idnits 2.17.1 draft-ietf-ippm-alt-mark-04.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- ** 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 682: '...correlation mechanism SHOULD be in use...' RFC 2119 keyword, line 737: '... 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 (March 1, 2017) is 2614 days in the past. Is this intentional? Checking references for intended status: Experimental ---------------------------------------------------------------------------- == Missing Reference: 'RFC5905' is mentioned on line 705, but not defined == Missing Reference: 'IEEE1588' is mentioned on line 706, 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reference: RFC 2679 (Obsoleted by RFC 7679) ** Obsolete normative reference: RFC 2680 (Obsoleted by RFC 7680) == Outdated reference: A later version (-04) exists of draft-bryant-mpls-rfc6374-sfl-03 == Outdated reference: A later version (-05) exists of draft-bryant-mpls-sfl-framework-02 == Outdated reference: A later version (-01) exists of draft-fioccola-ippm-alt-mark-active-00 == Outdated reference: A later version (-12) exists of draft-ietf-bier-mpls-encapsulation-06 == Outdated reference: A later version (-15) exists of draft-ietf-bier-pmmm-oam-01 == Outdated reference: A later version (-07) exists of draft-ietf-mpls-flow-ident-04 Summary: 3 errors (**), 0 flaws (~~), 9 warnings (==), 1 comment (--).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: September 2, 2017 L. Castaldelli 6 Telecom Italia 7 M. Chen, Ed. 8 L. Zheng, Ed. 9 Huawei Technologies 10 G. Mirsky, Ed. 11 ZTE 12 T. Mizrahi, Ed. 13 Marvell 14 March 1, 2017 16 Alternate Marking method for passive performance monitoring 17 draft-ietf-ippm-alt-mark-04 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 September 2, 2017. 45 Copyright Notice 47 Copyright (c) 2017 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. Double marking methodology . . . . . . . . . . . . . 12 70 3.3. Delay variation measurement . . . . . . . . . . . . . . . 14 71 4. Considerations . . . . . . . . . . . . . . . . . . . . . . . 14 72 4.1. Synchronization . . . . . . . . . . . . . . . . . . . . . 14 73 4.2. Data Correlation . . . . . . . . . . . . . . . . . . . . 15 74 4.3. Packet Re-ordering . . . . . . . . . . . . . . . . . . . 16 75 5. Implementation and deployment . . . . . . . . . . . . . . . . 16 76 5.1. Report on the operational experiment at Telecom Italia . 17 77 5.1.1. Coloring the packets . . . . . . . . . . . . . . . . 18 78 5.1.2. Counting the packets . . . . . . . . . . . . . . . . 19 79 5.1.3. Collecting data and calculating packet loss . . . . . 20 80 5.1.4. Metric transparency . . . . . . . . . . . . . . . . . 21 81 5.2. IP flow performance measurement (IPFPM) . . . . . . . . . 21 82 5.3. Performance Measurement Marking Method in BIER Domain . . 21 83 5.4. Overlay OAM Passive Performance Measurement . . . . . . . 21 84 5.5. RFC6374 Use Case . . . . . . . . . . . . . . . . . . . . 22 85 5.6. Application to active performance measurement . . . . . . 22 86 6. Hybrid measurement . . . . . . . . . . . . . . . . . . . . . 22 87 7. Compliance with RFC6390 guidelines . . . . . . . . . . . . . 22 88 8. Security Considerations . . . . . . . . . . . . . . . . . . . 24 89 9. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . 25 90 10. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 26 91 11. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 26 92 12. References . . . . . . . . . . . . . . . . . . . . . . . . . 26 93 12.1. Normative References . . . . . . . . . . . . . . . . . . 26 94 12.2. Informative References . . . . . . . . . . . . . . . . . 27 95 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 29 97 1. Introduction 99 Nowadays, most of the traffic in Service Providers' networks carries 100 real time content. These contents are highly sensitive to packet 101 loss [RFC2680], while interactive contents are sensitive to delay 102 [RFC2679], and jitter [RFC3393]. 104 In view of this scenario, Service Providers need methodologies and 105 tools to monitor and measure network performances with an adequate 106 accuracy, in order to constantly control the quality of experience 107 perceived by their customers. On the other hand, performance 108 monitoring provides useful information for improving network 109 management (e.g. isolation of network problems, troubleshooting, 110 etc.). 112 A lot of work related to OAM, that includes also performance 113 monitoring techniques, has been done by Standards Developing 114 Organizations(SDOs):: [RFC7276] provides a good overview of existing 115 OAM mechanisms defined in IETF, ITU-T and IEEE. Considering IETF, a 116 lot of work has been done on fault detection and connectivity 117 verification, while a minor effort has been dedicated so far to 118 performance monitoring. The IPPM WG has defined standard metrics to 119 measure network performance; however, the methods developed in this 120 WG mainly refer to focus on active measurement techniques. More 121 recently, the MPLS WG has defined mechanisms for measuring packet 122 loss, one-way and two-way delay, and delay variation in MPLS 123 networks[RFC6374], but their applicability to passive measurements 124 has some limitations, especially for pure connection-less networks. 126 The lack of adequate tools to measure packet loss with the desired 127 accuracy drove an effort to design a new method for the performance 128 monitoring of live traffic, possibly easy to implement and deploy. 129 The effort led to the method described in this document: basically, 130 it is a passive performance monitoring technique, potentially 131 applicable to any kind of packet based traffic, including Ethernet, 132 IP, and MPLS, both unicast and multicast. The method addresses 133 primarily packet loss measurement, but it can be easily extended to 134 one-way delay and delay variation measurements as well. 136 The method has been explicitly designed for passive measurements but 137 it can also be used with active probes. Passive measurements are 138 usually more easily understood by customers and provide a much better 139 accuracy, especially for packet loss measurements. 141 This document is organized as follows: 143 o Section 2 gives an overview of the method, including a comparison 144 with different measurement strategies; 146 o Section 3 describes the method in detail; 148 o Section 4 reports considerations about synchronization, data 149 correlation and packet re-ordering; 151 o Section 5 reports examples of implementation and deployment of the 152 method. Furthermore the operational experiment done at Telecom 153 Italia is described; 155 o Section 8 includes some security aspects; 157 o Section 9 finally summarizes some concluding remarks. 159 2. Overview of the method 161 In order to perform packet loss measurements on a live traffic flow, 162 different approaches exist. The most intuitive one consists in 163 numbering the packets, so that each router that receives the flow can 164 immediately detect a packet missing. This approach, though very 165 simple in theory, is not simple to achieve: it requires the insertion 166 of a sequence number into each packet and the devices must be able to 167 extract the number and check it in real time. Such a task can be 168 difficult to implement on live traffic: if UDP is used as the 169 transport protocol, the sequence number is not available; on the 170 other hand, if a higher layer sequence number (e.g. in the RTP 171 header) is used, extracting that information from each packet and 172 process it in real time could overload the device. 