idnits 2.17.1 draft-ietf-ippm-alt-mark-02.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 document seems to lack a both a reference to RFC 2119 and the recommended RFC 2119 boilerplate, even if it appears to use RFC 2119 keywords. RFC 2119 keyword, line 661: '...correlation mechanism SHOULD be in use...' RFC 2119 keyword, line 716: '... SHOULD provide a way to configure t...' Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (October 28, 2016) is 2731 days in the past. Is this intentional? Checking references for intended status: Experimental ---------------------------------------------------------------------------- == Missing Reference: 'RFC5905' is mentioned on line 684, but not defined == Missing Reference: 'IEEE1588' is mentioned on line 685, but not defined ** Obsolete normative reference: RFC 2679 (Obsoleted by RFC 7679) ** Obsolete normative reference: RFC 2680 (Obsoleted by RFC 7680) == Outdated reference: A later version (-04) exists of draft-bryant-mpls-rfc6374-sfl-01 == Outdated reference: A later version (-05) exists of draft-bryant-mpls-sfl-framework-02 == Outdated reference: A later version (-01) exists of draft-fioccola-ippm-alt-mark-active-00 == Outdated reference: A later version (-12) exists of draft-ietf-bier-mpls-encapsulation-05 == Outdated reference: A later version (-15) exists of draft-ietf-bier-pmmm-oam-00 == Outdated reference: A later version (-07) exists of draft-ietf-mpls-flow-ident-02 == Outdated reference: A later version (-02) exists of draft-ooamdt-rtgwg-ooam-requirement-01 Summary: 3 errors (**), 0 flaws (~~), 10 warnings (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network Working Group G. Fioccola, Ed. 3 Internet-Draft A. Capello, Ed. 4 Intended status: Experimental M. Cociglio 5 Expires: May 1, 2017 L. Castaldelli 6 Telecom Italia 7 M. Chen, Ed. 8 L. Zheng, Ed. 9 Huawei Technologies 10 G. Mirsky, Ed. 11 Ericsson 12 T. Mizrahi, Ed. 13 Marvell 14 October 28, 2016 16 Alternate Marking method for passive performance monitoring 17 draft-ietf-ippm-alt-mark-02 19 Abstract 21 This document describes a passive method to perform packet loss, 22 delay and jitter measurements on live traffic. This method is based 23 on Alternate Marking (Coloring) technique. A report on the 24 operational experiment done at Telecom Italia is explained in order 25 to give an example and show the method applicability. This technique 26 can be applied in various situations as detailed in this document. 28 Status of This Memo 30 This Internet-Draft is submitted in full conformance with the 31 provisions of BCP 78 and BCP 79. 33 Internet-Drafts are working documents of the Internet Engineering 34 Task Force (IETF). Note that other groups may also distribute 35 working documents as Internet-Drafts. The list of current Internet- 36 Drafts is at http://datatracker.ietf.org/drafts/current/. 38 Internet-Drafts are draft documents valid for a maximum of six months 39 and may be updated, replaced, or obsoleted by other documents at any 40 time. It is inappropriate to use Internet-Drafts as reference 41 material or to cite them other than as "work in progress." 43 This Internet-Draft will expire on May 1, 2017. 45 Copyright Notice 47 Copyright (c) 2016 IETF Trust and the persons identified as the 48 document authors. All rights reserved. 50 This document is subject to BCP 78 and the IETF Trust's Legal 51 Provisions Relating to IETF Documents 52 (http://trustee.ietf.org/license-info) in effect on the date of 53 publication of this document. Please review these documents 54 carefully, as they describe your rights and restrictions with respect 55 to this document. Code Components extracted from this document must 56 include Simplified BSD License text as described in Section 4.e of 57 the Trust Legal Provisions and are provided without warranty as 58 described in the Simplified BSD License. 60 Table of Contents 62 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 63 2. Overview of the method . . . . . . . . . . . . . . . . . . . 4 64 3. Detailed description of the method . . . . . . . . . . . . . 5 65 3.1. Packet loss measurement . . . . . . . . . . . . . . . . . 5 66 3.1.1. Timing aspects . . . . . . . . . . . . . . . . . . . 9 67 3.2. One-way delay measurement . . . . . . . . . . . . . . . . 10 68 3.2.1. Single marking methodology . . . . . . . . . . . . . 10 69 3.2.2. Mean delay . . . . . . . . . . . . . . . . . . . . . 12 70 3.2.3. Double marking methodology . . . . . . . . . . . . . 12 71 3.3. Delay variation measurement . . . . . . . . . . . . . . . 13 72 4. Considerations . . . . . . . . . . . . . . . . . . . . . . . 14 73 4.1. Synchronization . . . . . . . . . . . . . . . . . . . . . 14 74 4.2. Data Correlation . . . . . . . . . . . . . . . . . . . . 14 75 4.3. Packet Re-ordering . . . . . . . . . . . . . . . . . . . 15 76 5. Implementation and deployment . . . . . . . . . . . . . . . . 16 77 5.1. Report on the operational experiment at Telecom Italia . 16 78 5.1.1. Coloring the packets . . . . . . . . . . . . . . . . 18 79 5.1.2. Counting the packets . . . . . . . . . . . . . . . . 19 80 5.1.3. Collecting data and calculating packet loss . . . . . 20 81 5.1.4. Metric transparency . . . . . . . . . . . . . . . . . 20 82 5.2. IP flow performance measurement (IPFPM) . . . . . . . . . 21 83 5.3. Performance Measurement Marking Method in BIER Domain . . 21 84 5.4. Overlay OAM Passive Performance Measurement . . . . . . . 21 85 5.5. RFC6374 Use Case . . . . . . . . . . . . . . . . . . . . 21 86 5.6. Application to active performance measurement . . . . . . 22 87 6. Hybrid measurement . . . . . . . . . . . . . . . . . . . . . 22 88 7. Compliance with RFC6390 guidelines . . . . . . . . . . . . . 22 89 8. Security Considerations . . . . . . . . . . . . . . . . . . . 24 90 9. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . 25 91 10. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 26 92 11. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 26 93 12. References . . . . . . . . . . . . . . . . . . . . . . . . . 26 94 12.1. Normative References . . . . . . . . . . . . . . . . . . 26 95 12.2. Informative References . . . . . . . . . . . . . . . . . 26 96 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 29 98 1. Introduction 100 Nowadays, most of the traffic in Service Providers' networks carries 101 real time content. These contents are highly sensitive to packet 102 loss [RFC2680], while interactive contents are sensitive to delay 103 [RFC2679], and jitter [RFC3393]. 105 In view of this scenario, Service Providers need methodologies and 106 tools to monitor and measure network performances with an adequate 107 accuracy, in order to constantly control the quality of experience 108 perceived by their customers. On the other hand, performance 109 monitoring provides useful information for improving network 110 management (e.g. isolation of network problems, troubleshooting, 111 etc.). 113 A lot of work related to OAM, that includes also performance 114 monitoring techniques, has been done by Standards Developing 115 Organizations(SDOs):: [RFC7276] provides a good overview of existing 116 OAM mechanisms defined in IETF, ITU-T and IEEE. Considering IETF, a 117 lot of work has been done on fault detection and connectivity 118 verification, while a minor effort has been dedicated so far to 119 performance monitoring. The IPPM WG has defined standard metrics to 120 measure network performance; however, the methods developed in this 121 WG mainly refer to focus on active measurement techniques. More 122 recently, the MPLS WG has defined mechanisms for measuring packet 123 loss, one-way and two-way delay, and delay variation in MPLS 124 networks[RFC6374], but their applicability to passive measurements 125 has some limitations, especially for pure connection-less networks. 127 The lack of adequate tools to measure packet loss with the desired 128 accuracy drove an effort to design a new method for the performance 129 monitoring of live traffic, possibly easy to implement and deploy. 130 The effort led to the method described in this document: basically, 131 it is a passive performance monitoring technique, potentially 132 applicable to any kind of packet based traffic, including Ethernet, 133 IP, and MPLS, both unicast and multicast. The method addresses 134 primarily packet loss measurement, but it can be easily extended to 135 one-way delay and delay variation measurements as well. 137 The method has been explicitly designed for passive measurements but 138 it can also be used with active probes. Passive measurements are 139 usually more easily understood by customers and provide a much better 140 accuracy, especially for packet loss measurements. 