. Geometric Stream: Loss Episode Frequency . . . . . . . . . 18 133 6.3.1. Metric Name . . . . . . . . . . . . . . . . . . . . . 18 134 6.3.2. Metric Parameters . . . . . . . . . . . . . . . . . . 18 135 6.3.3. Metric Units . . . . . . . . . . . . . . . . . . . . . 18 136 6.3.4. Metric Definition . . . . . . . . . . . . . . . . . . 18 137 6.3.5. Discussion . . . . . . . . . . . . . . . . . . . . . . 19 138 6.3.6. Methodologies . . . . . . . . . . . . . . . . . . . . 19 139 6.3.7. Errors and Uncertainties . . . . . . . . . . . . . . . 19 140 6.3.8. Reporting the Metric . . . . . . . . . . . . . . . . . 19 141 7. Applicability of Loss Episode Metrics . . . . . . . . . . . . 19 142 7.1. Relation to Gilbert Model . . . . . . . . . . . . . . . . 19 143 8. IPR Considerations . . . . . . . . . . . . . . . . . . . . . . 20 144 9. Security Considerations . . . . . . . . . . . . . . . . . . . 20 145 10. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 21 146 11. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 21 147 12. References . . . . . . . . . . . . . . . . . . . . . . . . . . 21 148 12.1. Normative References . . . . . . . . . . . . . . . . . . . 21 149 12.2. Informative References . . . . . . . . . . . . . . . . . . 21 150 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 22 152 1. Introduction 154 1.1. Background and Motivation 156 Packet loss in the Internet is a complex phenomenon due to the bursty 157 nature of traffic and congestion processes, influenced by both end- 158 users and applications, and the operation of transport protocols such 159 as TCP. For these reasons, the simplest model of packet loss--the 160 single parameter Bernoulli (independent) loss model--does not 161 represent the complexity of packet loss over periods of time. 162 Correspondingly, a single loss metric--the average packet loss ratio 163 over some period of time--arising, e.g., from a stream of Poisson 164 probes as in [RFC2680] is not sufficient to determine the effect of 165 packet loss on traffic in general. 167 Moving beyond single parameter loss models, Markovian and Markov- 168 modulated loss models involving transitions between a good and bad 169 state, each with an associated loss rate, have been proposed by 170 Gilbert [Gilbert] and more generally by Elliot [Elliot]. In 171 principle, Markovian models can be formulated over state spaces 172 involving patterns of loss of any desired number of packets. However 173 further increase in the size of the state space makes such models 174 cumbersome both for parameter estimation (accuracy decreases) and 175 prediction in practice (due to computational complexity and 176 sensitivity to parameter inaccuracy). In general, the relevance and 177 importance of particular models can change in time, e.g. in response 178 to the advent of new applications and services. For this reason we 179 are drawn to empirical metrics that do not depend on a particular 180 model for their interpretation. 182 An empirical measure of packet loss complexity, the index of 183 dispersion of counts (IDC), comprise, for each t >0, the ratio v(t) / 184 a(t) of the variance v(t) and average a(t) of the number of losses 185 over successive measurement windows of a duration t. However, a full 186 characterization of packet loss over time requires specification of 187 the IDC for each window size t>0. 189 In the standards arena, loss pattern sample metrics are defined in 190 [RFC3357]. Following the Gilbert-Elliot model, burst metrics 191 specific for VoIP that characterize complete episodes of lost, 192 transmitted and discarded packets are defined in [RFC3611] 194 All these considerations motivate formulating empirical metrics of 195 one-way packet loss that provide the simplest generalization of the 196 successful [RFC2680] that can capture deviations from independent 197 packet loss in a robust model-independent manner, and, to define 198 efficient measurement methodologies for these metrics. 