idnits 2.17.00 (12 Aug 2021) /tmp/idnits1680/draft-ietf-ippm-2680-bis-05.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 : ---------------------------------------------------------------------------- -- The draft header indicates that this document obsoletes RFC2680, but the abstract doesn't seem to directly say this. It does mention RFC2680 though, so this could be OK. 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 date (August 20, 2015) is 2459 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) ** Downref: Normative reference to an Informational RFC: RFC 2330 ** Obsolete normative reference: RFC 2679 (Obsoleted by RFC 7679) ** Obsolete normative reference: RFC 2680 (Obsoleted by RFC 7680) ** Downref: Normative reference to an Informational RFC: RFC 7312 == Outdated reference: draft-ietf-lmap-framework has been published as RFC 7594 == Outdated reference: A later version (-02) exists of draft-morton-ippm-2330-stdform-typep-00 Summary: 4 errors (**), 0 flaws (~~), 4 warnings (==), 2 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network Working Group G. Almes 3 Internet-Draft Texas A&M 4 Obsoletes: 2680 (if approved) S. Kalidindi 5 Intended status: Standards Track Ixia 6 Expires: February 21, 2016 M. Zekauskas 7 Internet2 8 A. Morton, Ed. 9 AT&T Labs 10 August 20, 2015 12 A One-Way Loss Metric for IPPM 13 draft-ietf-ippm-2680-bis-05 15 Abstract 17 This memo (RFC 2680 bis) defines a metric for one-way loss of packets 18 across Internet paths. It builds on notions introduced and discussed 19 in the IPPM Framework document, RFC 2330; the reader is assumed to be 20 familiar with that document. This memo makes RFC 2680 obsolete. 22 Status of This Memo 24 This Internet-Draft is submitted in full conformance with the 25 provisions of BCP 78 and BCP 79. 27 Internet-Drafts are working documents of the Internet Engineering 28 Task Force (IETF). Note that other groups may also distribute 29 working documents as Internet-Drafts. The list of current Internet- 30 Drafts is at http://datatracker.ietf.org/drafts/current/. 32 Internet-Drafts are draft documents valid for a maximum of six months 33 and may be updated, replaced, or obsoleted by other documents at any 34 time. It is inappropriate to use Internet-Drafts as reference 35 material or to cite them other than as "work in progress." 37 This Internet-Draft will expire on February 21, 2016. 39 Copyright Notice 41 Copyright (c) 2015 IETF Trust and the persons identified as the 42 document authors. All rights reserved. 44 This document is subject to BCP 78 and the IETF Trust's Legal 45 Provisions Relating to IETF Documents 46 (http://trustee.ietf.org/license-info) in effect on the date of 47 publication of this document. Please review these documents 48 carefully, as they describe your rights and restrictions with respect 49 to this document. Code Components extracted from this document must 50 include Simplified BSD License text as described in Section 4.e of 51 the Trust Legal Provisions and are provided without warranty as 52 described in the Simplified BSD License. 54 Table of Contents 56 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2 57 1.1. Motivation . . . . . . . . . . . . . . . . . . . . . . . 3 58 1.2. General Issues Regarding Time . . . . . . . . . . . . . . 4 59 2. A Singleton Definition for One-way Packet Loss . . . . . . . 6 60 2.1. Metric Name: . . . . . . . . . . . . . . . . . . . . . . 6 61 2.2. Metric Parameters: . . . . . . . . . . . . . . . . . . . 6 62 2.3. Metric Units: . . . . . . . . . . . . . . . . . . . . . . 6 63 2.4. Definition: . . . . . . . . . . . . . . . . . . . . . . . 6 64 2.5. Discussion: . . . . . . . . . . . . . . . . . . . . . . . 6 65 2.6. Methodologies: . . . . . . . . . . . . . . . . . . . . . 7 66 2.7. Errors and Uncertainties: . . . . . . . . . . . . . . . . 8 67 2.8. Reporting the metric: . . . . . . . . . . . . . . . . . . 9 68 2.8.1. Type-P . . . . . . . . . . . . . . . . . . . . . . . 10 69 2.8.2. Loss Threshold . . . . . . . . . . . . . . . . . . . 10 70 2.8.3. Calibration Results . . . . . . . . . . . . . . . . . 10 71 2.8.4. Path . . . . . . . . . . . . . . . . . . . . . . . . 10 72 3. A Definition for Samples of One-way Packet Loss . . . . . . . 11 73 3.1. Metric Name: . . . . . . . . . . . . . . . . . . . . . . 11 74 3.2. Metric Parameters: . . . . . . . . . . . . . . . . . . . 11 75 3.3. Metric Units: . . . . . . . . . . . . . . . . . . . . . . 11 76 3.4. Definition: . . . . . . . . . . . . . . . . . . . . . . . 12 77 3.5. Discussion: . . . . . . . . . . . . . . . . . . . . . . . 12 78 3.6. Methodologies: . . . . . . . . . . . . . . . . . . . . . 13 79 3.7. Errors and Uncertainties: . . . . . . . . . . . . . . . . 13 80 3.8. Reporting the metric: . . . . . . . . . . . . . . . . . . 14 81 4. Some Statistics Definitions for One-way Packet Loss . . . . . 14 82 4.1. Type-P-One-way-Packet Loss-Ratio . . . . . . . . . . . . 14 83 5. Security Considerations . . . . . . . . . . . . . . . . . . . 15 84 6. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 16 85 7. Changes from RFC 2680 . . . . . . . . . . . . . . . . . . . . 16 86 8. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 18 87 9. References . . . . . . . . . . . . . . . . . . . . . . . . . 18 88 9.1. Normative References . . . . . . . . . . . . . . . . . . 18 89 9.2. Informative References . . . . . . . . . . . . . . . . . 