142 This document is organized as follows: 144 o Section 2 gives an overview of the method, including a comparison 145 with different measurement strategies; 147 o Section 3 describes the method in detail; 149 o Section 4 reports considerations about synchronization, data 150 correlation and packet re-ordering; 152 o Section 5 reports examples of implementation and deployment of the 153 method. Furthermore the operational experiment done at Telecom 154 Italia is described; 156 o Section 8 includes some security aspects; 158 o Section 9 finally summarizes some concluding remarks. 160 2. Overview of the method 162 In order to perform packet loss measurements on a live traffic flow, 163 different approaches exist. The most intuitive one consists in 164 numbering the packets, so that each router that receives the flow can 165 immediately detect a packet missing. This approach, though very 166 simple in theory, is not simple to achieve: it requires the insertion 167 of a sequence number into each packet and the devices must be able to 168 extract the number and check it in real time. Such a task can be 169 difficult to implement on live traffic: if UDP is used as the 170 transport protocol, the sequence number is not available; on the 171 other hand, if a higher layer sequence number (e.g. in the RTP 172 header) is used, extracting that information from each packet and 173 process it in real time could overload the device. 175 An alternate approach is to count the number of packets sent on one 176 end, the number of packets received on the other end, and to compare 177 the two values. This operation is much simpler to implement, but 178 requires that the devices performing the measurement are in sync: in 179 order to compare two counters it is required that they refer exactly 180 to the same set of packets. Since a flow is continuous and cannot be 181 stopped when a counter has to be read, it could be difficult to 182 determine exactly when to read the counter. A possible solution to 183 overcome this problem is to virtually split the flow in consecutive 184 blocks by inserting periodically a delimiter so that each counter 185 refers exactly to the same block of packets. The delimiter could be 186 for example a special packet inserted artificially into the flow. 187 However, delimiting the flow using specific packets has some 188 limitations. First, it requires generating additional packets within 189 the flow and requires the equipment to be able to process those 190 packets. In addition, the method is vulnerable to out of order 191 reception of delimiting packets and, to a lesser extent, to their 192 loss. 194 The method proposed in this document follows the second approach, but 195 it doesn't use additional packets to virtually split the flow in 196 blocks. Instead, it "colors" the packets so that the packets 197 belonging to the same block will have the same color, whilst 198 consecutive blocks will have different colors. Each change of color 199 represents a sort of auto-synchronization signal that guarantees the 200 consistency of measurements taken by different devices along the 201 path. 203 Figure 1 represents a very simple network and shows how the method 204 can be used to measure packet loss on different network segments: by 205 enabling the measurement on several interfaces along the path, it is 206 possible to perform link monitoring, node monitoring or end-to-end 207 monitoring. The method is flexible enough to measure packet loss on 208 any segment of the network and can be used to isolate the faulty 209 element. 211 Traffic flow 212 ========================================================> 213 +------+ +------+ +------+ +------+ 214 ---<> R1 <>-----<> R2 <>-----<> R3 <>-----<> R4 <>--- 215 +------+ +------+ +------+ +------+ 216 . . . . . . 217 . . . . . . 218 . <------> <-------> . 219 . Node Packet Loss Link Packet Loss . 220 . . 221 <---------------------------------------------------> 222 End-to-End Packet loss 224 Figure 1: Available measurements 226 3. Detailed description of the method 228 This section describes in detail how the method operate. A special 229 emphasis is given to the measurement of packet loss, that represents 230 the core application of the method, but applicability to delay and 231 jitter measurements is also considered. 233 3.1. Packet loss measurement 235 The basic idea is to virtually split traffic flows into consecutive 236 blocks: each block represents a measurable entity unambiguously 237 recognizable by all network devices along the path. By counting the 238 number of packets in each block and comparing the values measured by 239 different network devices along the path, it is possible to measure 240 packet loss occurred in any single block between any two points. 242 As discussed in the previous section, a simple way to create the 243 blocks is to "color" the traffic (two colors are sufficient) so that 244 packets belonging to different consecutive blocks will have different 245 colors. Whenever the color changes, the previous block terminates 246 and the new one begins. Hence, all the packets belonging to the same 247 block will have the same color and packets of different consecutive 248 blocks will have different colors. The number of packets in each 249 block depends on the criterion used to create the blocks: if the 250 color is switched after a fixed number of packets, then each block 251 will contain the same number of packets (except for any losses); but 252 if the color is switched according to a fixed timer, then the number 253 of packets may be different in each block depending on the packet 254 rate. 256 The following figure shows how a flow looks like when it is split in 257 traffic blocks with colored packets. 259 A: packet with A coloring 260 B: packet with B coloring 262 | | | | | 263 | | Traffic flow | | 264 -------------------------------------------------------------------> 265 BBBBBBB AAAAAAAAAAA BBBBBBBBBBB AAAAAAAAAAA BBBBBBBBBBB AAAAAAA 266 -------------------------------------------------------------------> 267 ... | Block 5 | Block 4 | Block 3 | Block 2 | Block 1 268 | | | | | 270 Figure 2: Traffic coloring 272 Figure 3 shows how the method can be used to measure link packet loss 273 between two adjacent nodes. 275 Referring to the figure, let's assume we want to monitor the packet 276 loss on the link between two routers: router R1 and router R2. 277 According to the method, the traffic is colored alternatively with 278 two different colors, A and B. Whenever the color changes, the 279 transition generates a sort of square-wave signal, as depicted in the 280 following figure. 282 Color A ----------+ +-----------+ +---------- 283 | | | | 284 Color B +-----------+ +-----------+ 285 Block n ... Block 3 Block 2 Block 1 286 <---------> <---------> <---------> <---------> <---------> 288 Traffic flow 289 ===========================================================> 290 Color ...AAAAAAAAAAA BBBBBBBBBBB AAAAAAAAAAA BBBBBBBBBBB AAAAAAA... 291 ===========================================================> 293 Figure 3: Computation of link packet loss 295 Traffic coloring could be done by R1 itself or by an upward router. 296 R1 needs two counters, C(A)R1 and C(B)R1, on its egress interface: 297 C(A)R1 counts the packets with color A and C(B)R1 counts those with 298 color B. As long as traffic is colored A, only counter C(A)R1 will 299 be incremented, while C(B)R1 is not incremented; vice versa, when the 300 traffic is colored as B, only C(B)R1 is incremented. C(A)R1 and 301 C(B)R1 can be used as reference values to determine the packet loss 302 from R1 to any other measurement point down the path. Router R2, 303 similarly, will need two counters on its ingress interface, C(A)R2 304 and C(B)R2, to count the packets received on that interface and 305 colored with color A and B respectively. When an A block ends, it is 306 possible to compare C(A)R1 and C(A)R2 and calculate the packet loss 307 within the block; similarly, when the successive B block terminates, 308 it is possible to compare C(B)R1 with C(B)R2, and so on for every 309 successive block. 311 Likewise, by using two counters on R2 egress interface it is possible 312 to count the packets sent out of R2 interface and use them as 313 reference values to calculate the packet loss from R2 to any 314 measurement point down R2. 