200 1.2. Loss Episode Metrics and Bi-Packet Probes 202 The losses experienced by the packet stream can be viewed as 203 occurring in loss episodes, i.e., maximal set of consecutively lost 204 packets. This memo describes one-way loss episode metrics: their 205 frequency and average duration. Although the average loss ratio can 206 be expressed in terms of these quantities, they go further in 207 characterizing the statistics of the patterns of packet loss within 208 the stream of probes. This is useful information in understanding 209 the effect of packet losses on application performance, since 210 different applications can have different sensitivities to patterns 211 of loss, being sensitive not only to the long term average loss rate, 212 but how losses are distributed in time. As an example: MPEG video 213 traffic may be sensitive to loss involving the I-frame in a group of 214 pictures, but further losses within an episode of sufficiently short 215 duration have no further impact; the damage is already done. 217 The loss episode metrics presented here have the following useful 218 properties: 220 1. the metrics are empirical and do not depend on an underlying 221 model; e.g., the loss process is not assumed to be Markovian. On 222 the other hand, it turns out that the metrics of this memo can be 223 related to the special case of the Gilbert Model parameters; see 224 Section 7. 226 2. the metric units can be directly compared with applications or 227 user requirements or tolerance for network loss performance, in 228 the frequency and duration of loss episodes, as well as the usual 229 packet loss ratio, which can be recovered from the loss episode 230 metrics upon dividing the average loss episode duration by the 231 loss episode frequency. 233 3. the metrics provide the smallest possible increment in complexity 234 beyond, but in the spirit of, the IPPM average packet loss ratio 235 metrics [RFC2680] i.e., moving from a single metric (average 236 packet loss ratio) to a pair of metrics (loss episode frequency 237 and average loss episode duration). 239 The document also describes a probing methodology under which loss 240 episode metrics are to be measured. The methodology comprises 241 sending probe packets in pairs, where packets within each probe pair 242 have a fixed separation, and the time between pairs takes the form of 243 a geometric distributed number multiplied by the same separation. 244 This can be regarded a generalization of Poisson probing where the 245 probes are pairs rather than single packets as in [RFC2680], and also 246 of geometric probing described in [RFC2330]. However, it should be 247 distinguished from back to back packet pairs whose change in 248 separation on traversing a link is used to probe bandwidth. In this 249 document, the separation between the packets in a pair is the 250 temporal resolution at which different loss episodes are to be 251 distinguished. The methodology does not measure episodes of loss of 252 consecutive background packets on the measured path. One key feature 253 of this methodology is its efficiency: it estimates the average 254 length of loss episodes without directly measuring the complete 255 episodes themselves. Instead, this information is encoded in the 256 observed relative frequencies of the 4 possible outcomes arising from 257 the loss or successful transmission of each of the two packets of the 258 probe pairs. This is distinct from the approach of [RFC3611] that 259 reports on directly measured episodes. 261 The metrics defined in this memo are "derived metrics", according to 262 Section 6.1 of [RFC2330] the IPPM framework. They are based on the 263 singleton loss metric defined in Section 2 of [RFC2680] . 265 1.3. Outline and Contents 267 o Section 2 defines the fundamental singleton metric for the 268 possible outcomes of a probe pair: Type-P-One-way-Bi-Packet-Loss. 270 o Section 3 defines sample sets of this metric derived from a 271 general probe stream: Type-P-One-way-Bi-Packet-Loss-Stream. 273 o Section 4 defines the prime example of the Bi-Packet-Loss-Stream 274 metrics, specifically Type-P-One-way-Bi-Packet-Loss-Geometric- 275 Stream arising from the geometric stream of packet-pair probes 276 that was described informally in Section 1. 278 o Section 5 defines Loss episode proto-metrics that summarize the 279 outcomes from a stream metrics as an intermediate step to forming 280 the loss episode metrics; they need not be reported in general. 282 o Section 6 defines the final loss episode metrics that are the 283 focus of this memo, the new metrics 285 * Type-P-One-way-Bi-Packet-Loss-Geometric-Stream-Episode- 286 Duration, the average duration, in seconds, of a loss episode 288 * Type-P-One-way-Bi-Packet-Loss-Geometric-Stream-Episode- 289 Frequency, the average frequency, per second, at which loss 290 episodes start. 