19 90 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 20 92 1. Introduction 94 This memo defines a metric for one-way packet loss across Internet 95 paths. It builds on notions introduced and discussed in the IPPM 96 Framework document, [RFC2330]; the reader is assumed to be familiar 97 with that document, and its recent update [RFC7312]. 99 This memo is intended to be parallel in structure to a companion 100 document for One-way Delay ("A One-way Delay Metric for IPPM") 101 [RFC2679]; the reader is assumed to be familiar with that document. 103 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 104 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 105 document are to be interpreted as described in[RFC2119]. Although 106 [RFC2119] was written with protocols in mind, the key words are used 107 in this document for similar reasons. They are used to ensure the 108 results of measurements from two different implementations are 109 comparable, and to note instances when an implementation could 110 perturb the network. 112 The structure of the memo is as follows: 114 + A 'singleton' analytic metric, called Type-P-One-way-Packet-Loss, 115 is introduced to measure a single observation of packet transmission 116 or loss. 118 + Using this singleton metric, a 'sample', called Type-P-One-way- 119 Packet-Loss-Poisson-Stream, is introduced to measure a sequence of 120 singleton transmissions and/or losses measured at times taken from a 121 Poisson process. 123 + Using this sample, several 'statistics' of the sample are defined 124 and discussed. 126 This progression from singleton to sample to statistics, with clear 127 separation among them, is important. 129 Whenever a technical term from the IPPM Framework document is first 130 used in this memo, it will be tagged with a trailing asterisk. For 131 example, "term*" indicates that "term" is defined in the Framework. 133 1.1. Motivation 135 Understanding one-way packet loss of Type-P* packets from a source 136 host* to a destination host is useful for several reasons: 138 + Some applications do not perform well (or at all) if end-to-end 139 loss between hosts is large relative to some threshold value. 141 + Excessive packet loss may make it difficult to support certain 142 real-time applications (where the precise threshold of "excessive" 143 depends on the application). 145 + The larger the value of packet loss, the more difficult it is for 146 transport-layer protocols to sustain high bandwidths. 148 + The sensitivity of real-time applications and of transport-layer 149 protocols to loss become especially important when very large delay- 150 bandwidth products must be supported. 152 The measurement of one-way loss instead of round-trip loss is 153 motivated by the following factors: 155 + In today's Internet, the path from a source to a destination may be 156 different than the path from the destination back to the source 157 ("asymmetric paths"), such that different sequences of routers are 158 used for the forward and reverse paths. Therefore round-trip 159 measurements actually measure the performance of two distinct paths 160 together. Measuring each path independently highlights the 161 performance difference between the two paths which may traverse 162 different Internet service providers, and even radically different 163 types of networks (for example, research versus commodity networks, 164 or networks with asymmetric link capacities, or wireless vs. wireline 165 access). 167 + Even when the two paths are symmetric, they may have radically 168 different performance characteristics due to asymmetric queueing. 170 + Performance of an application may depend mostly on the performance 171 in one direction. For example, a TCP-based communication will 172 experience reduced throughput if congestion occurs in one direction 173 of its communication. Trouble shooting may be simplified if the 174 congested direction of TCP transmission can be identified. 176 + In quality-of-service (QoS) enabled networks, provisioning in one 177 direction may be radically different than provisioning in the reverse 178 direction, and thus the QoS guarantees differ. Measuring the paths 179 independently allows the verification of both guarantees. 181 It is outside the scope of this document to say precisely how loss 182 metrics would be applied to specific problems. 184 1.2. General Issues Regarding Time 186 {Comment: the terminology below differs from that defined by ITU-T 187 documents (e.g., G.810, "Definitions and terminology for 188 synchronization networks" and I.356, "B-ISDN ATM layer cell transfer 189 performance"), but is consistent with the IPPM Framework document. 190 In general, these differences derive from the different backgrounds; 191 the ITU-T documents historically have a telephony origin, while the 192 authors of this document (and the Framework) have a computer systems 193 background. Although the terms defined below have no direct 194 equivalent in the ITU-T definitions, after our definitions we will 195 provide a rough mapping. However, note one potential confusion: our 196 definition of "clock" is the computer operating systems definition 197 denoting a time-of-day clock, while the ITU-T definition of clock 198 denotes a frequency reference.