316 Using a fixed timer for color switching offers a better control over 317 the method: the (time) length of the blocks can be chosen large 318 enough to simplify the collection and the comparison of measures 319 taken by different network devices. It's preferable to read the 320 value of the counters not immediately after the color switch: some 321 packets could arrive out of order and increment the counter 322 associated to the previous block (color), so it is worth waiting for 323 some time. A safe choice is to wait L/2 time units (where L is the 324 duration for each block) after the color switch, to read the still 325 counter of the previous color, so the possibility to read a running 326 counter instead of a still one is minimized. The drawback is that 327 the longer the duration of the block, the less frequent the 328 measurement can be taken. 330 The following table shows how the counters can be used to calculate 331 the packet loss between R1 and R2. The first column lists the 332 sequence of traffic blocks while the other columns contain the 333 counters of A-colored packets and B-colored packets for R1 and R2. 334 In this example, we assume that the values of the counters are reset 335 to zero whenever a block ends and its associated counter has been 336 read: with this assumption, the table shows only relative values, 337 that is the exact number of packets of each color within each block. 338 If the values of the counters were not reset, the table would contain 339 cumulative values, but the relative values could be determined simply 340 by difference from the value of the previous block of the same color. 342 The color is switched on the basis of a fixed timer (not shown in the 343 table), so the number of packets in each block is different. 345 +-------+--------+--------+--------+--------+------+ 346 | Block | C(A)R1 | C(B)R1 | C(A)R2 | C(B)R2 | Loss | 347 +-------+--------+--------+--------+--------+------+ 348 | 1 | 375 | 0 | 375 | 0 | 0 | 349 | | | | | | | 350 | 2 | 0 | 388 | 0 | 388 | 0 | 351 | | | | | | | 352 | 3 | 382 | 0 | 381 | 0 | 1 | 353 | | | | | | | 354 | 4 | 0 | 377 | 0 | 374 | 3 | 355 | | | | | | | 356 | ... | ... | ... | ... | ... | ... | 357 | | | | | | | 358 | n | 0 | 387 | 0 | 387 | 0 | 359 | | | | | | | 360 | n+1 | 379 | 0 | 377 | 0 | 2 | 361 +-------+--------+--------+--------+--------+------+ 363 Table 1: Evaluation of counters for packet loss measurements 365 During an A block (blocks 1, 3 and n+1), all the packets are 366 A-colored, therefore the C(A) counters are incremented to the number 367 seen on the interface, while C(B) counters are zero. Vice versa, 368 during a B block (blocks 2, 4 and n), all the packets are B-colored: 369 C(A) counters are zero, while C(B) counters are incremented. 371 When a block ends (because of color switching) the relative counters 372 stop incrementing and it is possible to read them, compare the values 373 measured on router R1 and R2 and calculate the packet loss within 374 that block. 376 For example, looking at the table above, during the first block 377 (A-colored), C(A)R1 and C(A)R2 have the same value (375), which 378 corresponds to the exact number of packets of the first block (no 379 loss). Also during the second block (B-colored) R1 and R2 counters 380 have the same value (388), which corresponds to the number of packets 381 of the second block (no loss). During blocks three and four, R1 and 382 R2 counters are different, meaning that some packets have been lost: 383 in the example, one single packet (382-381) was lost during block 384 three and three packets (377-374) were lost during block four. 386 The method applied to R1 and R2 can be extended to any other router 387 and applied to more complex networks, as far as the measurement is 388 enabled on the path followed by the traffic flow(s) being observed. 390 3.1.1. Timing aspects 392 This document introduces two color switching method: one is based on 393 fixed number of packet, the other is based on fixed timer. But the 394 method based on fixed timer is preferable because is more 395 deterministic, and will be considered in the rest of the dcoument. 397 By considering the clock error between network devices R1 and R2, 398 they must be synchronized to the same clock reference with an 399 accuracy of +/- L/2 time units, where L is the time duration of the 400 block. So each colored packet can be assigned to the right batch by 401 each router. This is because the minimum time distance between two 402 packets of the same color but belonging to different batches is L 403 time units. 405 In practice, there are also out of order at batch boundaries, 406 strictly related to the delay between measurement points. This means 407 that, without considering clock error, we wait L/2 after color 408 switching to be sure to take a still counter. 410 In summary we need to take into account two contributions: clock 411 error between network devices and the interval we need to wait to 412 avoid out of order because of network delay. 414 The following figure eplains both issues. 416 ...BBBBBBBBB | AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA | BBBBBBBBB... 417 |<======================================>| 418 | L | 419 ...=========>|<==================><==================>|<==========... 420 | L/2 L/2 | 421 |<===>| |<===>| 422 d | | d 423 |<==========================>| 424 available counting interval 426 Figure 4: Timing aspects 428 It is assumed that all network devices are synchronized to a common 429 reference time with an accuracy of +/- A/2. Thus, the difference 430 between the clock values of any two network devices is bounded by A. 432 The guardband d is given by: 434 d = A + D_max - D_min, 436 where A is the clock accuracy, D_max is an upper bound on the network 437 delay between the network devices, and D_min is a lower bound on the 438 delay. 440 The available counting interval is L - 2d that must be > 0. 442 The condition that must be satisfied and is a requirement on the 443 synchronization accuracy is: 445 d < L/2. 447 3.2. One-way delay measurement 449 The same principle used to measure packet loss can be applied also to 450 one-way delay measurement. There are three alternatives, as 451 described hereinafter. 453 3.2.1. Single marking methodology 455 The alternation of colors can be used as a time reference to 456 calculate the delay. Whenever the color changes (that means that a 457 new block has started) a network device can store the timestamp of 458 the first packet of the new block; that timestamp can be compared 459 with the timestamp of the same packet on a second router to compute 460 packet delay. Considering Figure 2, R1 stores a timestamp TS(A1)R1 461 when it sends the first packet of block 1 (A-colored), a timestamp 462 TS(B2)R1 when it sends the first packet of block 2 (B-colored) and so 463 on for every other block. R2 performs the same operation on the 464 receiving side, recording TS(A1)R2, TS(B2)R2 and so on. Since the 465 timestamps refer to specific packets (the first packet of each block) 466 we are sure that timestamps compared to compute delay refer to the 467 same packets. By comparing TS(A1)R1 with TS(A1)R2 (and similarly 468 TS(B2)R1 with TS(B2)R2 and so on) it is possible to measure the delay 469 between R1 and R2. In order to have more measurements, it is 470 possible to take and store more timestamps, referring to other 471 packets within each block. 473 In order to coherently compare timestamps collected on different 474 routers, the network nodes must be in sync. Furthermore, a 475 measurement is valid only if no packet loss occurs and if packet 476 misordering can be avoided, otherwise the first packet of a block on 477 R1 could be different from the first packet of the same block on R2 478 (f.i. if that packet is lost between R1 and R2 or it arrives after 479 the next one). 481 The following table shows how timestamps can be used to calculate the 482 delay between R1 and R2. The first column lists the sequence of 483 blocks while other columns contain the timestamp referring to the 484 first packet of each block on R1 and R2. The delay is computed as a 485 difference between timestamps. For the sake of simplicity, all the 486 values are expressed in milliseconds. 