292 * Type-P-One-way-Bi-Packet-Loss-Geometric-Stream-Ratio, which is 293 the average packet loss ratio metric arising from the geometric 294 stream probing methodology 296 o Section 7 details applications and relations to existing loss 297 models. 299 2. Singleton Definition for Type-P-One-way Bi-Packet Loss 301 2.1. Metric Name 303 Type-P-One-way-Bi-Packet-Loss 305 2.2. Metric Parameters 307 o Src, the IP address of a source host 309 o Dst, the IP address of a destination host 311 o T1, a sending time of the first packet 313 o T2, a sending time of the second packet, with T2>T1 315 o F, a selection function defining unambiguously the two packets 316 from the stream selected for the metric. 318 o P, the specification of the packet type, over and above the source 319 and destination addresses 321 2.3. Metric Units 323 A Loss Pair is pair (l1, l2) where each of l1 and l2 is a binary 324 value 0 or 1, where 0 signifies successful transmission of a packet 325 and 1 signifies loss. 327 The metric unit of Type-P-One-way-Bi-Packet-Loss is a Loss Pair. 329 2.4. Metric Definition 331 1. "The Type-P-One-way-Bi-Packet-Loss with parameters (Src, Dst, T1, 332 T2, F, P) is (1,1)" means that Src sent the first bit of a Type-P 333 packet to Dst at wire-time T1 and the first bit of a Type-P 334 packet to Dst at wire-time T2>T1, and that neither packet was 335 received at Dst. 337 2. The Type-P-One-way-Bi-Packet-Loss with parameters (Src, Dst, T1, 338 T2, F, P) is (1,0)" means that Src sent the first bit of a Type-P 339 packet to Dst at wire-time T1 and the first bit of a Type-P 340 packet to Dst at wire-time T2>T1, and that the first packet was 341 not received at Dst, and the second packet was received at Dst 343 3. The Type-P-One-way-Bi-Packet-Loss with parameters (Src, Dst, T1, 344 T2, F, P) is (0,1)" means that Src sent the first bit of a Type-P 345 packet to Dst at wire-time T1 and the first bit of a Type-P 346 packet to Dst at wire-time T2>T1, and that the first packet was 347 received at Dst, and the second packet was not received at Dst 349 4. The Type-P-One-way-Bi-Packet-Loss with parameters (Src, Dst, T1, 350 T2, F, P) is (0,0)" means that Src sent the first bit of a Type-P 351 packet to Dst at wire-time T1 and the first bit of a Type-P 352 packet to Dst at wire-time T2>T1, and that both packets were 353 received at Dst. 355 2.5. Discussion 357 The purpose of the selection function is to specify exactly which 358 packets are to be used for measurement. The notion is taken from 359 Section 2.5 of [RFC3393], where examples are discussed. 361 2.6. Methodologies 363 The methodologies related to the Type-P-One-way-Packet-Loss metric in 364 Section 2.6 of [RFC2680] are similar for the Type-P-One-way-Bi- 365 Packet-Loss metric described above. In particular, the methodologies 366 described in RFC 2680 apply to both packets of the pair. 368 2.7. Errors and Uncertainties 370 Sources of error for the Type-P-One-way-Packet-Loss metric in Section 371 2.7 of [RFC2680] apply to each packet of the pair for the Type-P-One- 372 way-Bi-Packet-Loss metric. 374 2.8. Reporting the Metric 376 Refer to Section 2.8 of [RFC2680]. 378 3. General Definition of samples for Type-P-One-way-Bi-Packet-Loss 380 Given the singleton metric for Type-P-One-way-Bi-Packet-Loss, we now 381 define examples of samples of singletons. The basic idea is as 382 follows. We first specify a set of times T1 < T2 <...1 598 o 0 if N(0,1) + N(1,0) + N(1,1) = 0 (no probe packets lost) 600 o Undefined if N(0,1) + N(1,0) + N(0,0) = 0 (all probe packets lost) 602 Note N(0,1) + N(1,0) is zero if there are no transitions between loss 603 and no-loss outcomes. 605 5.4. Bi-Packet-Loss-Episode-Frequency-Number 607 The Bi-Packet-Loss-Episode-Frequency-Number associated with a set of 608 n loss pairs L1,,,,Ln is defined in terms of their Loss-Pair-Counts 609 as Bi-Packet-Loss-Ratio / Bi-Packet-Loss-Episode-Duration-Number, 610 when this can be defined, specifically, it is: 612 o (N(1,0)+N(1,1)) * (N(0,1)+N(1,0)) / (2*N(1,1)+N(0,1)+N(1,0) ) / n 613 if N(0,1)+N(0,1) > 0 615 o 0 if N(0,1)+N(1,0) +N(1,1) = 0 (no probe packets lost) 617 o 1 if N(0,1) +N(1,0) +N(0,0) = 0 (all probe packets lost) 619 6. Loss Episode Metrics derived from Bi-Packet Loss Probing 621 Metrics for the time frequency and time duration of loss episodes are 622 now defined as functions of set of n loss pairs L1,....,Ln. Although 623 a loss episode is defined as a maximal set of successive lost 624 packets, the loss episode metrics are not defined directly in terms 625 of the sequential patterns of packet loss exhibited by loss pairs. 