174 An alternate approach is to count the number of packets sent on one 175 end, the number of packets received on the other end, and to compare 176 the two values. This operation is much simpler to implement, but 177 requires that the devices performing the measurement are in sync: in 178 order to compare two counters it is required that they refer exactly 179 to the same set of packets. Since a flow is continuous and cannot be 180 stopped when a counter has to be read, it could be difficult to 181 determine exactly when to read the counter. A possible solution to 182 overcome this problem is to virtually split the flow in consecutive 183 blocks by inserting periodically a delimiter so that each counter 184 refers exactly to the same block of packets. The delimiter could be 185 for example a special packet inserted artificially into the flow. 186 However, delimiting the flow using specific packets has some 187 limitations. First, it requires generating additional packets within 188 the flow and requires the equipment to be able to process those 189 packets. In addition, the method is vulnerable to out of order 190 reception of delimiting packets and, to a lesser extent, to their 191 loss. 193 The method proposed in this document follows the second approach, but 194 it doesn't use additional packets to virtually split the flow in 195 blocks. Instead, it "colors" the packets so that the packets 196 belonging to the same block will have the same color, whilst 197 consecutive blocks will have different colors. Each change of color 198 represents a sort of auto-synchronization signal that guarantees the 199 consistency of measurements taken by different devices along the 200 path. 202 Figure 1 represents a very simple network and shows how the method 203 can be used to measure packet loss on different network segments: by 204 enabling the measurement on several interfaces along the path, it is 205 possible to perform link monitoring, node monitoring or end-to-end 206 monitoring. The method is flexible enough to measure packet loss on 207 any segment of the network and can be used to isolate the faulty 208 element. 210 Traffic flow 211 ========================================================> 212 +------+ +------+ +------+ +------+ 213 ---<> R1 <>-----<> R2 <>-----<> R3 <>-----<> R4 <>--- 214 +------+ +------+ +------+ +------+ 215 . . . . . . 216 . . . . . . 217 . <------> <-------> . 218 . Node Packet Loss Link Packet Loss . 219 . . 220 <---------------------------------------------------> 221 End-to-End Packet loss 223 Figure 1: Available measurements 225 3. Detailed description of the method 227 This section describes in detail how the method operate. A special 228 emphasis is given to the measurement of packet loss, that represents 229 the core application of the method, but applicability to delay and 230 jitter measurements is also considered. 232 3.1. Packet loss measurement 234 The basic idea is to virtually split traffic flows into consecutive 235 blocks: each block represents a measurable entity unambiguously 236 recognizable by all network devices along the path. By counting the 237 number of packets in each block and comparing the values measured by 238 different network devices along the path, it is possible to measure 239 packet loss occurred in any single block between any two points. 241 As discussed in the previous section, a simple way to create the 242 blocks is to "color" the traffic (two colors are sufficient) so that 243 packets belonging to different consecutive blocks will have different 244 colors. Whenever the color changes, the previous block terminates 245 and the new one begins. Hence, all the packets belonging to the same 246 block will have the same color and packets of different consecutive 247 blocks will have different colors. The number of packets in each 248 block depends on the criterion used to create the blocks: if the 249 color is switched after a fixed number of packets, then each block 250 will contain the same number of packets (except for any losses); but 251 if the color is switched according to a fixed timer, then the number 252 of packets may be different in each block depending on the packet 253 rate. 255 The following figure shows how a flow looks like when it is split in 256 traffic blocks with colored packets. 258 A: packet with A coloring 259 B: packet with B coloring 261 | | | | | 262 | | Traffic flow | | 263 -------------------------------------------------------------------> 264 BBBBBBB AAAAAAAAAAA BBBBBBBBBBB AAAAAAAAAAA BBBBBBBBBBB AAAAAAA 265 -------------------------------------------------------------------> 266 ... | Block 5 | Block 4 | Block 3 | Block 2 | Block 1 267 | | | | | 269 Figure 2: Traffic coloring 271 Figure 3 shows how the method can be used to measure link packet loss 272 between two adjacent nodes. 274 Referring to the figure, let's assume we want to monitor the packet 275 loss on the link between two routers: router R1 and router R2. 276 According to the method, the traffic is colored alternatively with 277 two different colors, A and B. Whenever the color changes, the 278 transition generates a sort of square-wave signal, as depicted in the 279 following figure. 281 Color A ----------+ +-----------+ +---------- 282 | | | | 283 Color B +-----------+ +-----------+ 284 Block n ... Block 3 Block 2 Block 1 285 <---------> <---------> <---------> <---------> <---------> 287 Traffic flow 288 ===========================================================> 289 Color ...AAAAAAAAAAA BBBBBBBBBBB AAAAAAAAAAA BBBBBBBBBBB AAAAAAA... 290 ===========================================================> 292 Figure 3: Computation of link packet loss 294 Traffic coloring could be done by R1 itself or by an upward router. 295 R1 needs two counters, C(A)R1 and C(B)R1, on its egress interface: 296 C(A)R1 counts the packets with color A and C(B)R1 counts those with 297 color B. As long as traffic is colored A, only counter C(A)R1 will 298 be incremented, while C(B)R1 is not incremented; vice versa, when the 299 traffic is colored as B, only C(B)R1 is incremented. C(A)R1 and 300 C(B)R1 can be used as reference values to determine the packet loss 301 from R1 to any other measurement point down the path. Router R2, 302 similarly, will need two counters on its ingress interface, C(A)R2 303 and C(B)R2, to count the packets received on that interface and 304 colored with color A and B respectively. When an A block ends, it is 305 possible to compare C(A)R1 and C(A)R2 and calculate the packet loss 306 within the block; similarly, when the successive B block terminates, 307 it is possible to compare C(B)R1 with C(B)R2, and so on for every 308 successive block. 310 Likewise, by using two counters on R2 egress interface it is possible 311 to count the packets sent out of R2 interface and use them as 312 reference values to calculate the packet loss from R2 to any 313 measurement point down R2. 315 Using a fixed timer for color switching offers a better control over 316 the method: the (time) length of the blocks can be chosen large 317 enough to simplify the collection and the comparison of measures 318 taken by different network devices. It's preferable to read the 319 value of the counters not immediately after the color switch: some 320 packets could arrive out of order and increment the counter 321 associated to the previous block (color), so it is worth waiting for 322 some time. A safe choice is to wait L/2 time units (where L is the 323 duration for each block) after the color switch, to read the still 324 counter of the previous color, so the possibility to read a running 325 counter instead of a still one is minimized. The drawback is that 326 the longer the duration of the block, the less frequent the 327 measurement can be taken. 329 The following table shows how the counters can be used to calculate 330 the packet loss between R1 and R2. The first column lists the 331 sequence of traffic blocks while the other columns contain the 332 counters of A-colored packets and B-colored packets for R1 and R2. 333 In this example, we assume that the values of the counters are reset 334 to zero whenever a block ends and its associated counter has been 335 read: with this assumption, the table shows only relative values, 336 that is the exact number of packets of each color within each block. 337 If the values of the counters were not reset, the table would contain 338 cumulative values, but the relative values could be determined simply 339 by difference from the value of the previous block of the same color. 341 The color is switched on the basis of a fixed timer (not shown in the 342 table), so the number of packets in each block is different. 