488 +-------+---------+---------+---------+---------+-------------+ 489 | Block | TS(A)R1 | TS(B)R1 | TS(A)R2 | TS(B)R2 | Delay R1-R2 | 490 +-------+---------+---------+---------+---------+-------------+ 491 | 1 | 12.483 | - | 15.591 | - | 3.108 | 492 | | | | | | | 493 | 2 | - | 6.263 | - | 9.288 | 3.025 | 494 | | | | | | | 495 | 3 | 27.556 | - | 30.512 | - | 2.956 | 496 | | | | | | | 497 | | - | 18.113 | - | 21.269 | 3.156 | 498 | | | | | | | 499 | ... | ... | ... | ... | ... | ... | 500 | | | | | | | 501 | n | 77.463 | - | 80.501 | - | 3.038 | 502 | | | | | | | 503 | n+1 | - | 24.333 | - | 27.433 | 3.100 | 504 +-------+---------+---------+---------+---------+-------------+ 506 Table 2: Evaluation of timestamps for delay measurements 508 The first row shows timestamps taken on R1 and R2 respectively and 509 referring to the first packet of block 1 (which is A-colored). Delay 510 can be computed as a difference between the timestamp on R2 and the 511 timestamp on R1. Similarly, the second row shows timestamps (in 512 milliseconds) taken on R1 and R2 and referring to the first packet of 513 block 2 (which is B-colored). Comparing timestamps taken on 514 different nodes in the network and referring to the same packets 515 (identified using the alternation of colors) it is possible to 516 measure delay on different network segments. 518 For the sake of simplicity, in the above example a single measurement 519 is provided within a block, taking into account only the first packet 520 of each block. The number of measurements can be easily increased by 521 considering multiple packets in the block: for instance, a timestamp 522 could be taken every N packets, thus generating multiple delay 523 measurements. Taking this to the limit, in principle the delay could 524 be measured for each packet, by taking and comparing the 525 corresponding timestamps (possible but impractical from an 526 implementation point of view). 528 3.2.2. Mean delay 530 As mentioned before, the method previously exposed for measuring the 531 delay is sensitive to out of order reception of packets. In order to 532 overcome this problem, a different approach has been considered: it 533 is based on the concept of mean delay. The mean delay is calculated 534 by considering the average arrival time of the packets within a 535 single block. The network device locally stores a timestamp for each 536 packet received within a single block: summing all the timestamps and 537 dividing by the total number of packets received, the average arrival 538 time for that block of packets can be calculated. By subtracting the 539 average arrival times of two adjacent devices it is possible to 540 calculate the mean delay between those nodes. This method is robust 541 to out of order packets and also to packet loss (only a small error 542 is introduced). Moreover, it greatly reduces the number of 543 timestamps (only one per block for each network device) that have to 544 be collected by the management system. On the other hand, it only 545 gives one measure for the duration of the block (f.i. 5 minutes), and 546 it doesn't give the minimum, maximum and median delay values (RFC 547 6703 [RFC6703]). This limitation could be overcome by reducing the 548 duration of the block (f.i. from 5 minutes to a few seconds), that 549 implicates an highly optimized implementation of the method. 551 By summing the mean delays of the two directions of a path, it is 552 also possible to measure the two-way delay (round-trip delay). 554 3.2.3. Double marking methodology 556 The Single marking methodology for one-way delay measurement is 557 sensitive to out of order reception of packets. The first approach 558 to overcome this problem is described before and is based on the 559 concept of mean delay. But the limitation of mean delay is that it 560 doesn't give information about the delay values distribution for the 561 duration of the block. Additionally it may be useful to have not 562 only the mean delay but also the minimum and maximum delay values 563 and, in wider terms, to know more about the statistic distribution of 564 delay values. So in order to have more information about the delay 565 and to overcome out of order issues, a different approach can be 566 introduced: it is based on double marking methodology. 568 Basically, the idea is to use the first marking to create the 569 alternate flow and, within this colored flow, a second marking to 570 select the packets for measuring delay/jitter. The first marking is 571 needed for packet loss and mean delay measurement. The second 572 marking creates a new set of marked packets that are fully identified 573 over the network, so that a network device can store the timestamps 574 of these packets; these timestamps can be compared with the 575 timestamps of the same packets on a second router to compute packet 576 delay values for each packet. The number of measurements can be 577 easily increased by changing the frequency of the second marking. 578 But the frequency of the second marking must be not too high in order 579 to avoid out of order issues. Between packets with the second 580 marking there should be a security time gap (e.g. this gap could be, 581 at the minimum, the mean network delay calculated with the previous 582 methodology) to avoid out of order issues and also to have a number 583 of measurement packets that is rate independent. If a second marking 584 packet is lost, the delay measurement for the considered block is 585 corrupted and should be discarded. 587 3.3. Delay variation measurement 589 Similarly to one-way delay measurement (both for single marking and 590 double marking), the method can also be used to measure the inter- 591 arrival jitter. We refer to the definition in RFC 3393 [RFC3393]. 592 The alternation of colors, for single marking method, can be used as 593 a time reference to measure delay variations. In case of double 594 marking, the time reference is given by the second marked packets. 595 Considering the example depicted in Figure 2, R1 stores a timestamp 596 TS(A)R1 whenever it sends the first packet of a block and R2 stores a 597 timestamp TS(B)R2 whenever it receives the first packet of a block. 598 The inter-arrival jitter can be easily derived from one-way delay 599 measurement, by evaluating the delay variation of consecutive 600 samples. 602 The concept of mean delay can also be applied to delay variation, by 603 evaluating the average variation of the interval between consecutive 604 packets of the flow from R1 to R2. 606 4. Considerations 608 This section highlights some considerations about the methodology. 610 4.1. Synchronization 612 The Alternate Marking technique does not require a strong 613 synchronization, especially for packet loss and two-way delay 614 measurement. Only one-way delay measurement requires network devices 615 to have synchronized clocks. 617 The color switching is the reference for all the network devices, and 618 the only requirement to be achieved is that all network devices have 619 to recognize the right batch along the path. 621 If the length of the measurement period is L time units, then all 622 network devices must be synchronized to the same clock reference with 623 an accuracy of +/- L/2 time units (without considering network 624 delay). This level of accuracy guarantees that all network devices 625 consistently match the color bit to the correct block. For example, 626 if the color is toggeled every second (L = 1 second), then clocks 627 must be synchronized with an accuracy of +/- 0.5 second to a common 628 time reference. 630 This synchronization requirement can be satisfied even with a 631 relatively inaccurate synchronization method. This is true for 632 packet loss and two-way delay measurement, instead, for one-way delay 633 measurement clock synchronization must be accurate. 635 Therefore, a system that uses only packet loss and two-way delay 636 measurement does not require synchronization. This is because the 637 value of the clocks of network devices does not affect the 638 computation of the two-way delay measurement. 640 4.2. Data Correlation 642 Data Correlation is the mechanism to compare counters and timestamps 643 for packet loss, delay and delay variation calculation. It could be 644 performed in several ways depending on the alternate marking 645 application and use case. 647 o A possibility is to use a centralized solution using Network 648 Management System (NMS) to correlate data; 650 o Another possibility is to define a protocol based distributed 651 solution, by defining a new protocol or by extending the existing 652 protocols (e.g. RFC6374, TWAMP, OWAMP) in order to communicate 653 the counters and timestamps between nodes. 