idnits 2.17.1 draft-fioccola-ippm-multipoint-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 abstract seems to contain references ([RFC8321], [RFC5644]), which it shouldn't. Please replace those with straight textual mentions of the documents in question. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year == The document doesn't use any RFC 2119 keywords, yet seems to have RFC 2119 boilerplate text. -- The document date (June 29, 2018) is 2090 days in the past. Is this intentional? Checking references for intended status: Experimental ---------------------------------------------------------------------------- ** Obsolete normative reference: RFC 8321 (Obsoleted by RFC 9341) == Outdated reference: A later version (-05) exists of draft-mizrahi-ippm-compact-alternate-marking-01 Summary: 2 errors (**), 0 flaws (~~), 3 warnings (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 IPPM Working Group G. Fioccola, Ed. 3 Internet-Draft M. Cociglio 4 Intended status: Experimental Telecom Italia 5 Expires: December 31, 2018 A. Sapio 6 R. Sisto 7 Politecnico di Torino 8 June 29, 2018 10 Multipoint Alternate Marking method for passive and hybrid performance 11 monitoring 12 draft-fioccola-ippm-multipoint-alt-mark-04 14 Abstract 16 The Alternate Marking method, as presented in RFC 8321 [RFC8321], can 17 be applied only to point-to-point flows because it assumes that all 18 the packets of the flow measured on one node are measured again by a 19 single second node. This document aims to generalize and expand this 20 methodology to measure any kind of unicast flows, whose packets can 21 follow several different paths in the network, in wider terms a 22 multipoint-to-multipoint network. For this reason the technique here 23 described is called Multipoint Alternate Marking. Some definitions 24 here introduced extend the scope of RFC 5644 [RFC5644] in the context 25 of alternate marking schema. 27 Requirements Language 29 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 30 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 31 document are to be interpreted as described in RFC 2119 [RFC2119]. 33 Status of This Memo 35 This Internet-Draft is submitted in full conformance with the 36 provisions of BCP 78 and BCP 79. 38 Internet-Drafts are working documents of the Internet Engineering 39 Task Force (IETF). Note that other groups may also distribute 40 working documents as Internet-Drafts. The list of current Internet- 41 Drafts is at https://datatracker.ietf.org/drafts/current/. 43 Internet-Drafts are draft documents valid for a maximum of six months 44 and may be updated, replaced, or obsoleted by other documents at any 45 time. It is inappropriate to use Internet-Drafts as reference 46 material or to cite them other than as "work in progress." 48 This Internet-Draft will expire on December 31, 2018. 50 Copyright Notice 52 Copyright (c) 2018 IETF Trust and the persons identified as the 53 document authors. All rights reserved. 55 This document is subject to BCP 78 and the IETF Trust's Legal 56 Provisions Relating to IETF Documents 57 (https://trustee.ietf.org/license-info) in effect on the date of 58 publication of this document. Please review these documents 59 carefully, as they describe your rights and restrictions with respect 60 to this document. Code Components extracted from this document must 61 include Simplified BSD License text as described in Section 4.e of 62 the Trust Legal Provisions and are provided without warranty as 63 described in the Simplified BSD License. 65 Table of Contents 67 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2 68 2. Correlation with RFC5644 . . . . . . . . . . . . . . . . . . 4 69 3. Flow classification . . . . . . . . . . . . . . . . . . . . . 4 70 4. Multipoint Performance Measurement . . . . . . . . . . . . . 6 71 4.1. Monitoring Network . . . . . . . . . . . . . . . . . . . 7 72 5. Multipoint Packet Loss . . . . . . . . . . . . . . . . . . . 8 73 6. Network Clustering . . . . . . . . . . . . . . . . . . . . . 9 74 6.1. Algorithm for Cluster partition . . . . . . . . . . . . . 10 75 7. Timing Aspects . . . . . . . . . . . . . . . . . . . . . . . 12 76 8. Multipoint Delay and Delay Variation . . . . . . . . . . . . 14 77 8.1. Delay measurements on multipoint paths basis . . . . . . 14 78 8.1.1. Single Marking measurement . . . . . . . . . . . . . 14 79 8.2. Delay measurements on single packets basis . . . . . . . 14 80 8.2.1. Single and Double Marking measurement . . . . . . . . 14 81 8.2.2. Hashing selection method . . . . . . . . . . . . . . 15 82 9. An SDN enabled Performance Management . . . . . . . . . . . . 17 83 10. Examples of application . . . . . . . . . . . . . . . . . . . 17 84 11. Security Considerations . . . . . . . . . . . . . . . . . . . 18 85 12. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 18 86 13. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 18 87 14. References . . . . . . . . . . . . . . . . . . . . . . . . . 18 88 14.1. Normative References . . . . . . . . . . . . . . . . . . 18 89 14.2. Informative References . . . . . . . . . . . . . . . . . 