} 200 Whenever a time (i.e., a moment in history) is mentioned here, it is 201 understood to be measured in seconds (and fractions) relative to UTC. 203 As described more fully in the Framework document, there are four 204 distinct, but related notions of clock uncertainty: 206 synchronization* 208 measures the extent to which two clocks agree on what time it is. 209 For example, the clock on one host might be 5.4 msec ahead of the 210 clock on a second host. {Comment: A rough ITU-T equivalent is "time 211 error".} 213 accuracy* 215 measures the extent to which a given clock agrees with UTC. For 216 example, the clock on a host might be 27.1 msec behind UTC. {Comment: 217 A rough ITU-T equivalent is "time error from UTC".} 219 resolution* 221 specification of the smallest unit by which the clock's time is 222 updated. It gives a lower bound on the clock's uncertainty. For 223 example, the clock on an old Unix host might tick only once every 10 224 msec, and thus have a resolution of only 10 msec. {Comment: A very 225 rough ITU-T equivalent is "sampling period".} 227 skew* 229 measures the change of accuracy, or of synchronization, with time. 230 For example, the clock on a given host might gain 1.3 msec per hour 231 and thus be 27.1 msec behind UTC at one time and only 25.8 msec an 232 hour later. In this case, we say that the clock of the given host 233 has a skew of 1.3 msec per hour relative to UTC, which threatens 234 accuracy. We might also speak of the skew of one clock relative to 235 another clock, which threatens synchronization. {Comment: A rough 236 ITU-T equivalent is "time drift".} 238 2. A Singleton Definition for One-way Packet Loss 240 2.1. Metric Name: 242 Type-P-One-way-Packet-Loss 244 2.2. Metric Parameters: 246 + Src, the IP address of a host 248 + Dst, the IP address of a host 250 + T, a time 252 + Tmax, a loss threshold waiting time 254 2.3. Metric Units: 256 The value of a Type-P-One-way-Packet-Loss is either a zero 257 (signifying successful transmission of the packet) or a one 258 (signifying loss). 260 2.4. Definition: 262 >>The *Type-P-One-way-Packet-Loss* from Src to Dst at T is 0<< means 263 that Src sent the first bit of a Type-P packet to Dst at wire-time* T 264 and that Dst received that packet. 266 >>The *Type-P-One-way-Packet-Loss* from Src to Dst at T is 1<< means 267 that Src sent the first bit of a type-P packet to Dst at wire-time T 268 and that Dst did not receive that packet (within the loss threshold 269 waiting time, Tmax). 271 2.5. Discussion: 273 Thus, Type-P-One-way-Packet-Loss is 0 exactly when Type-P-One-way- 274 Delay is a finite value, and it is 1 exactly when Type-P-One-way- 275 Delay is undefined. 277 The following issues are likely to come up in practice: 279 + A given methodology will have to include a way to distinguish 280 between a packet loss and a very large (but finite) delay. As noted 281 by Mahdavi and Paxson [RFC2678], simple upper bounds (such as the 255 282 seconds theoretical upper bound on the lifetimes of IP packets 283 [RFC0791]) could be used, but good engineering, including an 284 understanding of packet lifetimes, will be needed in practice. 285 {Comment: Note that, for many applications of these metrics, there 286 may be no harm in treating a large delay as packet loss. An audio 287 playback packet, for example, that arrives only after the playback 288 point may as well have been lost. See section 4.1.1 of [RFC6703] for 289 examination of unusual packet delays and application performance 290 estimation.} 292 + If the packet arrives, but is corrupted, then it is counted as 293 lost. {Comment: one is tempted to count the packet as received since 294 corruption and packet loss are related but distinct phenomena. If 295 the IP header is corrupted, however, one cannot be sure about the 296 source or destination IP addresses and is thus on shaky grounds about 297 knowing that the corrupted received packet corresponds to a given 298 sent test packet. Similarly, if other parts of the packet needed by 299 the methodology to know that the corrupted received packet 300 corresponds to a given sent test packet, then such a packet would 301 have to be counted as lost. Counting these packets as lost but 302 packet with corruption in other parts of the packet as not lost would 303 be inconsistent.} Section 15 of [RFC2330] defines the "standard- 304 formed" packet which is applicable to all metrics. Note: At this 305 time, the definition of standard-formed packets only applies to IPv4, 306 but also see [I-D.morton-ippm-2330-stdform-typep]. 308 + If the packet is duplicated along the path (or paths) so that 309 multiple non-corrupt copies arrive at the destination, then the 310 packet is counted as received. 312 + If the packet is fragmented and if, for whatever reason, reassembly 313 does not occur, then the packet will be deemed lost. 315 2.6. Methodologies: 317 As with other Type-P-* metrics, the detailed methodology will depend 318 on the Type-P (e.g., protocol number, UDP/TCP port number, size, 319 Differentiated Services (DS) Field [RFC2780])). 321 Generally, for a given Type-P, one possible methodology would proceed 322 as follows: 324 + Arrange that Src and Dst have clocks that are synchronized with 325 each other. The degree of synchronization is a parameter of the 326 methodology, and depends on the threshold used to determine loss (see 327 below). 