655 In the following paragraphs an example data correlation mechanism is 656 explained and could be use independently of the adopted solutions. 658 When data is collected on the upstream and downstream node, e.g., 659 packet counts for packet loss measurement or timestamps for packet 660 delay measurement, and periodically reported to or pulled by other 661 nodes or NMS, a certain data correlation mechanism SHOULD be in use 662 to help the nodes or NMS to tell whether any two or more packet 663 counts are related to the same block of markers, or any two 664 timestamps are related to the same marked packet. 666 The alternate marking method described in this document literally 667 split the packets of the measured flow into different measurement 668 blocks, in addition a Block Number could be assigned to each of such 669 measurement block. The BN is generated each time a node reads the 670 data (packet counts or timestamps), and is associated with each 671 packet count and timestamp reported to or pulled by other nodes or 672 NMS. The value of BN could be calculated as the modulo of the local 673 time (when the data are read) and the interval of the marking time 674 period. 676 When the nodes or NMS see, for example, same BNs associated with two 677 packet counts from an upstream and a downstream node respectively, it 678 considers that these two packet counts corresponding to the same 679 block, i.e. that these two packet counts belong to the same block of 680 markers from the upstream and downstream node. The assumption of 681 this BN mechanism is that the measurement nodes are time 682 synchronized. This requires the measurement nodes to have a certain 683 time synchronization capability (e.g., the Network Time Protocol 684 (NTP) [RFC5905], or the IEEE 1588 Precision Time Protocol (PTP) 685 [IEEE1588]). Synchronization aspects are further discussed in 686 Section 4. 688 4.3. Packet Re-ordering 690 Due to ECMP, packet re-ordering is very common in IP network. The 691 accuracy of marking based PM, especially packet loss measurement, may 692 be affected by packet re-ordering. Take a look at the following 693 example: 695 Block : 1 | 2 | 3 | 4 | 5 |... 696 --------|---------|---------|---------|---------|---------|--- 697 Node R1 : AAAAAAA | BBBBBBB | AAAAAAA | BBBBBBB | AAAAAAA |... 698 Node R2 : AAAAABB | AABBBBA | AAABAAA | BBBBBBA | ABAAABA |... 700 Figure 5: Packet Reordering 702 In the following paragraphs an example of data correlation mechanism 703 is explained and could be use independently of the adopted solutions. 705 Most of the packet re-ordering occur at the edge of adjacent blocks, 706 and they are easy to handle if the interval of each block is 707 sufficient large. Then, it can assume that the packets with 708 different marker belong to the block that they are more close to. If 709 the interval is small, it is difficult and sometime impossible to 710 determine to which block a packet belongs. See above example, the 711 packet with the marker of "B" in block 3, there is no safe way to 712 tell whether the packet belongs to block 2 or block 4. 714 To choose a proper interval is important and how to choose a proper 715 interval is out of the scope of this document. But an implementation 716 SHOULD provide a way to configure the interval and allow a certain 717 degree of packet re-ordering. 719 5. Implementation and deployment 721 The methodology described in the previous sections can be applied in 722 various situations. Basically Alternate Marking technique could be 723 used in many cases for performance measurement. The only requirement 724 is to select and mark the flow to be monitored; in this way packets 725 are batched by the sender and each batch is alternately marked such 726 that can be easily recognized by the receiver. 728 An example of implementation and deployment is explained in the next 729 section, just to clarify how the method can work. 731 5.1. Report on the operational experiment at Telecom Italia 733 The method described in this document, also called PNPM (Packet 734 Network Performance Monitoring), has been invented and engineered in 735 Telecom Italia and it's currently being used in Telecom Italia's 736 network. The methodology has been applied by leveraging functions 737 and tools available on IP routers and it's currently being used to 738 monitor packet loss in some portions of Telecom Italia's network. 739 The application of the method to delay measurement is currently being 740 evaluated in Telecom Italia's labs. This section describes how the 741 features currently available on existing routing platforms can be 742 used to apply the method, in order to give an example of 743 implementation and deployment. 745 The fundamental steps for this implementation of the method can be 746 summarized in the following items: 748 o coloring the packets; 749 o counting the packets; 751 o collecting data and calculating the packet loss. 753 o metric transparency. 755 Before going deeper into the implementation details, it's worth 756 mentioning two different strategies that can be used when 757 implementing the method: 759 o flow-based: the flow-based strategy is used when only a limited 760 number of traffic flows need to be monitored. This could be the 761 case, for example, of IPTV channels or other specific applications 762 traffic with high QoS requirements (i.e. Mobile Backhauling 763 traffic). According to this strategy, only a subset of the flows 764 is colored. Counters for packet loss measurements can be 765 instantiated for each single flow, or for the set as a whole, 766 depending on the desired granularity. A relevant problem with 767 this approach is the necessity to know in advance the path 768 followed by flows that are subject to measurement. Path rerouting 769 and traffic load-balancing increase the issue complexity, 770 especially for unicast traffic. The problem is easier to solve 771 for multicast traffic where load balancing is seldom used, 772 especially for IPTV traffic where static joins are frequently used 773 to force traffic forwarding and replication. Another application 774 is on Mobile Backhauling, implemented with a VPN MPLS in Telecom 775 Italia's network; in this case the problem with unicast traffic is 776 overcome by monitoring just the two Provider Edge nodes of the VPN 777 MPLS. 779 o link-based: measurements are performed on all the traffic on a 780 link by link basis. The link could be a physical link or a 781 logical link (for instance an Ethernet VLAN or a MPLS PW). 782 Counters could be instantiated for the traffic as a whole or for 783 each traffic class (in case it is desired to monitor each class 784 separately), but in the second case a couple of counters is needed 785 for each class. 787 The current implementation in Telecom Italia uses the first strategy. 788 As mentioned, the flow-based measurement requires the identification 789 of the flow to be monitored and the discovery of the path followed by 790 the selected flow. It is possible to monitor a single flow or 791 multiple flows grouped together, but in this case measurement is 792 consistent only if all the flows in the group follow the same path. 793 Moreover, a Service Provider should be aware that, if a measurement 794 is performed by grouping many flows, it is not possible to determine 795 exactly which flow was affected by packets loss. In order to have 796 measures per single flow it is necessary to configure counters for 797 each specific flow. Once the flow(s) to be monitored have been 798 identified, it is necessary to configure the monitoring on the proper 799 nodes. Configuring the monitoring means configuring the policy to 800 intercept the traffic and configuring the counters to count the 801 packets. To have just an end-to-end monitoring, it is sufficient to 802 enable the monitoring on the first and the last hop routers of the 803 path: the mechanism is completely transparent to intermediate nodes 804 and independent from the path followed by traffic flows. On the 805 contrary, to monitor the flow on a hop-by-hop basis along its whole 806 path it is necessary to enable the monitoring on every node from the 807 source to the destination. In case the exact path followed by the 808 flow is not known a priori (i.e. the flow has multiple paths to reach 809 the destination) it is necessary to enable the monitoring system on 810 every path: counters on interfaces traversed by the flow will report 811 packet count, counters on other interfaces will be null. 813 5.1.1. Coloring the packets 815 The coloring operation is fundamental in order to create packet 816 blocks. This implies choosing where to activate the coloring and how 817 to color the packets. 