344 +-------+--------+--------+--------+--------+------+ 345 | Block | C(A)R1 | C(B)R1 | C(A)R2 | C(B)R2 | Loss | 346 +-------+--------+--------+--------+--------+------+ 347 | 1 | 375 | 0 | 375 | 0 | 0 | 348 | | | | | | | 349 | 2 | 0 | 388 | 0 | 388 | 0 | 350 | | | | | | | 351 | 3 | 382 | 0 | 381 | 0 | 1 | 352 | | | | | | | 353 | 4 | 0 | 377 | 0 | 374 | 3 | 354 | | | | | | | 355 | ... | ... | ... | ... | ... | ... | 356 | | | | | | | 357 | n | 0 | 387 | 0 | 387 | 0 | 358 | | | | | | | 359 | n+1 | 379 | 0 | 377 | 0 | 2 | 360 +-------+--------+--------+--------+--------+------+ 362 Table 1: Evaluation of counters for packet loss measurements 364 During an A block (blocks 1, 3 and n+1), all the packets are 365 A-colored, therefore the C(A) counters are incremented to the number 366 seen on the interface, while C(B) counters are zero. Vice versa, 367 during a B block (blocks 2, 4 and n), all the packets are B-colored: 368 C(A) counters are zero, while C(B) counters are incremented. 370 When a block ends (because of color switching) the relative counters 371 stop incrementing and it is possible to read them, compare the values 372 measured on router R1 and R2 and calculate the packet loss within 373 that block. 375 For example, looking at the table above, during the first block 376 (A-colored), C(A)R1 and C(A)R2 have the same value (375), which 377 corresponds to the exact number of packets of the first block (no 378 loss). Also during the second block (B-colored) R1 and R2 counters 379 have the same value (388), which corresponds to the number of packets 380 of the second block (no loss). During blocks three and four, R1 and 381 R2 counters are different, meaning that some packets have been lost: 382 in the example, one single packet (382-381) was lost during block 383 three and three packets (377-374) were lost during block four. 385 The method applied to R1 and R2 can be extended to any other router 386 and applied to more complex networks, as far as the measurement is 387 enabled on the path followed by the traffic flow(s) being observed. 389 3.1.1. Timing aspects 391 This document introduces two color switching method: one is based on 392 fixed number of packet, the other is based on fixed timer. But the 393 method based on fixed timer is preferable because is more 394 deterministic, and will be considered in the rest of the dcoument. 396 By considering the clock error between network devices R1 and R2, 397 they must be synchronized to the same clock reference with an 398 accuracy of +/- L/2 time units, where L is the time duration of the 399 block. So each colored packet can be assigned to the right batch by 400 each router. This is because the minimum time distance between two 401 packets of the same color but belonging to different batches is L 402 time units. 404 In practice, there are also out of order at batch boundaries, 405 strictly related to the delay between measurement points. This means 406 that, without considering clock error, we wait L/2 after color 407 switching to be sure to take a still counter. 409 In summary we need to take into account two contributions: clock 410 error between network devices and the interval we need to wait to 411 avoid out of order because of network delay. 413 The following figure eplains both issues. 415 ...BBBBBBBBB | AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA | BBBBBBBBB... 416 |<======================================>| 417 | L | 418 ...=========>|<==================><==================>|<==========... 419 | L/2 L/2 | 420 |<===>| |<===>| 421 d | | d 422 |<==========================>| 423 available counting interval 425 Figure 4: Timing aspects 427 It is assumed that all network devices are synchronized to a common 428 reference time with an accuracy of +/- A/2. Thus, the difference 429 between the clock values of any two network devices is bounded by A. 431 The guardband d is given by: 433 d = A + D_max - D_min, 435 where A is the clock accuracy, D_max is an upper bound on the network 436 delay between the network devices, and D_min is a lower bound on the 437 delay. 439 The available counting interval is L - 2d that must be > 0. 441 The condition that must be satisfied and is a requirement on the 442 synchronization accuracy is: 444 d < L/2. 446 3.2. One-way delay measurement 448 The same principle used to measure packet loss can be applied also to 449 one-way delay measurement. There are three alternatives, as 450 described hereinafter. 452 3.2.1. Single marking methodology 454 The alternation of colors can be used as a time reference to 455 calculate the delay. Whenever the color changes (that means that a 456 new block has started) a network device can store the timestamp of 457 the first packet of the new block; that timestamp can be compared 458 with the timestamp of the same packet on a second router to compute 459 packet delay. Considering Figure 2, R1 stores a timestamp TS(A1)R1 460 when it sends the first packet of block 1 (A-colored), a timestamp 461 TS(B2)R1 when it sends the first packet of block 2 (B-colored) and so 462 on for every other block. R2 performs the same operation on the 463 receiving side, recording TS(A1)R2, TS(B2)R2 and so on. Since the 464 timestamps refer to specific packets (the first packet of each block) 465 we are sure that timestamps compared to compute delay refer to the 466 same packets. By comparing TS(A1)R1 with TS(A1)R2 (and similarly 467 TS(B2)R1 with TS(B2)R2 and so on) it is possible to measure the delay 468 between R1 and R2. In order to have more measurements, it is 469 possible to take and store more timestamps, referring to other 470 packets within each block. 472 In order to coherently compare timestamps collected on different 473 routers, the network nodes must be in sync. Furthermore, a 474 measurement is valid only if no packet loss occurs and if packet 475 misordering can be avoided, otherwise the first packet of a block on 476 R1 could be different from the first packet of the same block on R2 477 (f.i. if that packet is lost between R1 and R2 or it arrives after 478 the next one). 480 The following table shows how timestamps can be used to calculate the 481 delay between R1 and R2. The first column lists the sequence of 482 blocks while other columns contain the timestamp referring to the 483 first packet of each block on R1 and R2. The delay is computed as a 484 difference between timestamps. For the sake of simplicity, all the 485 values are expressed in milliseconds. 487 +-------+---------+---------+---------+---------+-------------+ 488 | Block | TS(A)R1 | TS(B)R1 | TS(A)R2 | TS(B)R2 | Delay R1-R2 | 489 +-------+---------+---------+---------+---------+-------------+ 490 | 1 | 12.483 | - | 15.591 | - | 3.108 | 491 | | | | | | | 492 | 2 | - | 6.263 | - | 9.288 | 3.025 | 493 | | | | | | | 494 | 3 | 27.556 | - | 30.512 | - | 2.956 | 495 | | | | | | | 496 | | - | 18.113 | - | 21.269 | 3.156 | 497 | | | | | | | 498 | ... | ... | ... | ... | ... | ... | 499 | | | | | | | 500 | n | 77.463 | - | 80.501 | - | 3.038 | 501 | | | | | | | 502 | n+1 | - | 24.333 | - | 27.433 | 3.100 | 503 +-------+---------+---------+---------+---------+-------------+ 505 Table 2: Evaluation of timestamps for delay measurements 507 The first row shows timestamps taken on R1 and R2 respectively and 508 referring to the first packet of block 1 (which is A-colored). Delay 509 can be computed as a difference between the timestamp on R2 and the 510 timestamp on R1. Similarly, the second row shows timestamps (in 511 milliseconds) taken on R1 and R2 and referring to the first packet of 512 block 2 (which is B-colored). Comparing timestamps taken on 513 different nodes in the network and referring to the same packets 514 (identified using the alternation of colors) it is possible to 515 measure delay on different network segments. 517 For the sake of simplicity, in the above example a single measurement 518 is provided within a block, taking into account only the first packet 519 of each block. The number of measurements can be easily increased by 520 considering multiple packets in the block: for instance, a timestamp 521 could be taken every N packets, thus generating multiple delay 522 measurements. Taking this to the limit, in principle the delay could 523 be measured for each packet, by taking and comparing the 524 corresponding timestamps (possible but impractical from an 525 implementation point of view). 