18 90 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 19 92 1. Introduction 94 The alternate marking method, as presented until now, is applicable 95 to a point-to-point path; so the extension proposed in this document 96 explains the most general case of multipoint-to-multipoint path and 97 enables flexible and adaptive performance measurements in a managed 98 network. 100 The Alternate Marking methodology described in RFC 8321 [RFC8321] has 101 the property to synchronize measurements in different points 102 maintaining the coherence of the counters. So it is possible to show 103 what is happening in every marking period for each monitored flow. 104 The monitoring parameters are the packet counter and timestamps of a 105 flow for each marking period. 107 There are some applications of the alternate marking method where 108 there are a lot of monitored flows and nodes. Multipoint Alternate 109 Marking aims to reduce these values and makes the performance 110 monitoring more flexible in case a detailed analysis is not needed. 111 For instance, by considering n measurement points and m monitored 112 flows,the order of magnitude of the packet counters for each time 113 interval is n*m*2 (1 per color). If both n and m are high values the 114 packet counters increase a lot and Multipoint Alternate Marking 115 offers a tool to control these parameters. 117 The approach presented in this document is applied only to unicast 118 flows and not to multicast. BUM (Boradcast Unkown Unicast Multicast) 119 traffic is not considered here, because traffic replication is not 120 covered by the Multipoint Alternate Marking method. 122 Alternate Marking method works by definition for multipoint to 123 multipoint paths but the network clustering approach presented in 124 this document is the formalization of how to implement this property 125 and it allows a flexible and optimized performance measurement 126 support. 128 Without network clustering, it is possible to apply alternate marking 129 only for all the network or per single flow. Instead, with network 130 clustering, it is possible to use the network clusters partition at 131 different levels to perform the needed degree of detail. In some 132 circumstances it is possible to monitor a Multipoint Network by 133 analyzing the Network Clustering, without examining in depth. In 134 case of problems (packet loss is measured or the delay is too high) 135 the filtering criteria could be specified more in order to perform a 136 detailed analysis by using a different combination of clusters up to 137 a per-flow measurement as described in RFC 8321 [RFC8321]. 139 An application could be the Software Defined Network (SDN) paradigm 140 where the SDN Controllers are the brains of the network and can 141 manage flow control to the switches and routers and, in the same way, 142 can calibrate the performance measurements depending on the 143 necessity. An SDN Controller Application can orchestrate how deep 144 the network performance monitoring is setup. 146 2. Correlation with RFC5644 148 RFC 5644 [RFC5644] is limited to active measurements using a single 149 source packet or stream, and observations of corresponding packets 150 along the path (spatial), at one or more destinations (one-to-group), 151 or both. Instead, the scope of this memo is to define multiparty 152 metrics for passive and hybrid measurements in a group-to-group 153 topology with multiple sources and destinations. 155 RFC 5644 [RFC5644] introduces metric names that can be reused also 156 here but have to be extended and rephrased to be applied to the 157 alternate marking schema: 159 a. the multiparty metrics are not only one-to-group metrics but can 160 be also group-to-group metrics; 162 b. the spatial metrics, used for measuring the performance of 163 segments of a source to destination path, are applied here to 164 group-to-group segments (called Clusters). 166 3. Flow classification 168 An unicast flow is identified by all the packets having a set of 169 common characteristics. This definition is inspired by RFC 7011 170 [RFC7011]. 172 As an example, by considering a flow as all the packets sharing the 173 same source IP address or the same destination IP address, it is easy 174 to understand that the resulting pattern will not be a point-to-point 175 connection, but a point-to-multipoint or multipoint-to-point 176 connection. 178 In general a flow can be defined by a set of selection rules used to 179 match a subset of the packets processed by the network device. These 180 rules specify a set of headers fields (Identification Fields) and the 181 relative values that must be found in matching packets. 183 The choice of the identification fields directly affects the type of 184 paths that the flow would follow in the network. In fact, it is 185 possible to relate a set of identification fields with the pattern of 186 the resulting graphs, as listed in Figure 1. 