329 + At the Src host, select Src and Dst IP addresses, and form a test 330 packet of Type-P with these addresses. 332 + At the Dst host, arrange to receive the packet. 334 + At the Src host, place a timestamp in the prepared Type-P packet, 335 and send it towards Dst (ideally minimizing time before sending). 337 + If the packet arrives within a reasonable period of time, the one- 338 way packet-loss is taken to be zero (and take a timestamp as soon as 339 possible upon the receipt of the packet). 341 + If the packet fails to arrive within a reasonable period of time, 342 Tmax, the one-way packet-loss is taken to be one. Note that the 343 threshold of "reasonable" here is a parameter of the metric. 345 {Comment: The definition of reasonable is intentionally vague, and is 346 intended to indicate a value "Th" so large that any value in the 347 closed interval [Th-delta, Th+delta] is an equivalent threshold for 348 loss. Here, delta encompasses all error in clock synchronization and 349 timestamp acquisition and assignment along the measured path. If 350 there is a single value, Tmax, after which the packet must be counted 351 as lost, then we reintroduce the need for a degree of clock 352 synchronization similar to that needed for one-way delay, and 353 virtually all practical measurement systems combine methods for delay 354 and loss. Therefore, if a measure of packet loss parameterized by a 355 specific non-huge "reasonable" time-out value is needed, one can 356 always measure one-way delay and see what percentage of packets from 357 a given stream exceed a given time-out value. This point is examined 358 in detail in [RFC6703], including analysis preferences to assign 359 undefined delay to packets that fail to arrive with the difficulties 360 emerging from the informal "infinite delay" assignment, and an 361 estimation of an upper bound on waiting time for packets in transit. 362 Further, enforcing a specific constant waiting time on stored 363 singletons of one-way delay is compliant with this specification and 364 may allow the results to serve more than one reporting audience.} 366 Issues such as the packet format, the means by which Dst knows when 367 to expect the test packet, and the means by which Src and Dst are 368 synchronized are outside the scope of this document. {Comment: We 369 plan to document elsewhere our own work in describing such more 370 detailed implementation techniques and we encourage others to as 371 well.} 373 2.7. Errors and Uncertainties: 375 The description of any specific measurement method should include an 376 accounting and analysis of various sources of error or uncertainty. 377 The Framework document provides general guidance on this point. 379 For loss, there are three sources of error: 381 + Synchronization between clocks on Src and Dst. 383 + The packet-loss threshold (which is related to the synchronization 384 between clocks). 386 + Resource limits in the network interface or software on the 387 receiving instrument. 389 The first two sources are interrelated and could result in a test 390 packet with finite delay being reported as lost. Type-P-One-way- 391 Packet-Loss is 1 if the test packet does not arrive, or if it does 392 arrive and the difference between Src timestamp and Dst timestamp is 393 greater than the "reasonable period of time", or loss threshold. If 394 the clocks are not sufficiently synchronized, the loss threshold may 395 not be "reasonable" - the packet may take much less time to arrive 396 than its Src timestamp indicates. Similarly, if the loss threshold 397 is set too low, then many packets may be counted as lost. The loss 398 threshold must be high enough, and the clocks synchronized well 399 enough so that a packet that arrives is rarely counted as lost. (See 400 the discussions in the previous two sections.) 402 Since the sensitivity of packet loss measurement alone to lack of 403 clock synchronization is less than for delay, we refer the reader to 404 the treatment of synchronization errors in the One-way Delay metric 405 [RFC2330] for more details. 407 The last source of error, resource limits, cause the packet to be 408 dropped by the measurement instrument, and counted as lost when in 409 fact the network delivered the packet in reasonable time. 411 The measurement instruments should be calibrated such that the loss 412 threshold is reasonable for application of the metrics and the clocks 413 are synchronized enough so the loss threshold remains reasonable. 415 In addition, the instruments should be checked to ensure the that the 416 possibility a packet arrives at the network interface, but is lost 417 due to congestion on the interface or to other resource exhaustion 418 (e.g., buffers) on the instrument is low. 420 2.8. Reporting the metric: 422 The calibration and context in which the metric is measured MUST be 423 carefully considered, and SHOULD always be reported along with metric 424 results. We now present four items to consider: Type-P of the test 425 packets, the loss threshold, instrument calibration, and the path 426 traversed by the test packets. This list is not exhaustive; any 427 additional information that could be useful in interpreting 428 applications of the metrics should also be reported (see [RFC6703] 429 for extensive discussion of reporting considerations for different 430 audiences). 