626 This is because samples, including Type-P-One-way-Bi-Packet-Loss- 627 Geometric-Stream, generally do not report all lost packets in each 628 episode. Instead, the metrics are defined as functions of the Loss- 629 Pair-Counts of the sample, for reasons that are now described. 631 Consider an idealized Type-P-One-way-Bi-Packet-Loss-Geometric-Stream 632 sample in which the launch probability q =1. It is shown in [SBDR08] 633 that the average number of packets in a loss episode of this ideal 634 sample is exactly the Bi-Packet-Loss-Episode-Duration derived from 635 its set of loss pairs. Note this computation makes no reference to 636 the position of lost packet in the sequence of probes. 638 A general Type-P-One-way-Bi-Packet-Loss-Geometric-Stream sample with 639 launch probability q < 1, independently samples, with probability q, 640 each loss pair of an idealized sample. On average, the Loss-Pair- 641 Counts (if normalized by the total number of pairs) will be the same 642 as in the idealized sample. The loss episode metrics in the general 643 case are thus estimators of those for the idealized case; the 644 statistical properties of this estimation, including a derivation of 645 the estimation variance, is provided in [SBDR08]. 647 6.1. Geometric Stream: Loss Ratio 649 6.1.1. Metric Name 651 Type-P-One-way-Bi-Packet-Loss-Geometric-Stream-Ratio 653 6.1.2. Metric Parameters 655 o Src, the IP address of a source host 657 o Dst, the IP address of a destination host 659 o T0, the randomly selected starting time [RFC3432] for periodic 660 launch opportunities 662 o d, the time spacing between potential launch times, Ti and Ti+1 664 o n, a count of potential measurement instants 666 o q, a launch probability 668 o F, a selection function defining unambiguously the two packets 669 from the stream selected for the metric. 671 o P, the specification of the packet type, over and above the source 672 and destination address 674 6.1.3. Metric Units 676 A decimal number in the interval [0,1] 678 6.1.4. Metric Definition 680 The result obtained by computing the Bi-Packet-Loss-Ratio over a 681 Type-P-One-way-Bi-Packet-Loss-Geometric-Stream sample with the metric 682 parameters. 684 6.1.5. Discussion 686 Type-P-One-way-Bi-Packet-Loss-Geometric-Stream-Ratio estimates the 687 fraction of packets lost from the geometric stream of Bi-Packet 688 probes. 690 6.1.6. Methodologies 692 Refer to Section 4.6 694 6.1.7. Errors and Uncertainties 696 Because Type-P-One-way-Bi-Packet-Loss-Geometric-Stream is sampled in 697 general (when the launch probability q <1) the metrics described in 698 this Section can be regarded as statistical estimators of the 699 corresponding idealized version corresponding to q = 1. Estimation 700 variance as it applies to Type-P-One-way-Bi-Packet-Loss-Geometric- 701 Stream-Loss-Ratio is described in [SBDR08]. 703 For other issues refer to Section 4.7 705 6.1.8. Reporting the Metric 707 Refer to Section 4.8 709 6.2. Geometric Stream: Loss Episode Duration 711 6.2.1. Metric Name 713 Type-P-One-way-Bi-Packet-Loss-Geometric-Stream-Episode-Duration 715 6.2.2. Metric Parameters 716 o Src, the IP address of a source host 718 o Dst, the IP address of a destination host 720 o T0, the randomly selected starting time [RFC3432] for periodic 721 launch opportunities 723 o d, the time spacing between potential launch times, Ti and Ti+1 725 o n, a count of potential measurement instants 727 o q, a launch probability 729 o F, a selection function defining unambiguously the two packets 730 from the stream selected for the metric. 732 o P, the specification of the packet type, over and above the source 733 and destination address 735 6.2.3. Metric Units 737 A non-negative number of seconds. 