527 3.2.1.1. Mean delay 529 As mentioned before, the method previously exposed for measuring the 530 delay is sensitive to out of order reception of packets. In order to 531 overcome this problem, a different approach has been considered: it 532 is based on the concept of mean delay. The mean delay is calculated 533 by considering the average arrival time of the packets within a 534 single block. The network device locally stores a timestamp for each 535 packet received within a single block: summing all the timestamps and 536 dividing by the total number of packets received, the average arrival 537 time for that block of packets can be calculated. By subtracting the 538 average arrival times of two adjacent devices it is possible to 539 calculate the mean delay between those nodes. This method is robust 540 to out of order packets and also to packet loss (only a small error 541 is introduced). Moreover, it greatly reduces the number of 542 timestamps (only one per block for each network device) that have to 543 be collected by the management system. On the other hand, it only 544 gives one measure for the duration of the block (f.i. 5 minutes), and 545 it doesn't give the minimum, maximum and median delay values (RFC 546 6703 [RFC6703]). This limitation could be overcome by reducing the 547 duration of the block (f.i. from 5 minutes to a few seconds), that 548 implicates an highly optimized implementation of the method. 550 By summing the mean delays of the two directions of a path, it is 551 also possible to measure the two-way mean delay (round-trip delay). 553 3.2.2. Double marking methodology 555 The Single marking methodology for one-way delay measurement is 556 sensitive to out of order reception of packets. The first approach 557 to overcome this problem is described before and is based on the 558 concept of mean delay. But the limitation of mean delay is that it 559 doesn't give information about the delay values distribution for the 560 duration of the block. Additionally it may be useful to have not 561 only the mean delay but also the minimum, maximum and median delay 562 values and, in wider terms, to know more about the statistic 563 distribution of delay values. So in order to have more information 564 about the delay and to overcome out of order issues, a different 565 approach can be introduced: it is based on double marking 566 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 Mean delay is calculated on all the packets of a sample and is a 588 simple computation to be performed for single marking method. In 589 some cases the mean delay measure is not sufficient to characterize 590 the sample, and more statistics of delay extent data are needed, e.g. 591 percentiles, variance and median delay values. The conventional 592 range (maximum-minimum) should be avoided for several reasons, 593 including stability of the maximum delay due to the influence by 594 outliers. RFC 5481 [RFC5481] section 6.5 highlights how the 99.9th 595 percentile of delay and delay variation is more helpful to 596 performance planners. To overcome this drawback the idea is to 597 couple the mean delay measure for the entire batch with double 598 marking method, where a subset of batch packets are selected for 599 extensive delay calculation by using a second marking. In this way 600 it is possible to perform a detailed analysis on these double marked 601 packets. Please note that there are classic algorithms for median 602 and variance calculation, but are out of the scope of this document. 603 The comparison between the mean delay for the entire batch and the 604 mean delay on these double marked packets gives an useful information 605 since it is possible to understand if the double marking measurements 606 are actually representative of the delay trends. 608 3.3. Delay variation measurement 610 Similarly to one-way delay measurement (both for single marking and 611 double marking), the method can also be used to measure the inter- 612 arrival jitter. We refer to the definition in RFC 3393 [RFC3393]. 613 The alternation of colors, for single marking method, can be used as 614 a time reference to measure delay variations. In case of double 615 marking, the time reference is given by the second marked packets. 616 Considering the example depicted in Figure 2, R1 stores a timestamp 617 TS(A)R1 whenever it sends the first packet of a block and R2 stores a 618 timestamp TS(B)R2 whenever it receives the first packet of a block. 619 The inter-arrival jitter can be easily derived from one-way delay 620 measurement, by evaluating the delay variation of consecutive 621 samples. 623 The concept of mean delay can also be applied to delay variation, by 624 evaluating the average variation of the interval between consecutive 625 packets of the flow from R1 to R2. 627 4. Considerations 629 This section highlights some considerations about the methodology. 631 4.1. Synchronization 633 The Alternate Marking technique does not require a strong 634 synchronization, especially for packet loss and two-way delay 635 measurement. Only one-way delay measurement requires network devices 636 to have synchronized clocks. 638 The color switching is the reference for all the network devices, and 639 the only requirement to be achieved is that all network devices have 640 to recognize the right batch along the path. 642 If the length of the measurement period is L time units, then all 643 network devices must be synchronized to the same clock reference with 644 an accuracy of +/- L/2 time units (without considering network 645 delay). This level of accuracy guarantees that all network devices 646 consistently match the color bit to the correct block. For example, 647 if the color is toggeled every second (L = 1 second), then clocks 648 must be synchronized with an accuracy of +/- 0.5 second to a common 649 time reference. 651 This synchronization requirement can be satisfied even with a 652 relatively inaccurate synchronization method. This is true for 653 packet loss and two-way delay measurement, instead, for one-way delay 654 measurement clock synchronization must be accurate. 656 Therefore, a system that uses only packet loss and two-way delay 657 measurement does not require synchronization. This is because the 658 value of the clocks of network devices does not affect the 659 computation of the two-way delay measurement. 661 4.2. Data Correlation 663 Data Correlation is the mechanism to compare counters and timestamps 664 for packet loss, delay and delay variation calculation. It could be 665 performed in several ways depending on the alternate marking 666 application and use case. 668 o A possibility is to use a centralized solution using Network 669 Management System (NMS) to correlate data; 671 o Another possibility is to define a protocol based distributed 672 solution, by defining a new protocol or by extending the existing 673 protocols (e.g. RFC6374, TWAMP, OWAMP) in order to communicate 674 the counters and timestamps between nodes. 676 In the following paragraphs an example data correlation mechanism is 677 explained and could be use independently of the adopted solutions. 679 When data is collected on the upstream and downstream node, e.g., 680 packet counts for packet loss measurement or timestamps for packet 681 delay measurement, and periodically reported to or pulled by other 682 nodes or NMS, a certain data correlation mechanism SHOULD be in use 683 to help the nodes or NMS to tell whether any two or more packet 684 counts are related to the same block of markers, or any two 685 timestamps are related to the same marked packet. 687 The alternate marking method described in this document literally 688 split the packets of the measured flow into different measurement 689 blocks, in addition a Block Number could be assigned to each of such 690 measurement block. The BN is generated each time a node reads the 691 data (packet counts or timestamps), and is associated with each 692 packet count and timestamp reported to or pulled by other nodes or 693 NMS. The value of BN could be calculated as the modulo of the local 694 time (when the data are read) and the interval of the marking time 695 period. 