432 2.8.1. Type-P 434 As noted in the Framework document, section 13 of [RFC2330], the 435 value of the metric may depend on the type of IP packets used to make 436 the measurement, or "Type-P". The value of Type-P-One-way-Delay 437 could change if the protocol (UDP or TCP), port number, size, or 438 arrangement for special treatment (e.g., IP DS Field [RFC2780], ECN 439 [RFC3168], or RSVP) changes. Additional packet distinctions 440 identified in future extensions of the Type-P definition will apply. 441 The exact Type-P used to make the measurements MUST be accurately 442 reported. 444 2.8.2. Loss Threshold 446 The threshold, Tmax, (or methodology to distinguish) between a large 447 finite delay and loss MUST be reported. 449 2.8.3. Calibration Results 451 The degree of synchronization between the Src and Dst clocks MUST be 452 reported. If possible, possibility that a test packet that arrives 453 at the Dst network interface is reported as lost due to resource 454 exhaustion on Dst SHOULD be reported. 456 2.8.4. Path 458 Finally, the path traversed by the packet SHOULD be reported, if 459 possible. In general it is impractical to know the precise path a 460 given packet takes through the network. The precise path may be 461 known for certain Type-P on short or stable paths. If Type-P 462 includes the record route (or loose-source route) option in the IP 463 header, and the path is short enough, and all routers* on the path 464 support record (or loose-source) route, then the path will be 465 precisely recorded. This is impractical because the route must be 466 short enough, many routers do not support (or are not configured for) 467 record route, and use of this feature would often artificially worsen 468 the performance observed by removing the packet from common-case 469 processing. However, partial information is still valuable context. 470 For example, if a host can choose between two links* (and hence two 471 separate routes from Src to Dst), then the initial link used is 472 valuable context. {Comment: Backbone path selection services come and 473 go. A historical example was Merit's NetNow setup, where a Src on 474 one NAP can reach a Dst on another NAP by either of several different 475 backbone networks.} 477 3. A Definition for Samples of One-way Packet Loss 479 Given the singleton metric Type-P-One-way-Packet-Loss, we now define 480 one particular sample of such singletons. The idea of the sample is 481 to select a particular binding of the parameters Src, Dst, and Type- 482 P, then define a sample of values of parameter T. The means for 483 defining the values of T is to select a beginning time T0, a final 484 time Tf, and an average rate lambda, then define a pseudo-random 485 Poisson process of rate lambda, whose values fall between T0 and Tf. 486 The time interval between successive values of T will then average 1/ 487 lambda. 489 Note that Poisson sampling is only one way of defining a sample. 490 Poisson has the advantage of limiting bias, but other methods of 491 sampling will be appropriate for different situations. For example, 492 a truncated Poisson distribution may be needed to avoid reactive 493 network state changes during intervals of inactivity, see section 4.6 494 of [RFC7312]. Sometimes, the goal is sampling with a known bias, and 495 [RFC3432] describes a method for periodic sampling with random start 496 times. 498 3.1. Metric Name: 500 Type-P-One-way-Packet-Loss-Poisson-Stream 502 3.2. Metric Parameters: 504 + Src, the IP address of a host 506 + Dst, the IP address of a host 508 + T0, a time 510 + Tf, a time 512 + Tmax, a loss threshold waiting time 514 + lambda, a rate in reciprocal seconds 516 3.3. Metric Units: 518 A sequence of pairs; the elements of each pair are: 520 + T, a time, and 522 + L, either a zero or a one 523 The values of T in the sequence are monotonic increasing. Note that 524 T would be a valid parameter to Type-P-One-way-Packet-Loss, and that 525 L would be a valid value of Type-P-One-way-Packet-Loss. 527 3.4. Definition: 529 Given T0, Tf, and lambda, we compute a pseudo-random Poisson process 530 beginning at or before T0, with average arrival rate lambda, and 531 ending at or after Tf. Those time values greater than or equal to T0 532 and less than or equal to Tf are then selected. At each of the times 533 in this process, we obtain the value of Type-P-One-way-Packet-Loss at 534 this time. The value of the sample is the sequence made up of the 535 resulting pairs. If there are no such pairs, the 536 sequence is of length zero and the sample is said to be empty. 538 3.5. Discussion: 540 The reader should be familiar with the in-depth discussion of Poisson 541 sampling in the Framework document [RFC2330], which includes methods 542 to compute and verify the pseudo-random Poisson process. 544 We specifically do not constrain the value of lambda, except to note 545 the extremes. If the rate is too large, then the measurement traffic 546 will perturb the network, and itself cause congestion. If the rate 547 is too small, then you might not capture interesting network 548 behavior. {Comment: We expect to document our experiences with, and 549 suggestions for, lambda elsewhere, culminating in a "best current 550 practices" document.} 552 Since a pseudo-random number sequence is employed, the sequence of 553 times, and hence the value of the sample, is not fully specified. 