739 6.2.4. Metric Definition 741 The result obtained by computing the Bi-Packet-Loss-Episode-Duration- 742 Number over a Type-P-One-way-Bi-Packet-Loss-Geometric-Stream sample 743 with the metric parameters, then multiplying the result by the launch 744 spacing parameter d. 746 6.2.5. Discussion 748 Type-P-One-way-Bi-Packet-Loss-Geometric-Stream-Episode-Duration 749 estimates the average duration of a loss episode, measured in 750 seconds. The duration measured in packets is obtained by dividing 751 the metric value by the packet launch spacing parameter d. 753 6.2.6. Methodologies 755 Refer to Section 4.6 757 6.2.7. Errors and Uncertainties 759 Because Type-P-One-way-Bi-Packet-Loss-Geometric-Stream is sampled in 760 general (when the launch probability q <1) the metrics described in 761 this Section can be regarded as statistical estimators of the 762 corresponding idealized version corresponding to q = 1. Estimation 763 variance as it applies to Type-P-One-way-Bi-Packet-Loss-Geometric- 764 Stream-Episode-Duration is described in [SBDR08]. 766 For other issues refer to Section 4.7 768 6.2.8. Reporting the Metric 770 Refer to Section 4.8 772 6.3. Geometric Stream: Loss Episode Frequency 774 6.3.1. Metric Name 776 Type-P-One-way-Bi-Packet-Loss-Geometric-Stream-Episode-Frequency 778 6.3.2. Metric Parameters 780 o Src, the IP address of a source host 782 o Dst, the IP address of a destination host 784 o T0, the randomly selected starting time [RFC3432] for periodic 785 launch opportunities 787 o d, the time spacing between potential launch times, Ti and Ti+1 789 o n, a count of potential measurement instants 791 o q, a launch probability 793 o F, a selection function defining unambiguously the two packets 794 from the stream selected for the metric. 796 o P, the specification of the packet type, over and above the source 797 and destination address 799 6.3.3. Metric Units 801 A positive number. 803 6.3.4. Metric Definition 805 The result obtained by computing the Bi-Packet-Loss-Episode-Frequency 806 over a Type-P-One-way-Bi-Packet-Loss-Geometric-Stream sample with the 807 metric parameters, then dividing the result by the launch spacing 808 parameter d. 810 6.3.5. Discussion 812 Type-P-One-way-Bi-Packet-Loss-Geometric-Stream-Episode-Frequency 813 estimates the average frequency per unit time with which loss 814 episodes start (or finish). The frequency relative to the count of 815 potential probe launches is obtained by multiplying the metric value 816 by the packet launch spacing parameter d. 818 6.3.6. Methodologies 820 Refer to Section 4.6 822 6.3.7. Errors and Uncertainties 824 Because Type-P-One-way-Bi-Packet-Loss-Geometric-Stream is sampled in 825 general (when the launch probability q <1) the metrics described in 826 this Section can be regarded as statistical estimators of the 827 corresponding idealized version corresponding to q = 1. Estimation 828 variance as it applies to Type-P-One-way-Bi-Packet-Loss-Geometric- 829 Stream-Episode-Frequency is described in [SBDR08]. 831 For other issues refer to Section 4.7 833 6.3.8. Reporting the Metric 835 Refer to Section 4.8 837 7. Applicability of Loss Episode Metrics 839 7.1. Relation to Gilbert Model 841 The general Gilbert-Elliot model is a discrete time Markov chain over 842 two states, Good (g) and Bad (b), each with its own independent 843 packet loss rate. In the simplest case, the Good loss rate is 0 844 while the Bad loss rate is 1. Correspondingly, there are two 845 independent parameters, the Markov transition probabilities P(g|b) = 846 1- P(b|b) and P(b|g) = 1- P(g|g), where P(i|j) is the probability to 847 transition from state j and step n to state i at step n+1. With 848 these parameters, the fraction of steps spent in the bad state is 849 P(b|g)/(P(b|g) + P(g|b)) while the average duration of a sojourn in 850 the bad state is 1/P(g|b) steps. 852 Now identify the steps of the Markov chain with the possible sending 853 times of packets for a Type-P-One-way-Bi-Packet-Loss-Geometric-Stream 854 with launch spacing d. Suppose the loss episode metrics Type-P-One- 855 way-Bi-Packet-Loss-Geometric-Stream-Ratio and Type-P-One-way-Bi- 856 Packet-Loss-Geometric-Stream-Episode-Duration take the values r and m 857 respectively. Then from the discussion in Section 6.2.5 the 858 following can be equated: 860 r = P(b|g)/(P(b|g) + P(g|b)) and m/d = 1/P(g|b). 