697 When the nodes or NMS see, for example, same BNs associated with two 698 packet counts from an upstream and a downstream node respectively, it 699 considers that these two packet counts corresponding to the same 700 block, i.e. that these two packet counts belong to the same block of 701 markers from the upstream and downstream node. The assumption of 702 this BN mechanism is that the measurement nodes are time 703 synchronized. This requires the measurement nodes to have a certain 704 time synchronization capability (e.g., the Network Time Protocol 705 (NTP) [RFC5905], or the IEEE 1588 Precision Time Protocol (PTP) 706 [IEEE1588]). Synchronization aspects are further discussed in 707 Section 4. 709 4.3. Packet Re-ordering 711 Due to ECMP, packet re-ordering is very common in IP network. The 712 accuracy of marking based PM, especially packet loss measurement, may 713 be affected by packet re-ordering. Take a look at the following 714 example: 716 Block : 1 | 2 | 3 | 4 | 5 |... 717 --------|---------|---------|---------|---------|---------|--- 718 Node R1 : AAAAAAA | BBBBBBB | AAAAAAA | BBBBBBB | AAAAAAA |... 719 Node R2 : AAAAABB | AABBBBA | AAABAAA | BBBBBBA | ABAAABA |... 721 Figure 5: Packet Reordering 723 In the following paragraphs an example of data correlation mechanism 724 is explained and could be use independently of the adopted solutions. 726 Most of the packet re-ordering occur at the edge of adjacent blocks, 727 and they are easy to handle if the interval of each block is 728 sufficient large. Then, it can assume that the packets with 729 different marker belong to the block that they are more close to. If 730 the interval is small, it is difficult and sometime impossible to 731 determine to which block a packet belongs. See above example, the 732 packet with the marker of "B" in block 3, there is no safe way to 733 tell whether the packet belongs to block 2 or block 4. 735 To choose a proper interval is important and how to choose a proper 736 interval is out of the scope of this document. But an implementation 737 SHOULD provide a way to configure the interval and allow a certain 738 degree of packet re-ordering. 740 5. Implementation and deployment 742 The methodology described in the previous sections can be applied in 743 various situations. Basically Alternate Marking technique could be 744 used in many cases for performance measurement. The only requirement 745 is to select and mark the flow to be monitored; in this way packets 746 are batched by the sender and each batch is alternately marked such 747 that can be easily recognized by the receiver. 749 An example of implementation and deployment is explained in the next 750 section, just to clarify how the method can work. 752 5.1. Report on the operational experiment at Telecom Italia 754 The method described in this document, also called PNPM (Packet 755 Network Performance Monitoring), has been invented and engineered in 756 Telecom Italia and it's currently being used in Telecom Italia's 757 network. The methodology has been applied by leveraging functions 758 and tools available on IP routers and it's currently being used to 759 monitor packet loss in some portions of Telecom Italia's network. 760 The application of the method to delay measurement is currently being 761 evaluated in Telecom Italia's labs. This section describes how the 762 features currently available on existing routing platforms can be 763 used to apply the method, in order to give an example of 764 implementation and deployment. 766 The fundamental steps for this implementation of the method can be 767 summarized in the following items: 769 o coloring the packets; 771 o counting the packets; 773 o collecting data and calculating the packet loss. 775 o metric transparency. 777 Before going deeper into the implementation details, it's worth 778 mentioning two different strategies that can be used when 779 implementing the method: 781 o flow-based: the flow-based strategy is used when only a limited 782 number of traffic flows need to be monitored. This could be the 783 case, for example, of IPTV channels or other specific applications 784 traffic with high QoS requirements (i.e. Mobile Backhauling 785 traffic). According to this strategy, only a subset of the flows 786 is colored. Counters for packet loss measurements can be 787 instantiated for each single flow, or for the set as a whole, 788 depending on the desired granularity. A relevant problem with 789 this approach is the necessity to know in advance the path 790 followed by flows that are subject to measurement. Path rerouting 791 and traffic load-balancing increase the issue complexity, 792 especially for unicast traffic. The problem is easier to solve 793 for multicast traffic where load balancing is seldom used, 794 especially for IPTV traffic where static joins are frequently used 795 to force traffic forwarding and replication. Another application 796 is on Mobile Backhauling, implemented with a VPN MPLS in Telecom 797 Italia's network; in this case the problem with unicast traffic is 798 overcome by monitoring just the two Provider Edge nodes of the VPN 799 MPLS. 801 o link-based: measurements are performed on all the traffic on a 802 link by link basis. The link could be a physical link or a 803 logical link (for instance an Ethernet VLAN or a MPLS PW). 804 Counters could be instantiated for the traffic as a whole or for 805 each traffic class (in case it is desired to monitor each class 806 separately), but in the second case a couple of counters is needed 807 for each class. 809 The current implementation in Telecom Italia uses the first strategy. 810 As mentioned, the flow-based measurement requires the identification 811 of the flow to be monitored and the discovery of the path followed by 812 the selected flow. It is possible to monitor a single flow or 813 multiple flows grouped together, but in this case measurement is 814 consistent only if all the flows in the group follow the same path. 815 Moreover, a Service Provider should be aware that, if a measurement 816 is performed by grouping many flows, it is not possible to determine 817 exactly which flow was affected by packets loss. In order to have 818 measures per single flow it is necessary to configure counters for 819 each specific flow. Once the flow(s) to be monitored have been 820 identified, it is necessary to configure the monitoring on the proper 821 nodes. Configuring the monitoring means configuring the policy to 822 intercept the traffic and configuring the counters to count the 823 packets. To have just an end-to-end monitoring, it is sufficient to 824 enable the monitoring on the first and the last hop routers of the 825 path: the mechanism is completely transparent to intermediate nodes 826 and independent from the path followed by traffic flows. On the 827 contrary, to monitor the flow on a hop-by-hop basis along its whole 828 path it is necessary to enable the monitoring on every node from the 829 source to the destination. In case the exact path followed by the 830 flow is not known a priori (i.e. the flow has multiple paths to reach 831 the destination) it is necessary to enable the monitoring system on 832 every path: counters on interfaces traversed by the flow will report 833 packet count, counters on other interfaces will be null. 835 5.1.1. Coloring the packets 837 The coloring operation is fundamental in order to create packet 838 blocks. This implies choosing where to activate the coloring and how 839 to color the packets. 841 In case of flow-based measurements, it is desirable, in general, to 842 have a single coloring node because it is easier to manage and 843 doesn't rise any risk of conflict (consider the case where two nodes 844 color the same flow). Thus it is necessary to color the flow as 845 close as possible to the source. In addition, coloring a flow close 846 to the source allows an end-to-end measure if a measurement point is 847 enabled on the last-hop router as well. The only requirement is that 848 the coloring must change periodically and every node along the path 849 must be able to identify unambiguously the colored packets. For 850 link-based measurements, all traffic needs to be colored when 851 transmitted on the link. If the traffic had already been colored, 852 then it has to be re-colored because the color must be consistent on 853 the link. This means that each hop along the path must (re-)color 854 the traffic; the color is not required to be consistent along 855 different links. 