819 In case of flow-based measurements, it is desirable, in general, to 820 have a single coloring node because it is easier to manage and 821 doesn't rise any risk of conflict (consider the case where two nodes 822 color the same flow). Thus it is necessary to color the flow as 823 close as possible to the source. In addition, coloring a flow close 824 to the source allows an end-to-end measure if a measurement point is 825 enabled on the last-hop router as well. The only requirement is that 826 the coloring must change periodically and every node along the path 827 must be able to identify unambiguously the colored packets. For 828 link-based measurements, all traffic needs to be colored when 829 transmitted on the link. If the traffic had already been colored, 830 then it has to be re-colored because the color must be consistent on 831 the link. This means that each hop along the path must (re-)color 832 the traffic; the color is not required to be consistent along 833 different links. 835 Traffic coloring can be implemented by setting a specific bit in the 836 packet header and changing the value of that bit periodically. With 837 current router implementations, only QoS related fields and features 838 offer the required flexibility to set bits in the packet header. In 839 case a Service Provider only uses the three most significant bits of 840 the DSCP field (corresponding to IP Precedence) for QoS 841 classification and queuing, it is possible to use the two less 842 significant bits of the DSCP field (bit 0 and bit 1) to implement the 843 method without affecting QoS policies. One of the two bits (bit 0) 844 could be used to identify flows subject to traffic monitoring (set to 845 1 if the flow is under monitoring, otherwise it is set to 0), while 846 the second (bit 1) can be used for coloring the traffic (switching 847 between values 0 and 1, corresponding to color A and B) and creating 848 the blocks. 850 In practice, coloring the traffic using the DSCP field can be 851 implemented by configuring on the router output interface an access 852 list that intercepts the flow(s) to be monitored and applies to them 853 a policy that sets the DSCP field accordingly. Since traffic 854 coloring has to be switched between the two values over time, the 855 policy needs to be modified periodically: an automatic script ca be 856 used perform this task on the basis of a fixed timer. In Telecom 857 Italia's implementation this timer is set to 5 minutes: this value 858 showed to be a good compromise between measurement frequency and 859 stability of the measurement (i.e. possibility to collect all the 860 measures referring to the same block). 862 5.1.2. Counting the packets 864 Assuming that the coloring of the packets is performed only by the 865 source node, the nodes between source and destination (included) have 866 to count the colored packets that they receive and forward: this 867 operation can be enabled on every router along the path or only on a 868 subset, depending on which network segment is being monitored (a 869 single link, a particular metro area, the backbone, the whole path). 871 Since the color switches periodically between two values, two 872 counters (one for each value) are needed: one counter for packets 873 with color A and one counter for packets with color B. For each flow 874 (or group of flows) being monitored and for every interface where the 875 monitoring is active, a couple od counters is needed. For example, 876 in order to monitor separately 3 flows on a router with 4 interfaces 877 involved, 24 counters are needed (2 counters for each of the 3 flows 878 on each of the 4 interfaces). If traffic is colored using the DSCP 879 field, as in Telecom Italia's implementation, an access-list that 880 matches specific DSCP values can be used to count the packets of the 881 flow(s) being monitored. 883 In case of link-based measurements the behaviour is similar except 884 that coloring and counting operations are performed on a link by link 885 basis at each endpoint of the link. 887 Another important aspect to take into consideration is when to read 888 the counters: in order to count the exact number of packets of a 889 block the routers must perform this operation when that block has 890 ended: in other words, the counter for color A must be read when the 891 current block has color B, in order to be sure that the value of the 892 counter is stable. This task can be accomplished in two ways. The 893 general approach suggests to read the counters periodically, many 894 times during a block duration, and to compare these successive 895 readings: when the counter stops incrementing means that the current 896 block has ended and its value can be elaborated safely. 897 Alternatively, if the coloring operation is performed on the basis of 898 a fixed timer, it is possible to configure the reading of the 899 counters according to that timer: for example, if each block is 5 900 minutes long, reading the counter for color A every 5 minute in the 901 middle of the subsequent block (with color B) is a safe choice. A 902 sufficient margin should be considered between the end of a block and 903 the reading of the counter, in order to take into account any out-of- 904 order packets. The choice of a 5 minutes timer for colore switching 905 was also inspired by these considerations. 907 5.1.3. Collecting data and calculating packet loss 909 The nodes enabled to perform performance monitoring collect the value 910 of the counters, but they are not able to directly use this 911 information to measure packet loss, because they only have their own 912 samples. For this reason, an external Network Management System 913 (NMS) is required to collect and elaborate data and to perform packet 914 loss calculation. The NMS compares the values of counters from 915 different nodes and can calculate if some packets were lost (even a 916 single packet) and also where packets were lost. 918 The value of the counters needs to be transmitted to the NMS as soon 919 as it has been read. This can be accomplished by using SNMP or FTP 920 and can be done in Push Mode or Polling Mode. In the first case, 921 each router periodically sends the information to the NMS, in the 922 latter case it is the NMS that periodically polls routers to collect 923 information. In any case, the NMS has to collect all the relevant 924 values from all the routers within one cycle of the timer (5 925 minutes). 927 If link-based measurement is used, it would be possible to use a 928 protocol to exchange values of counters between the two endpoints in 929 order to let them perform the packet loss calculation for each 930 traffic direction. A similar approach could be complicated if 931 applied to a flow-based measurement. 933 5.1.4. Metric transparency 935 In Telecom Italia's implementation the source node colors the packets 936 with a policy that is modified periodically via an automatic script 937 in order to alternate the DSCP field of the packets. The nodes 938 between source and destination (included) have to count with an 939 access-list the colored packets that they receive and forward. 941 Moreover the destination node has an important role: the colored 942 packets are intercepted and a policy restores and sets the DSCP field 943 of all the packets to the initial value. In this way the metric is 944 transparent because outside the section of the network under 945 monitoring the traffic flow is unchanged. 947 In such a case, thanks to this restoring technique, network elements 948 outside the Alternate Marking monitoring domain (e.g. the two 949 Provider Edge nodes of the Mobile Backhauling VPN MPLS) are totally 950 anaware that packets were marked. So this restoring technique makes 951 Alternate Marking completely transparent outside its monitoring 952 domain. 954 5.2. IP flow performance measurement (IPFPM) 956 This application of marking method is described in 957 [I-D.chen-ippm-coloring-based-ipfpm-framework]. 959 5.3. Performance Measurement Marking Method in BIER Domain 961 In [I-D.ietf-bier-mpls-encapsulation] two OAM bits from Bit Index 962 Explicit Replication (BIER) Header are reserved for the passive 963 performance measurement marking method. [I-D.ietf-bier-pmmm-oam] 964 details the measurement for multicast service over BIER domain. 