554 Pseudo-random number generators of good quality will be needed to 555 achieve the desired qualities. 557 The sample is defined in terms of a Poisson process both to avoid the 558 effects of self-synchronization and also capture a sample that is 559 statistically as unbiased as possible. The Poisson process is used 560 to schedule the loss measurements. The test packets will generally 561 not arrive at Dst according to a Poisson distribution, since they are 562 influenced by the network. Time-slotted links described in section 563 3.4 [RFC7312] can greatly modify the sample characteristics. The 564 main concern is that un-biased packet streams with randomized inter- 565 packet time intervals will be converted to some new distribution 566 after encountering a time-slotted links, possibly with strong 567 periodic characteristics instead. 569 {Comment: there is, of course, no claim that real Internet traffic 570 arrives according to a Poisson arrival process. 572 It is important to note that, in contrast to this metric, loss ratios 573 observed by transport connections do not reflect unbiased samples. 574 For example, TCP transmissions both (1) occur in bursts, which can 575 induce loss due to the burst volume that would not otherwise have 576 been observed, and (2) adapt their transmission rate in an attempt to 577 minimize the loss ratio observed by the connection.} 579 All the singleton Type-P-One-way-Packet-Loss metrics in the sequence 580 will have the same values of Src, Dst, and Type-P. 582 Note also that, given one sample that runs from T0 to Tf, and given 583 new time values T0' and Tf' such that T0 <= T0' <= Tf' <= Tf, the 584 subsequence of the given sample whose time values fall between T0' 585 and Tf' are also a valid Type-P-One-way-Packet-Loss-Poisson-Stream 586 sample. 588 3.6. Methodologies: 590 The methodologies follow directly from: 592 + the selection of specific times, using the specified Poisson 593 arrival process, and 595 + the methodologies discussion already given for the singleton Type- 596 P-One-way-Packet-Loss metric. 598 Care must be given to correctly handle out-of-order arrival of test 599 packets; it is possible that the Src could send one test packet at 600 TS[i], then send a second one (later) at TS[i+1], while the Dst could 601 receive the second test packet at TR[i+1], and then receive the first 602 one (later) at TR[i]. Metrics for reordering may be found in 603 [RFC4737]. 605 3.7. Errors and Uncertainties: 607 In addition to sources of errors and uncertainties associated with 608 methods employed to measure the singleton values that make up the 609 sample, care must be given to analyze the accuracy of the Poisson 610 arrival process of the wire-times of the sending of the test packets. 611 Problems with this process could be caused by several things, 612 including problems with the pseudo-random number techniques used to 613 generate the Poisson arrival process. The Framework document shows 614 how to use the Anderson-Darling test to verify the accuracy of the 615 Poisson process over small time frames. {Comment: The goal is to 616 ensure that the test packets are sent "close enough" to a Poisson 617 schedule, and avoid periodic behavior.} 619 3.8. Reporting the metric: 621 The calibration and context for the underlying singletons MUST be 622 reported along with the stream. (See "Reporting the metric" for 623 Type-P-One-way-Packet-Loss.) 625 4. Some Statistics Definitions for One-way Packet Loss 627 Given the sample metric Type-P-One-way-Packet-Loss-Poisson-Stream, we 628 now offer several statistics of that sample. These statistics are 629 offered mostly to be illustrative of what could be done. See 630 [RFC6703] for additional discussion of statistics that are relevant 631 to different audiences. 633 4.1. Type-P-One-way-Packet Loss-Ratio 635 Given a Type-P-One-way-Packet-Loss-Poisson-Stream, the average of all 636 the L values in the Stream is the ratio of losses to total packets in 637 the stream. In addition, the Type-P-One-way-Packet-Loss-Ratio is 638 undefined if the sample is empty. 640 Example: suppose we take a sample and the results are: 642 Stream1 = < 644 646 648 650 652 654 > 656 Then the average of loss results would be 0.2, the loss ratio. 658 Note that, since healthy Internet paths should be operating at loss 659 ratios below 1% (particularly if high delay-bandwidth products are to 660 be sustained), the sample sizes needed might be larger than one would 661 like. Thus, for example, if one wants to discriminate between 662 various fractions of 1% over one-minute periods, then several hundred 663 samples per minute might be needed. This would result in larger 664 values of lambda than one would ordinarily want. 666 Note that although the loss threshold should be set such that any 667 errors in loss are not significant, if the possibility that a packet 668 which arrived is counted as lost due to resource exhaustion is 669 significant compared to the loss ratio of interest, Type-P-One-way- 670 Packet-Loss-Ratio will be meaningless. 672 5. Security Considerations 674 Conducting Internet measurements raises both security and privacy 675 concerns. This memo does not specify an implementation of the 676 metrics, so it does not directly affect the security of the Internet 677 nor of applications which run on the Internet. However, 678 implementations of these metrics must be mindful of security and 679 privacy concerns. 