857 Traffic coloring can be implemented by setting a specific bit in the 858 packet header and changing the value of that bit periodically. With 859 current router implementations, only QoS related fields and features 860 offer the required flexibility to set bits in the packet header. In 861 case a Service Provider only uses the three most significant bits of 862 the DSCP field (corresponding to IP Precedence) for QoS 863 classification and queuing, it is possible to use the two less 864 significant bits of the DSCP field (bit 0 and bit 1) to implement the 865 method without affecting QoS policies. One of the two bits (bit 0) 866 could be used to identify flows subject to traffic monitoring (set to 867 1 if the flow is under monitoring, otherwise it is set to 0), while 868 the second (bit 1) can be used for coloring the traffic (switching 869 between values 0 and 1, corresponding to color A and B) and creating 870 the blocks. 872 In practice, coloring the traffic using the DSCP field can be 873 implemented by configuring on the router output interface an access 874 list that intercepts the flow(s) to be monitored and applies to them 875 a policy that sets the DSCP field accordingly. Since traffic 876 coloring has to be switched between the two values over time, the 877 policy needs to be modified periodically: an automatic script ca be 878 used perform this task on the basis of a fixed timer. In Telecom 879 Italia's implementation this timer is set to 5 minutes: this value 880 showed to be a good compromise between measurement frequency and 881 stability of the measurement (i.e. possibility to collect all the 882 measures referring to the same block). 884 5.1.2. Counting the packets 886 Assuming that the coloring of the packets is performed only by the 887 source node, the nodes between source and destination (included) have 888 to count the colored packets that they receive and forward: this 889 operation can be enabled on every router along the path or only on a 890 subset, depending on which network segment is being monitored (a 891 single link, a particular metro area, the backbone, the whole path). 893 Since the color switches periodically between two values, two 894 counters (one for each value) are needed: one counter for packets 895 with color A and one counter for packets with color B. For each flow 896 (or group of flows) being monitored and for every interface where the 897 monitoring is active, a couple od counters is needed. For example, 898 in order to monitor separately 3 flows on a router with 4 interfaces 899 involved, 24 counters are needed (2 counters for each of the 3 flows 900 on each of the 4 interfaces). If traffic is colored using the DSCP 901 field, as in Telecom Italia's implementation, an access-list that 902 matches specific DSCP values can be used to count the packets of the 903 flow(s) being monitored. 905 In case of link-based measurements the behaviour is similar except 906 that coloring and counting operations are performed on a link by link 907 basis at each endpoint of the link. 909 Another important aspect to take into consideration is when to read 910 the counters: in order to count the exact number of packets of a 911 block the routers must perform this operation when that block has 912 ended: in other words, the counter for color A must be read when the 913 current block has color B, in order to be sure that the value of the 914 counter is stable. This task can be accomplished in two ways. The 915 general approach suggests to read the counters periodically, many 916 times during a block duration, and to compare these successive 917 readings: when the counter stops incrementing means that the current 918 block has ended and its value can be elaborated safely. 919 Alternatively, if the coloring operation is performed on the basis of 920 a fixed timer, it is possible to configure the reading of the 921 counters according to that timer: for example, if each block is 5 922 minutes long, reading the counter for color A every 5 minute in the 923 middle of the subsequent block (with color B) is a safe choice. A 924 sufficient margin should be considered between the end of a block and 925 the reading of the counter, in order to take into account any out-of- 926 order packets. The choice of a 5 minutes timer for colore switching 927 was also inspired by these considerations. 929 5.1.3. Collecting data and calculating packet loss 931 The nodes enabled to perform performance monitoring collect the value 932 of the counters, but they are not able to directly use this 933 information to measure packet loss, because they only have their own 934 samples. For this reason, an external Network Management System 935 (NMS) is required to collect and elaborate data and to perform packet 936 loss calculation. The NMS compares the values of counters from 937 different nodes and can calculate if some packets were lost (even a 938 single packet) and also where packets were lost. 940 The value of the counters needs to be transmitted to the NMS as soon 941 as it has been read. This can be accomplished by using SNMP or FTP 942 and can be done in Push Mode or Polling Mode. In the first case, 943 each router periodically sends the information to the NMS, in the 944 latter case it is the NMS that periodically polls routers to collect 945 information. In any case, the NMS has to collect all the relevant 946 values from all the routers within one cycle of the timer (5 947 minutes). 949 If link-based measurement is used, it would be possible to use a 950 protocol to exchange values of counters between the two endpoints in 951 order to let them perform the packet loss calculation for each 952 traffic direction. A similar approach could be complicated if 953 applied to a flow-based measurement. 955 5.1.4. Metric transparency 957 In Telecom Italia's implementation the source node colors the packets 958 with a policy that is modified periodically via an automatic script 959 in order to alternate the DSCP field of the packets. The nodes 960 between source and destination (included) have to count with an 961 access-list the colored packets that they receive and forward. 963 Moreover the destination node has an important role: the colored 964 packets are intercepted and a policy restores and sets the DSCP field 965 of all the packets to the initial value. In this way the metric is 966 transparent because outside the section of the network under 967 monitoring the traffic flow is unchanged. 969 In such a case, thanks to this restoring technique, network elements 970 outside the Alternate Marking monitoring domain (e.g. the two 971 Provider Edge nodes of the Mobile Backhauling VPN MPLS) are totally 972 anaware that packets were marked. So this restoring technique makes 973 Alternate Marking completely transparent outside its monitoring 974 domain. 976 5.2. IP flow performance measurement (IPFPM) 978 This application of marking method is described in 979 [I-D.chen-ippm-coloring-based-ipfpm-framework]. 981 5.3. Performance Measurement Marking Method in BIER Domain 983 In [I-D.ietf-bier-mpls-encapsulation] two OAM bits from Bit Index 984 Explicit Replication (BIER) Header are reserved for the passive 985 performance measurement marking method. [I-D.ietf-bier-pmmm-oam] 986 details the measurement for multicast service over BIER domain. 988 5.4. Overlay OAM Passive Performance Measurement 990 The Overlay OAM Design Team is considering the preliminary OAM 991 requirements from NVO3, BIER, and SFC. Marking Method is the 992 preferred passive method to measure performance. 