966 5.4. Overlay OAM Passive Performance Measurement 968 The Overlay OAM Design Team is considering the preliminary OAM 969 requirements from NVO3, BIER, and SFC. Marking Method is the 970 preferred passive method to measure performance. 971 [I-D.ooamdt-rtgwg-ooam-requirement] and 972 [I-D.ooamdt-rtgwg-oam-gap-analysis] explain in deep this item. 974 5.5. RFC6374 Use Case 976 RFC6374 [RFC6374] uses the LM packet as the packet accounting 977 demarcation point. Unfortunately this gives rise to a number of 978 problems that may lead to significant packet accounting errors in 979 certain situations. [I-D.ietf-mpls-flow-ident] discusses the desired 980 capabilities for MPLS flow identification in order to perform a 981 better in-band performance monitoring of user data packets. A method 982 of accomplishing identification is Synonymous Flow Labels (SFL) 983 introduced in [I-D.bryant-mpls-sfl-framework], while 984 [I-D.bryant-mpls-rfc6374-sfl] describes RFC6374 performance 985 measurements with SFL. 987 5.6. Application to active performance measurement 989 [I-D.fioccola-ippm-alt-mark-active] describes how to extend the 990 existing Active Measurement Protocol, in order to implement alternate 991 marking methodology. [I-D.fioccola-ippm-rfc6812-alt-mark-ext] 992 describes an extension to the Cisco SLA Protocol Measurement-Type 993 UDP-Measurement. 995 6. Hybrid measurement 997 The method has been explicitly designed for passive measurements but 998 it can also be used with active measurements. In order to have both 999 end to end measurements and intermediate measurements (hybrid 1000 measurements) two end points can exchanges artificial traffic flows 1001 and apply alternate marking over these flows. In the intermediate 1002 points artificial traffic is managed in the same way as real traffic 1003 and measured as specified before. So the application of marking 1004 method can simplify also the active measurement, as explained in 1005 [I-D.fioccola-ippm-alt-mark-active]. 1007 7. Compliance with RFC6390 guidelines 1009 RFC6390 [RFC6390] defines a framework and a process for developing 1010 Performance Metrics for protocols above and below the IP layer (such 1011 as IP-based applications that operate over reliable or datagram 1012 transport protocols). 1014 This document doesn't aim to propose a new Performance Metric but a 1015 new method of measurement for a few Performance Metrics that have 1016 already been standardized. Nevertheless, it's worth applying 1017 [RFC6390] guidelines to the present document, in order to provide a 1018 more complete and coherent description of the proposed method. We 1019 used a subset of the Performance Metric Definition template defined 1020 by [RFC6390]. 1022 o Metric name and description: as already stated, this document 1023 doesn't propose any new Performance Metric. On the contrary, it 1024 describes a novel method for measuring packet loss [RFC2680]. The 1025 same concept, with small differences, can also be used to measure 1026 delay [RFC2679], and jitter [RFC3393]. The document mainly 1027 describes the applicability to packet loss measurement. 1029 o Method of Measurement or Calculation: according to the method 1030 described in the previous sections, the number of packets lost is 1031 calculated by subtracting the value of the counter on the source 1032 node from the value of the counter on the destination node. Both 1033 counters must refer to the same color. The calculation is 1034 performed when the value of the counters is in a steady state. 1036 o Units of Measurement: the method calculates and reports the exact 1037 number of packets sent by the source node and not received by the 1038 destination node. 1040 o Measurement Points: the measurement can be performed between 1041 adjacent nodes, on a per-link basis, or along a multi-hop path, 1042 provided that the traffic under measurement follows that path. In 1043 case of a multi-hop path, the measurements can be performed both 1044 end-to-end and hop-by-hop. 1046 o Measurement Timing: the method have a constraint on the frequency 1047 of measurements. In order to perform a measure, the counter must 1048 be in a steady state: this happens when the traffic is being 1049 colored with the alternate color; for example in the Telecom 1050 Italia application of the method the time interval is set to 5 1051 minutes. 1053 o Implementation: the Telecom Italia application of the method uses 1054 two encodings of the DSCP field to color the packets; this enables 1055 the use of policy configurations on the router to color the 1056 packets and accordingly configure the counter for each color. The 1057 path followed by traffic being measured should be known in advance 1058 in order to configure the counters along the path and be able to 1059 compare the correct values. 1061 o Use and Applications: the method can be used to measure packet 1062 loss with high precision on live traffic; moreover, by combining 1063 end-to-end and per-link measurements, the method is useful to 1064 pinpoint the single link that is experiencing loss events. 1066 o Reporting Model: the value of the counters has to be sent to a 1067 centralized management system that perform the calculations; such 1068 samples must contain a reference to the time interval they refer 1069 to, so that the management system can perform the correct 1070 correlation; the samples have to be sent while the corresponding 1071 counter is in a steady state (within a time interval), otherwise 1072 the value of the sample should be stored locally. 1074 o Dependencies: the values of the counters have to be correlated to 1075 the time interval they refer to; moreover, as far the Telecom 1076 Italia application of the method is based on DSCP values, there 1077 are significant dependencies on the usage of the DSCP field: it 1078 must be possible to rely on unused DSCP values without affecting 1079 QoS-related configuration and behavior; moreover, the intermediate 1080 nodes must not change the value of the DSCP field not to alter the 1081 measurement. 1083 o Organization of Results: the method of measurement produces 1084 singletons. 1086 o Parameters: currently, the main parameter of the method is the 1087 time interval used to alternate the colors and read the counters. 1089 8. Security Considerations 1091 This document specifies a method to perform measurements in the 1092 context of a Service Provider's network and has not been developed to 1093 conduct Internet measurements, so it does not directly affect 1094 Internet security nor applications which run on the Internet. 1095 However, implementation of this method must be mindful of security 1096 and privacy concerns. 1098 There are two types of security concerns: potential harm caused by 1099 the measurements and potential harm to the measurements. For what 1100 concerns the first point, the measurements described in this document 1101 are passive, so there are no packets injected into the network 1102 causing potential harm to the network itself and to data traffic. 1103 Nevertheless, the method implies modifications on the fly to the IP 1104 header of data packets: this must be performed in a way that doesn't 1105 alter the quality of service experienced by packets subject to 1106 measurements and that preserve stability and performance of routers 1107 doing the measurements. The measurements themselves could be harmed 1108 by routers altering the marking of the packets, or by an attacker 1109 injecting artificial traffic. Authentication techniques, such as 1110 digital signatures, may be used where appropriate to guard against 1111 injected traffic attacks. 1113 The privacy concerns of network measurement are limited because the 1114 method only relies on information contained in the IP header without 1115 any release of user data. 1117 The measurement itself may be affected by routers (or other network 1118 devices) along the path of IP packets intentionally altering the 1119 value of marking bits of packets. As mentioned above, the mechanism 1120 specified in this document is just in the context of one Service 1121 Provider's network, and thus the routers (or other network devices) 1122 are locally administered and this type of attack can be avoided. 