862 These relationships can be inverted in order to recover the Gilbert 863 model parameters: 865 P(g|b) = d/m and P(b|g)=d/m/(1/r - 1) 867 8. IPR Considerations 869 An IPR disclosure concerning some of the material covered in this 870 document has been made to the IETF: see 871 https://datatracker.ietf.org/ipr/1354/ 873 9. Security Considerations 875 Conducting Internet measurements raises both security and privacy 876 concerns. This memo does not specify an implementation of the 877 metrics, so it does not directly affect the security of the Internet 878 nor of applications which run on the Internet. 879 However,implementations of these metrics must be mindful of security 880 and privacy concerns. 882 There are two types of security concerns: potential harm caused by 883 the measurements, and potential harm to the measurements. The 884 measurements could cause harm because they are active, and inject 885 packets into the network. The measurement parameters MUST be 886 carefully selected so that the measurements inject trivial amounts of 887 additional traffic into the networks they measure. If they inject 888 "too much" traffic, they can skew the results of the measurement, and 889 in extreme cases cause congestion and denial of service. The 890 measurements themselves could be harmed by routers giving measurement 891 traffic a different priority than "normal" traffic, or by an attacker 892 injecting artificial measurement traffic. If routers can recognize 893 measurement traffic and treat it separately, the measurements may not 894 reflect actual user traffic. If an attacker injects artificial 895 traffic that is accepted as legitimate, the loss rate will be 896 artificially lowered. Therefore, the measurement methodologies 897 SHOULD include appropriate techniques to reduce the probability that 898 measurement traffic can be distinguished from "normal" traffic. 899 Authentication techniques, such as digital signatures, may be used 900 where appropriate to guard against injected traffic attacks. The 901 privacy concerns of network measurement are limited by the active 902 measurements described in this memo: they involve no release of user 903 data. 905 10. IANA Considerations 907 This document requests no actions from IANA. 909 11. Acknowledgements 911 12. References 913 12.1. Normative References 915 [RFC2680] Almes, G., Kalidindi, S., and M. Zekauskas, "A One-way 916 Packet Loss Metric for IPPM", RFC 2680, September 1999. 918 [RFC3393] Demichelis, C. and P. Chimento, "IP Packet Delay Variation 919 Metric for IP Performance Metrics (IPPM)", RFC 3393, 920 November 2002. 922 [RFC3611] Friedman, T., Caceres, R., and A. Clark, "RTP Control 923 Protocol Extended Reports (RTCP XR)", RFC 3611, 924 November 2003. 926 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 927 Requirement Levels", BCP 14, RFC 2119, March 1997. 929 [RFC3432] Raisanen, V., Grotefeld, G., and A. Morton, "Network 930 performance measurement with periodic streams", RFC 3432, 931 November 2002. 933 12.2. Informative References 935 [RFC2330] Paxson, V., Almes, G., Mahdavi, J., and M. Mathis, 936 "Framework for IP Performance Metrics", RFC 2330, 937 May 1998. 939 [RFC3357] Koodli, R. and R. Ravikanth, "One-way Loss Pattern Sample 940 Metrics", RFC 3357, August 2002. 942 [SBDR08] IEEE/ACM Transactions on Networking, 16(2): 307-320, "A 943 Geometric Approach to Improving Active Packet Loss 944 Measurement", 2008. 946 [Gilbert] Gilbert, E.N., "Capacity of a Burst-Noise Channel. Bell 947 System Technical Journal 39 pp 1253-1265", 1960. 949 [Elliot] Elliott, E.O., "Estimates of Error Rates for Codes on 950 Burst-Noise Channels. Bell System Technical Journal 42 pp 951 1977-1997", 1963. 953 Authors' Addresses 955 Nick Duffield 956 AT&T Labs-Research 957 180 Park Avenue 958 Florham Park, NJ 07932 959 USA 961 Phone: +1 973 360 8726 962 Fax: +1 973 360 8871 963 Email: duffield@research.att.com 964 URI: http://www.research.att.com/people/Duffield_Nicholas_G 966 Al Morton 967 AT&T Labs 968 200 Laurel Avenue South 969 Middletown,, NJ 07748 970 USA 972 Phone: +1 732 420 1571 973 Fax: +1 732 368 1192 974 Email: acmorton@att.com 975 URI: http://home.comcast.net/~acmacm/ 977 Joel Sommers 978 Colgate University 979 304 McGregory Hall 980 Hamilton, NY 13346 981 USA 983 Phone: +1 315 228 7587 984 Fax: 985 Email: jsommers@colgate.edu 986 URI: http://cs.colgate.edu/faculty/jsommers