994 [I-D.ooamdt-rtgwg-ooam-requirement] and 995 [I-D.ooamdt-rtgwg-oam-gap-analysis] explain in deep this item. 997 5.5. RFC6374 Use Case 999 RFC6374 [RFC6374] uses the LM packet as the packet accounting 1000 demarcation point. Unfortunately this gives rise to a number of 1001 problems that may lead to significant packet accounting errors in 1002 certain situations. [I-D.ietf-mpls-flow-ident] discusses the desired 1003 capabilities for MPLS flow identification in order to perform a 1004 better in-band performance monitoring of user data packets. A method 1005 of accomplishing identification is Synonymous Flow Labels (SFL) 1006 introduced in [I-D.bryant-mpls-sfl-framework], while 1007 [I-D.bryant-mpls-rfc6374-sfl] describes RFC6374 performance 1008 measurements with SFL. 1010 5.6. Application to active performance measurement 1012 [I-D.fioccola-ippm-alt-mark-active] describes how to extend the 1013 existing Active Measurement Protocol, in order to implement alternate 1014 marking methodology. [I-D.fioccola-ippm-rfc6812-alt-mark-ext] 1015 describes an extension to the Cisco SLA Protocol Measurement-Type 1016 UDP-Measurement. 1018 6. Hybrid measurement 1020 The method has been explicitly designed for passive measurements but 1021 it can also be used with active measurements. In order to have both 1022 end to end measurements and intermediate measurements (hybrid 1023 measurements) two end points can exchanges artificial traffic flows 1024 and apply alternate marking over these flows. In the intermediate 1025 points artificial traffic is managed in the same way as real traffic 1026 and measured as specified before. So the application of marking 1027 method can simplify also the active measurement, as explained in 1028 [I-D.fioccola-ippm-alt-mark-active]. 1030 7. Compliance with RFC6390 guidelines 1032 RFC6390 [RFC6390] defines a framework and a process for developing 1033 Performance Metrics for protocols above and below the IP layer (such 1034 as IP-based applications that operate over reliable or datagram 1035 transport protocols). 1037 This document doesn't aim to propose a new Performance Metric but a 1038 new method of measurement for a few Performance Metrics that have 1039 already been standardized. Nevertheless, it's worth applying 1040 [RFC6390] guidelines to the present document, in order to provide a 1041 more complete and coherent description of the proposed method. We 1042 used a subset of the Performance Metric Definition template defined 1043 by [RFC6390]. 1045 o Metric name and description: as already stated, this document 1046 doesn't propose any new Performance Metric. On the contrary, it 1047 describes a novel method for measuring packet loss [RFC2680]. The 1048 same concept, with small differences, can also be used to measure 1049 delay [RFC2679], and jitter [RFC3393]. The document mainly 1050 describes the applicability to packet loss measurement. 1052 o Method of Measurement or Calculation: according to the method 1053 described in the previous sections, the number of packets lost is 1054 calculated by subtracting the value of the counter on the source 1055 node from the value of the counter on the destination node. Both 1056 counters must refer to the same color. The calculation is 1057 performed when the value of the counters is in a steady state. 1059 o Units of Measurement: the method calculates and reports the exact 1060 number of packets sent by the source node and not received by the 1061 destination node. 1063 o Measurement Points: the measurement can be performed between 1064 adjacent nodes, on a per-link basis, or along a multi-hop path, 1065 provided that the traffic under measurement follows that path. In 1066 case of a multi-hop path, the measurements can be performed both 1067 end-to-end and hop-by-hop. 1069 o Measurement Timing: the method have a constraint on the frequency 1070 of measurements. In order to perform a measure, the counter must 1071 be in a steady state: this happens when the traffic is being 1072 colored with the alternate color; for example in the Telecom 1073 Italia application of the method the time interval is set to 5 1074 minutes. 1076 o Implementation: the Telecom Italia application of the method uses 1077 two encodings of the DSCP field to color the packets; this enables 1078 the use of policy configurations on the router to color the 1079 packets and accordingly configure the counter for each color. The 1080 path followed by traffic being measured should be known in advance 1081 in order to configure the counters along the path and be able to 1082 compare the correct values. 1084 o Use and Applications: the method can be used to measure packet 1085 loss with high precision on live traffic; moreover, by combining 1086 end-to-end and per-link measurements, the method is useful to 1087 pinpoint the single link that is experiencing loss events. 1089 o Reporting Model: the value of the counters has to be sent to a 1090 centralized management system that perform the calculations; such 1091 samples must contain a reference to the time interval they refer 1092 to, so that the management system can perform the correct 1093 correlation; the samples have to be sent while the corresponding 1094 counter is in a steady state (within a time interval), otherwise 1095 the value of the sample should be stored locally. 1097 o Dependencies: the values of the counters have to be correlated to 1098 the time interval they refer to; moreover, as far the Telecom 1099 Italia application of the method is based on DSCP values, there 1100 are significant dependencies on the usage of the DSCP field: it 1101 must be possible to rely on unused DSCP values without affecting 1102 QoS-related configuration and behavior; moreover, the intermediate 1103 nodes must not change the value of the DSCP field not to alter the 1104 measurement. 1106 o Organization of Results: the method of measurement produces 1107 singletons. 1109 o Parameters: currently, the main parameter of the method is the 1110 time interval used to alternate the colors and read the counters. 1112 8. Security Considerations 1114 This document specifies a method to perform measurements in the 1115 context of a Service Provider's network and has not been developed to 1116 conduct Internet measurements, so it does not directly affect 1117 Internet security nor applications which run on the Internet. 1118 However, implementation of this method must be mindful of security 1119 and privacy concerns. 1121 There are two types of security concerns: potential harm caused by 1122 the measurements and potential harm to the measurements. For what 1123 concerns the first point, the measurements described in this document 1124 are passive, so there are no packets injected into the network 1125 causing potential harm to the network itself and to data traffic. 1126 Nevertheless, the method implies modifications on the fly to the IP 1127 header of data packets: this must be performed in a way that doesn't 1128 alter the quality of service experienced by packets subject to 1129 measurements and that preserve stability and performance of routers 1130 doing the measurements. The measurements themselves could be harmed 1131 by routers altering the marking of the packets, or by an attacker 1132 injecting artificial traffic. Authentication techniques, such as 1133 digital signatures, may be used where appropriate to guard against 1134 injected traffic attacks. 1136 The privacy concerns of network measurement are limited because the 1137 method only relies on information contained in the IP header without 1138 any release of user data. 1140 The measurement itself may be affected by routers (or other network 1141 devices) along the path of IP packets intentionally altering the 1142 value of marking bits of packets. As mentioned above, the mechanism 1143 specified in this document is just in the context of one Service 1144 Provider's network, and thus the routers (or other network devices) 1145 are locally administered and this type of attack can be avoided. 1147 One of the main security threats in OAM protocols is network 1148 reconnaissance; an attacker can gather information about the network 1149 performance by passively eavesdropping to OAM messages. The 1150 advantage of the methods described in this document is that the 1151 marking bits are the only information that is exchanged between the 1152 network devices. Therefore, passive eavesdropping to data plane 1153 traffic does not allow attackers to gain information about the 1154 network performance. 