1124 One of the main security threats in OAM protocols is network 1125 reconnaissance; an attacker can gather information about the network 1126 performance by passively eavesdropping to OAM messages. The 1127 advantage of the methods described in this document is that the 1128 marking bits are the only information that is exchanged between the 1129 network devices. Therefore, passive eavesdropping to data plane 1130 traffic does not allow attackers to gain information about the 1131 network performance. 1133 Delay attacks are another potential threat in the context of this 1134 document. Delay measurement is performed using a specific packet in 1135 each block, marked by a dedicated color bit. Therefore, a man-in- 1136 the-middle attacker can selectively induce synthetic delay only to 1137 delay-colored packets, causing systematic error in the delay 1138 measurements. As discussed in previous sections, the methods 1139 described in this document rely on an underlying time synchronization 1140 protocol. Thus, by attacking the time protocol an attacker can 1141 potentially compromise the integrity of the measurement. A detailed 1142 discussion about the threats against time protocols and how to 1143 mitigate them is presented in RFC 7384 [RFC7384]. 1145 9. Conclusions 1147 The advantages of the method described in this document are: 1149 o easy implementation: it can be implemented using features already 1150 available on major routing platforms; 1152 o low computational effort: the additional load on processing is 1153 negligible; 1155 o accurate packet loss measurement: single packet loss granularity 1156 is achieved with a passive measurement; 1158 o potential applicability to any kind of packet/frame -based 1159 traffic: Ethernet, IP, MPLS, etc., both unicast and multicast; 1161 o robustness: the method can tolerate out of order packets and it's 1162 not based on "special" packets whose loss could have a negative 1163 impact; 1165 o no interoperability issues: the features required to implement the 1166 method are available on all current routing platforms. 1168 The method doesn't raise any specific need for protocol extension, 1169 but it could be further improved by means of some extension to 1170 existing protocols. Specifically, the use of DiffServ bits for 1171 coloring the packets could not be a viable solution in some cases: a 1172 standard method to color the packets for this specific application 1173 could be beneficial. 1175 10. IANA Considerations 1177 There are no IANA actions required. 1179 11. Acknowledgements 1181 The previous IETF drafts about this technique were: 1182 [I-D.cociglio-mboned-multicast-pm] and [I-D.tempia-opsawg-p3m]. 1183 There are some references to this methodology in other IETF works 1184 (e.g. [I-D.ietf-mpls-flow-ident], [I-D.bryant-mpls-sfl-framework] 1185 [I-D.bryant-mpls-rfc6374-sfl], [I-D.ietf-bier-mpls-encapsulation], 1186 [I-D.ietf-bier-pmmm-oam] 1187 [I-D.chen-ippm-coloring-based-ipfpm-framework]). 1189 In addition the authors would like to thank Domenico Laforgia, 1190 Daniele Accetta and Mario Bianchetti for their contribution to the 1191 definition and the implementation of the method. 1193 12. References 1195 12.1. Normative References 1197 [RFC2679] Almes, G., Kalidindi, S., and M. Zekauskas, "A One-way 1198 Delay Metric for IPPM", RFC 2679, DOI 10.17487/RFC2679, 1199 September 1999, . 1201 [RFC2680] Almes, G., Kalidindi, S., and M. Zekauskas, "A One-way 1202 Packet Loss Metric for IPPM", RFC 2680, 1203 DOI 10.17487/RFC2680, September 1999, 1204 . 1206 [RFC3393] Demichelis, C. and P. Chimento, "IP Packet Delay Variation 1207 Metric for IP Performance Metrics (IPPM)", RFC 3393, 1208 DOI 10.17487/RFC3393, November 2002, 1209 . 1211 12.2. Informative References 1213 [I-D.bryant-mpls-rfc6374-sfl] 1214 Bryant, S., Swallow, G., Sivabalan, S., Mirsky, G., Chen, 1215 M., and Z. Li, "RFC6374 Synonymous Flow Labels", draft- 1216 bryant-mpls-rfc6374-sfl-01 (work in progress), August 1217 2016. 1219 [I-D.bryant-mpls-sfl-framework] 1220 Bryant, S., Chen, M., Li, Z., Swallow, G., Sivabalan, S., 1221 and G. Mirsky, "Synonymous Flow Label Framework", draft- 1222 bryant-mpls-sfl-framework-02 (work in progress), October 1223 2016. 1225 [I-D.chen-ippm-coloring-based-ipfpm-framework] 1226 Chen, M., Zheng, L., Mirsky, G., Fioccola, G., and T. 1227 Mizrahi, "IP Flow Performance Measurement Framework", 1228 draft-chen-ippm-coloring-based-ipfpm-framework-06 (work in 1229 progress), March 2016. 1231 [I-D.cociglio-mboned-multicast-pm] 1232 Cociglio, M., Capello, A., Bonda, A., and L. Castaldelli, 1233 "A method for IP multicast performance monitoring", draft- 1234 cociglio-mboned-multicast-pm-01 (work in progress), 1235 October 2010. 1237 [I-D.fioccola-ippm-alt-mark-active] 1238 Fioccola, G., Clemm, A., Cociglio, M., Chandramouli, M., 1239 and A. Capello, "Alternate Marking Extension to Active 1240 Measurement Protocol", draft-fioccola-ippm-alt-mark- 1241 active-00 (work in progress), July 2016. 1243 [I-D.fioccola-ippm-rfc6812-alt-mark-ext] 1244 Fioccola, G., Clemm, A., Cociglio, M., Chandramouli, M., 1245 and A. Capello, "Alternate Marking Extension to Cisco SLA 1246 Protocol RFC6812", draft-fioccola-ippm-rfc6812-alt-mark- 1247 ext-01 (work in progress), March 2016. 1249 [I-D.ietf-bier-mpls-encapsulation] 1250 Wijnands, I., Rosen, E., Dolganow, A., Tantsura, J., 1251 Aldrin, S., and I. Meilik, "Encapsulation for Bit Index 1252 Explicit Replication in MPLS Networks", draft-ietf-bier- 1253 mpls-encapsulation-05 (work in progress), July 2016. 1255 [I-D.ietf-bier-pmmm-oam] 1256 Mirsky, G., Zheng, L., Chen, M., and G. Fioccola, 1257 "Performance Measurement (PM) with Marking Method in Bit 1258 Index Explicit Replication (BIER) Layer", draft-ietf-bier- 1259 pmmm-oam-00 (work in progress), July 2016. 1261 [I-D.ietf-mpls-flow-ident] 1262 Bryant, S., Chen, M., Li, Z., Pignataro, C., and G. 1263 Mirsky, "MPLS Flow Identification Considerations", draft- 1264 ietf-mpls-flow-ident-02 (work in progress), August 2016. 1266 [I-D.ooamdt-rtgwg-oam-gap-analysis] 1267 Mirsky, G., Nordmark, E., Pignataro, C., Kumar, N., Kumar, 1268 D., Chen, M., Yizhou, L., Mozes, D., Networks, J., and I. 1269 Bagdonas, "Operations, Administration and Maintenance 1270 (OAM) for Overlay Networks: Gap Analysis", draft-ooamdt- 1271 rtgwg-oam-gap-analysis-02 (work in progress), July 2016. 1273 [I-D.ooamdt-rtgwg-ooam-requirement] 1274 Kumar, N., Pignataro, C., Kumar, D., Mirsky, G., Chen, M., 1275 Nordmark, E., Networks, J., and D. Mozes, "Overlay OAM 1276 Requirements", draft-ooamdt-rtgwg-ooam-requirement-01 1277 (work in progress), July 2016. 1279 [I-D.tempia-opsawg-p3m] 1280 Capello, A., Cociglio, M., Castaldelli, L., and A. Bonda, 1281 "A packet based method for passive performance 1282 monitoring", draft-tempia-opsawg-p3m-04 (work in 1283 progress), February 2014. 1285 [RFC6374] Frost, D. and S. Bryant, "Packet Loss and Delay 1286 Measurement for MPLS Networks", RFC 6374, 1287 DOI 10.17487/RFC6374, September 2011, 1288 . 1290 [RFC6390] Clark, A. and B. Claise, "Guidelines for Considering New 1291 Performance Metric Development", BCP 170, RFC 6390, 1292 DOI 10.17487/RFC6390, October 2011, 1293 . 1295 [RFC6703] Morton, A., Ramachandran, G., and G. Maguluri, "Reporting 1296 IP Network Performance Metrics: Different Points of View", 1297 RFC 6703, DOI 10.17487/RFC6703, August 2012, 1298 . 1300 [RFC7276] Mizrahi, T., Sprecher, N., Bellagamba, E., and Y. 1301 Weingarten, "An Overview of Operations, Administration, 1302 and Maintenance (OAM) Tools", RFC 7276, 1303 DOI 10.17487/RFC7276, June 2014, 1304 . 1306 [RFC7384] Mizrahi, T., "Security Requirements of Time Protocols in 1307 Packet Switched Networks", RFC 7384, DOI 10.17487/RFC7384, 1308 October 2014, . 1310 Authors' Addresses 1312 Giuseppe Fioccola (editor) 1313 Telecom Italia 1314 Via Reiss Romoli, 274 1315 Torino 10148 1316 Italy 1318 Email: giuseppe.fioccola@telecomitalia.it 1320 Alessandro Capello (editor) 1321 Telecom Italia 1322 Via Reiss Romoli, 274 1323 Torino 10148 1324 Italy 1326 Email: alessandro.capello@telecomitalia.it 1328 Mauro Cociglio 1329 Telecom Italia 1330 Via Reiss Romoli, 274 1331 Torino 10148 1332 Italy 1334 Email: mauro.cociglio@telecomitalia.it 1336 Luca Castaldelli 1337 Telecom Italia 1338 Via Reiss Romoli, 274 1339 Torino 10148 1340 Italy 1342 Email: luca.castaldelli@telecomitalia.it 1344 Mach(Guoyi) Chen (editor) 1345 Huawei Technologies 1347 Email: mach.chen@huawei.com 1349 Lianshu Zheng (editor) 1350 Huawei Technologies 1352 Email: vero.zheng@huawei.com 1353 Greg Mirsky (editor) 1354 Ericsson 1355 USA 1357 Email: gregory.mirsky@ericsson.com 1359 Tal Mizrahi (editor) 1360 Marvell 1361 6 Hamada st. 1362 Yokneam 1363 Israel 1365 Email: talmi@marvell.com