681 There are two types of security concerns: potential harm caused by 682 the measurements, and potential harm to the measurements. The 683 measurements could cause harm because they are active, and inject 684 packets into the network. The measurement parameters MUST be 685 carefully selected so that the measurements inject trivial amounts of 686 additional traffic into the networks they measure. If they inject 687 "too much" traffic, they can skew the results of the measurement, and 688 in extreme cases cause congestion and denial of service. 690 The measurements themselves could be harmed by routers giving 691 measurement traffic a different priority than "normal" traffic, or by 692 an attacker injecting artificial measurement traffic. If routers can 693 recognize measurement traffic and treat it separately, the 694 measurements will not reflect actual user traffic. If an attacker 695 injects artificial traffic that is accepted as legitimate, the loss 696 ratio will be artificially lowered. Therefore, the measurement 697 methodologies SHOULD include appropriate techniques to reduce the 698 probability measurement traffic can be distinguished from "normal" 699 traffic. Authentication techniques, such as digital signatures, may 700 be used where appropriate to guard against injected traffic attacks. 702 When considering privacy of those involved in measurement or those 703 whose traffic is measured, the sensitive information available to 704 potential observers is greatly reduced when using active techniques 705 which are within this scope of work. Passive observations of user 706 traffic for measurement purposes raise many privacy issues. We refer 707 the reader to the privacy considerations described in the Large Scale 708 Measurement of Broadband Performance (LMAP) Framework 709 [I-D.ietf-lmap-framework], which covers active and passive 710 techniques. 712 Collecting measurements or using measurement results for 713 reconnaissance to assist in subsequent system attacks is quite 714 common. Access to measurement results, or control of the measurement 715 systems to perform reconnaissance should be guarded against. See 716 Section 7 of [I-D.ietf-lmap-framework] (security considerations of 717 the LMAP Framework) for system requirements that help to avoid 718 measurement system compromise. 720 6. Acknowledgements 722 For [RFC2680], thanks are due to Matt Mathis for encouraging this 723 work and for calling attention on so many occasions to the 724 significance of packet loss. Thanks are due also to Vern Paxson for 725 his valuable comments on early drafts, and to Garry Couch and Will 726 Leland for several useful suggestions. 728 For RFC 2680 bis, thanks to Joachim Fabini, Ruediger Geib, Nalini 729 Elkins, and Barry Constantine for sharing their measurement 730 experience as part of their careful reviews. Brian Carpenter and 731 Scott Bradner provided useful feedback at IETF Last Call. 733 7. Changes from RFC 2680 735 Note: This section's placement currently preserves minimal 736 differences between this memo and RFC 2680. The RFC Editor should 737 place this section in an appropriate place. 739 The text above constitutes RFC 2680 bis proposed for advancement on 740 the IETF Standards Track. 742 [RFC7290] provides the test plan and results supporting [RFC2680] 743 advancement along the standards track, according to the process in 744 [RFC6576]. The conclusions of [RFC7290] list four minor 745 modifications for inclusion: 747 1. Section 6.2.3 of [RFC7290] asserts that the assumption of post- 748 processing to enforce a constant waiting time threshold is 749 compliant, and that the text of the RFC should be revised 750 slightly to include this point. The applicability of post- 751 processing was added in the last list item of section 2.6, above. 753 2. Section 6.5 of [RFC7290] indicates that Type-P-One-way-Packet- 754 Loss-Average statistic is more commonly called Packet Loss Ratio, 755 so it is re-named in RFC2680bis (this small discrepancy does not 756 affect candidacy for advancement) The re-naming was implemented 757 in section 4.1, above. 759 3. The IETF has reached consensus on guidance for reporting metrics 760 in [RFC6703], and this memo should be referenced in RFC2680bis to 761 incorporate recent experience where appropriate. This reference 762 was added in the last list item of section 2.6, in section 2.8, 763 and in section 4 above. 765 4. There are currently two errata with status "Verified" and "Held 766 for document update" for [RFC2680], and these minor revisions 767 were incorporated in section 1 and section 2.7. 769 A number of updates to the [RFC2680] text have been implemented in 770 the text, to reference key IPPM RFCs that were approved after 771 [RFC2680] (see sections 3 and 3.6, above), and to address comments on 772 the IPPM mailing list describing current conditions and experience. 774 1. Near the end of section 1.1, update of a network example using 775 ATM and clarification of TCP's affect on queue occupation and 776 importance of one-way delay measurement. 778 2. Clarification of the definition of "resolution" in section 1.2. 780 3. Explicit inclusion of the maximum waiting time input parameter 781 in sections 2.2, 2.4, and 3.2, reflecting recognition of this 782 parameter in more recent RFCs and ITU-T Recommendation Y.1540. 