1156 Delay attacks are another potential threat in the context of this 1157 document. Delay measurement is performed using a specific packet in 1158 each block, marked by a dedicated color bit. Therefore, a man-in- 1159 the-middle attacker can selectively induce synthetic delay only to 1160 delay-colored packets, causing systematic error in the delay 1161 measurements. As discussed in previous sections, the methods 1162 described in this document rely on an underlying time synchronization 1163 protocol. Thus, by attacking the time protocol an attacker can 1164 potentially compromise the integrity of the measurement. A detailed 1165 discussion about the threats against time protocols and how to 1166 mitigate them is presented in RFC 7384 [RFC7384]. 1168 9. Conclusions 1170 The advantages of the method described in this document are: 1172 o easy implementation: it can be implemented using features already 1173 available on major routing platforms; 1175 o low computational effort: the additional load on processing is 1176 negligible; 1178 o accurate packet loss measurement: single packet loss granularity 1179 is achieved with a passive measurement; 1181 o potential applicability to any kind of packet/frame -based 1182 traffic: Ethernet, IP, MPLS, etc., both unicast and multicast; 1184 o robustness: the method can tolerate out of order packets and it's 1185 not based on "special" packets whose loss could have a negative 1186 impact; 1188 o no interoperability issues: the features required to implement the 1189 method are available on all current routing platforms. 1191 The method doesn't raise any specific need for protocol extension, 1192 but it could be further improved by means of some extension to 1193 existing protocols. Specifically, the use of DiffServ bits for 1194 coloring the packets could not be a viable solution in some cases: a 1195 standard method to color the packets for this specific application 1196 could be beneficial. 1198 10. IANA Considerations 1200 There are no IANA actions required. 1202 11. Acknowledgements 1204 The previous IETF drafts about this technique were: 1205 [I-D.cociglio-mboned-multicast-pm] and [I-D.tempia-opsawg-p3m]. 1206 There are some references to this methodology in other IETF works 1207 (e.g. [I-D.ietf-mpls-flow-ident], [I-D.bryant-mpls-sfl-framework] 1208 [I-D.bryant-mpls-rfc6374-sfl], [I-D.ietf-bier-mpls-encapsulation], 1209 [I-D.ietf-bier-pmmm-oam] 1210 [I-D.chen-ippm-coloring-based-ipfpm-framework]). 1212 In addition the authors would like to thank Domenico Laforgia, 1213 Daniele Accetta and Mario Bianchetti for their contribution to the 1214 definition and the implementation of the method. 1216 12. References 1218 12.1. Normative References 1220 [RFC2679] Almes, G., Kalidindi, S., and M. Zekauskas, "A One-way 1221 Delay Metric for IPPM", RFC 2679, DOI 10.17487/RFC2679, 1222 September 1999, . 1224 [RFC2680] Almes, G., Kalidindi, S., and M. Zekauskas, "A One-way 1225 Packet Loss Metric for IPPM", RFC 2680, 1226 DOI 10.17487/RFC2680, September 1999, 1227 . 1229 [RFC3393] Demichelis, C. and P. Chimento, "IP Packet Delay Variation 1230 Metric for IP Performance Metrics (IPPM)", RFC 3393, 1231 DOI 10.17487/RFC3393, November 2002, 1232 . 1234 12.2. Informative References 1236 [I-D.bryant-mpls-rfc6374-sfl] 1237 Bryant, S., Chen, M., Li, Z., Swallow, G., Sivabalan, S., 1238 Mirsky, G., and G. Fioccola, "RFC6374 Synonymous Flow 1239 Labels", draft-bryant-mpls-rfc6374-sfl-03 (work in 1240 progress), October 2016. 1242 [I-D.bryant-mpls-sfl-framework] 1243 Bryant, S., Chen, M., Li, Z., Swallow, G., Sivabalan, S., 1244 and G. Mirsky, "Synonymous Flow Label Framework", draft- 1245 bryant-mpls-sfl-framework-02 (work in progress), October 1246 2016. 1248 [I-D.chen-ippm-coloring-based-ipfpm-framework] 1249 Chen, M., Zheng, L., Mirsky, G., Fioccola, G., and T. 1250 Mizrahi, "IP Flow Performance Measurement Framework", 1251 draft-chen-ippm-coloring-based-ipfpm-framework-06 (work in 1252 progress), March 2016. 1254 [I-D.cociglio-mboned-multicast-pm] 1255 Cociglio, M., Capello, A., Bonda, A., and L. Castaldelli, 1256 "A method for IP multicast performance monitoring", draft- 1257 cociglio-mboned-multicast-pm-01 (work in progress), 1258 October 2010. 1260 [I-D.fioccola-ippm-alt-mark-active] 1261 Fioccola, G., Clemm, A., Cociglio, M., Chandramouli, M., 1262 and A. Capello, "Alternate Marking Extension to Active 1263 Measurement Protocol", draft-fioccola-ippm-alt-mark- 1264 active-00 (work in progress), July 2016. 1266 [I-D.fioccola-ippm-rfc6812-alt-mark-ext] 1267 Fioccola, G., Clemm, A., Cociglio, M., Chandramouli, M., 1268 and A. Capello, "Alternate Marking Extension to Cisco SLA 1269 Protocol RFC6812", draft-fioccola-ippm-rfc6812-alt-mark- 1270 ext-01 (work in progress), March 2016. 1272 [I-D.ietf-bier-mpls-encapsulation] 1273 Wijnands, I., Rosen, E., Dolganow, A., Tantsura, J., 1274 Aldrin, S., and I. Meilik, "Encapsulation for Bit Index 1275 Explicit Replication in MPLS and non-MPLS Networks", 1276 draft-ietf-bier-mpls-encapsulation-06 (work in progress), 1277 December 2016. 1279 [I-D.ietf-bier-pmmm-oam] 1280 Mirsky, G., Zheng, L., Chen, M., and G. Fioccola, 1281 "Performance Measurement (PM) with Marking Method in Bit 1282 Index Explicit Replication (BIER) Layer", draft-ietf-bier- 1283 pmmm-oam-01 (work in progress), January 2017. 1285 [I-D.ietf-mpls-flow-ident] 1286 Bryant, S., Pignataro, C., Chen, M., Li, Z., and G. 1287 Mirsky, "MPLS Flow Identification Considerations", draft- 1288 ietf-mpls-flow-ident-04 (work in progress), February 2017. 1290 [I-D.ooamdt-rtgwg-oam-gap-analysis] 1291 Mirsky, G., Nordmark, E., Pignataro, C., Kumar, N., Kumar, 1292 D., Chen, M., Yizhou, L., Mozes, D., Networks, J., and I. 1293 Bagdonas, "Operations, Administration and Maintenance 1294 (OAM) for Overlay Networks: Gap Analysis", draft-ooamdt- 1295 rtgwg-oam-gap-analysis-02 (work in progress), July 2016. 1297 [I-D.ooamdt-rtgwg-ooam-requirement] 1298 Kumar, N., Pignataro, C., Kumar, D., Mirsky, G., Chen, M., 1299 Nordmark, E., Networks, J., and D. Mozes, "Overlay OAM 1300 Requirements", draft-ooamdt-rtgwg-ooam-requirement-02 1301 (work in progress), January 2017. 1303 [I-D.tempia-opsawg-p3m] 1304 Capello, A., Cociglio, M., Castaldelli, L., and A. Bonda, 1305 "A packet based method for passive performance 1306 monitoring", draft-tempia-opsawg-p3m-04 (work in 1307 progress), February 2014. 1309 [RFC5481] Morton, A. and B. Claise, "Packet Delay Variation 1310 Applicability Statement", RFC 5481, DOI 10.17487/RFC5481, 1311 March 2009, . 1313 [RFC6374] Frost, D. and S. Bryant, "Packet Loss and Delay 1314 Measurement for MPLS Networks", RFC 6374, 1315 DOI 10.17487/RFC6374, September 2011, 1316 . 1318 [RFC6390] Clark, A. and B. Claise, "Guidelines for Considering New 1319 Performance Metric Development", BCP 170, RFC 6390, 1320 DOI 10.17487/RFC6390, October 2011, 1321 . 1323 [RFC6703] Morton, A., Ramachandran, G., and G. Maguluri, "Reporting 1324 IP Network Performance Metrics: Different Points of View", 1325 RFC 6703, DOI 10.17487/RFC6703, August 2012, 1326 . 1328 [RFC7276] Mizrahi, T., Sprecher, N., Bellagamba, E., and Y. 1329 Weingarten, "An Overview of Operations, Administration, 1330 and Maintenance (OAM) Tools", RFC 7276, 1331 DOI 10.17487/RFC7276, June 2014, 1332 . 1334 [RFC7384] Mizrahi, T., "Security Requirements of Time Protocols in 1335 Packet Switched Networks", RFC 7384, DOI 10.17487/RFC7384, 1336 October 2014, . 1338 Authors' Addresses 1340 Giuseppe Fioccola (editor) 1341 Telecom Italia 1342 Via Reiss Romoli, 274 1343 Torino 10148 1344 Italy 1346 Email: giuseppe.fioccola@telecomitalia.it 1348 Alessandro Capello (editor) 1349 Telecom Italia 1350 Via Reiss Romoli, 274 1351 Torino 10148 1352 Italy 1354 Email: alessandro.capello@telecomitalia.it 1356 Mauro Cociglio 1357 Telecom Italia 1358 Via Reiss Romoli, 274 1359 Torino 10148 1360 Italy 1362 Email: mauro.cociglio@telecomitalia.it 1363 Luca Castaldelli 1364 Telecom Italia 1365 Via Reiss Romoli, 274 1366 Torino 10148 1367 Italy 1369 Email: luca.castaldelli@telecomitalia.it 1371 Mach(Guoyi) Chen (editor) 1372 Huawei Technologies 1374 Email: mach.chen@huawei.com 1376 Lianshu Zheng (editor) 1377 Huawei Technologies 1379 Email: vero.zheng@huawei.com 1381 Greg Mirsky (editor) 1382 ZTE 1383 USA 1385 Email: gregimirsky@gmail.com 1387 Tal Mizrahi (editor) 1388 Marvell 1389 6 Hamada st. 1390 Yokneam 1391 Israel 1393 Email: talmi@marvell.com