784 4. Addition of reference to RFC 6703 in the discussion of packet 785 life time and application timeouts in section 2.5. 787 5. Replaced "precedence" with updated terminology (DS Field) in 2.6 788 and 2.8.1 (with reference). 790 6. Added parenthetical guidance on minimizing interval between 791 timestamp placement to send time or reception time in section 792 2.6. Also, the text now recognizes the timestamp acquisition 793 process and that practical systems measure both delay and loss 794 (thus require the max waiting time parameter). 796 7. Added reference to RFC 3432 Periodic sampling alongside Poisson 797 sampling in section 3, and also noting that a truncated Poisson 798 distribution may be needed with modern networks as described in 799 the IPPM Framework update, [RFC7312]. 801 8. Recognition that Time-slotted links described in [RFC7312] can 802 greatly modify the sample characteristics, in section 3.5. 804 9. Add reference to RFC 4737 Reordering metric in the related 805 discussion of section 3.6, Methodologies. 807 10. Expanded and updated the material on Privacy, and added cautions 808 on use of measurements for reconnaissance in section 5, Security 809 Considerations. 811 Section 5.4.4 of [RFC6390] suggests a common template for performance 812 metrics partially derived from previous IPPM and BMWG RFCs, but also 813 contains some new items. All of the [RFC6390] Normative points are 814 covered, but not quite in the same section names or orientation. 815 Several of the Informative points are covered. Maintaining the 816 familiar outline of IPPM literature has value and minimizes 817 unnecessary differences between this revised RFC and current/future 818 IPPM RFCs. 820 8. IANA Considerations 822 This memo makes no requests of IANA. 824 9. References 826 9.1. Normative References 828 [RFC0791] Postel, J., "Internet Protocol", STD 5, RFC 791, 829 DOI 10.17487/RFC0791, September 1981, 830 . 832 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 833 Requirement Levels", BCP 14, RFC 2119, 834 DOI 10.17487/RFC2119, March 1997, 835 . 837 [RFC2330] Paxson, V., Almes, G., Mahdavi, J., and M. Mathis, 838 "Framework for IP Performance Metrics", RFC 2330, 839 DOI 10.17487/RFC2330, May 1998, 840 . 842 [RFC2678] Mahdavi, J. and V. Paxson, "IPPM Metrics for Measuring 843 Connectivity", RFC 2678, DOI 10.17487/RFC2678, September 844 1999, . 846 [RFC2679] Almes, G., Kalidindi, S., and M. Zekauskas, "A One-way 847 Delay Metric for IPPM", RFC 2679, DOI 10.17487/RFC2679, 848 September 1999, . 850 [RFC2680] Almes, G., Kalidindi, S., and M. Zekauskas, "A One-way 851 Packet Loss Metric for IPPM", RFC 2680, 852 DOI 10.17487/RFC2680, September 1999, 853 . 855 [RFC2780] Bradner, S. and V. Paxson, "IANA Allocation Guidelines For 856 Values In the Internet Protocol and Related Headers", 857 BCP 37, RFC 2780, DOI 10.17487/RFC2780, March 2000, 858 . 860 [RFC3168] Ramakrishnan, K., Floyd, S., and D. Black, "The Addition 861 of Explicit Congestion Notification (ECN) to IP", 862 RFC 3168, DOI 10.17487/RFC3168, September 2001, 863 . 865 [RFC3432] Raisanen, V., Grotefeld, G., and A. Morton, "Network 866 performance measurement with periodic streams", RFC 3432, 867 DOI 10.17487/RFC3432, November 2002, 868 . 870 [RFC6576] Geib, R., Ed., Morton, A., Fardid, R., and A. Steinmitz, 871 "IP Performance Metrics (IPPM) Standard Advancement 872 Testing", BCP 176, RFC 6576, DOI 10.17487/RFC6576, March 873 2012, . 875 [RFC7312] Fabini, J. and A. Morton, "Advanced Stream and Sampling 876 Framework for IP Performance Metrics (IPPM)", RFC 7312, 877 DOI 10.17487/RFC7312, August 2014, 878 . 880 9.2. Informative References 882 [I-D.ietf-lmap-framework] 883 Eardley, P., Morton, A., Bagnulo, M., Burbridge, T., 884 Aitken, P., and A. Akhter, "A framework for Large-Scale 885 Measurement of Broadband Performance (LMAP)", draft-ietf- 886 lmap-framework-14 (work in progress), April 2015. 888 [I-D.morton-ippm-2330-stdform-typep] 889 Morton, A., Fabini, J., Elkins, N., Ackermann, M., and V. 890 Hegde, "Updates for IPPM's Active Metric Framework: 891 Packets of Type-P and Standard-Formed Packets", draft- 892 morton-ippm-2330-stdform-typep-00 (work in progress), 893 August 2015. 895 [RFC4737] Morton, A., Ciavattone, L., Ramachandran, G., Shalunov, 896 S., and J. Perser, "Packet Reordering Metrics", RFC 4737, 897 DOI 10.17487/RFC4737, November 2006, 898 . 900 [RFC6390] Clark, A. and B. Claise, "Guidelines for Considering New 901 Performance Metric Development", BCP 170, RFC 6390, 902 DOI 10.17487/RFC6390, October 2011, 903 . 905 [RFC6703] Morton, A., Ramachandran, G., and G. Maguluri, "Reporting 906 IP Network Performance Metrics: Different Points of View", 907 RFC 6703, DOI 10.17487/RFC6703, August 2012, 908 . 910 [RFC7290] Ciavattone, L., Geib, R., Morton, A., and M. Wieser, "Test 911 Plan and Results for Advancing RFC 2680 on the Standards 912 Track", RFC 7290, DOI 10.17487/RFC7290, July 2014, 913 . 915 Authors' Addresses 917 Guy Almes 918 Texas A&M 920 Email: almes@acm.org 922 Sunil Kalidindi 923 Ixia 925 Email: skalidindi@ixiacom.com 927 Matt Zekauskas 928 Internet2 930 Email: matt@internet2.edu 932 Al Morton (editor) 933 AT&T Labs 934 200 Laurel Avenue South 935 Middletown, NJ 07748 936 USA 938 Phone: +1 732 420 1571 939 Fax: +1 732 368 1192 940 Email: acmorton@att.com 941 URI: http://home.comcast.net/~acmacm/