188 A TCP 5-tuple usually identifies flows following either a single path 189 or a point-to-point multipath (in case of load balancing). On the 190 contrary, a single source address selects flows following a point-to- 191 multipoint, while a multipoint-to-point can be the result of a 192 matching on a single destination address. In case a selection rule 193 and its reverse are used for bidirectional measurements, they can 194 correspond to a point-to-multipoint in one direction and a 195 multipoint-to-point in the opposite direction. 197 In this way the flows to be monitored are selected into the 198 monitoring points using packet selection rules, that can also change 199 the pattern of the monitored network. 201 The alternate marking method is applicable only to a single path (and 202 partially to a one-to-one multipath), so the extension proposed in 203 this document is suitable also for the most general case of 204 multipoint-to-multipoint, which embraces all the other patterns of 205 Figure 1. 207 point-to-point single path 208 +------+ +------+ +------+ 209 ---<> R1 <>----<> R2 <>----<> R3 <>--- 210 +------+ +------+ +------+ 212 point-to-point multipath 213 +------+ 214 <> R2 <> 215 / +------+ \ 216 / \ 217 +------+ / \ +------+ 218 ---<> R1 <> <> R4 <>--- 219 +------+ \ / +------+ 220 \ / 221 \ +------+ / 222 <> R3 <> 223 +------+ 225 point-to-multipoint 226 +------+ 227 <> R4 <>--- 228 / +------+ 229 +------+ / 230 <> R2 <> 231 / +------+ \ 232 +------+ / \ +------+ 233 ---<> R1 <> <> R5 <>--- 234 +------+ \ +------+ 235 \ +------+ 236 <> R3 <> 237 +------+ \ 238 \ +------+ 239 <> R6 <>--- 240 +------+ 242 multipoint-to-point 243 +------+ 244 ---<> R1 <> 245 +------+ \ 246 \ +------+ 247 <> R4 <> 248 / +------+ \ 249 +------+ / \ +------+ 250 ---<> R2 <> <> R4 <>--- 251 +------+ / +------+ 252 +------+ / 253 <> R5 <> 254 / +------+ 255 +------+ / 256 ---<> R3 <> 257 +------+ 259 multipoint-to-multipoint 260 +------+ +------+ 261 ---<> R1 <> <> R6 <>--- 262 +------+ \ / +------+ 263 \ +------+ / 264 <> R4 <> 265 +------+ \ 266 +------+ \ +------+ 267 ---<> R2 <> <> R7 <>--- 268 +------+ \ / +------+ 269 \ +------+ / 270 <> R5 <> 271 / +------+ \ 272 +------+ / \ +------+ 273 ---<> R3 <> <> R8 <>--- 274 +------+ +------+ 276 Figure 1: Flow classification 278 4. Multipoint Performance Measurement 280 By Using the "traditional" alternate marking method only point-to- 281 point paths can be monitored. To have an IP (TCP/UDP) flow that 282 follows a point-to-point path we have to define, with a specific 283 value, 5 identification fields (IP Source, IP Destination, Transport 284 Protocol, Source Port, Destination Port). 286 Multipoint Alternate Marking enables the performance measurement for 287 multipoint flows selected by identification fields without any 288 constraints (even the entire network production traffic). It is also 289 possible to use multiple marking points for the same monitored flow. 291 4.1. Monitoring Network 293 The Monitoring Network is deduced from the Production Network, by 294 identifying the nodes of the graph that are the measurement points, 295 and the links that are the connections between measurement points. 297 There are some techniques that can help with the building of the 298 monitoring network (as an example it is possible to mention 299 [I-D.amf-ippm-route]). In general there are different options: the 300 monitoring network can be obtained by considering all the possible 301 paths for the traffic or also by checking the traffic sometimes and 302 update the graph consequently. 304 So a graph model of the monitoring network can be built according to 305 the alternate marking method: the monitored interfaces and links are 306 identified. Only the measurement points and links where the traffic 307 has flowed have to be represented in the graph. 309 The following figure shows a simple example of a Monitoring Network 310 graph: 312 +------+ 313 <> R6 <>--- 314 / +------+ 315 +------+ +------+ / 316 <> R2 <>---<> R4 <> 317 / +------+ \ +------+ \ 318 / \ \ +------+ 319 +------+ / +------+ \ +------+ <> R7 <>--- 320 ---<> R1 <>---<> R3 <>---<> R5 <> +------+ 321 +------+ \ +------+ \ +------+ \ 322 \ \ \ +------+ 323 \ \ <> R8 <>--- 324 \ \ +------+ 325 \ \ 326 \ \ +------+ 327 \ <> R9 <>--- 328 \ +------+ 329 \ 330 \ +------+ 331 <> R10 <>--- 332 +------+ 334 Figure 2: Monitoring Network Graph 336 Each monitoring point is characterized by the packet counter that 337 refers only to a marking period of the monitored flow. 339 The same is applicable also for the delay but it will be described in 340 the following sections. 342 5. Multipoint Packet Loss 344 Since all the packets of the considered flow leaving the network have 345 previously entered the network, the number of packets counted by all 346 the input nodes is always greater or equal than the number of packets 347 counted by all the output nodes. 349 And in case of no packet loss occurring in the marking period, if all 350 the input and output points of the network domain to be monitored are 351 measurement points, the sum of the number of packets on all the 352 ingress interfaces and on all the egress interfaces is the same. In 353 this circumstance, if no packet loss occurs, the intermediate 354 measurement points have only the task to split the measurement. 356 It is possible to define the Network Packet Loss (for 1 flow, for 1 357 period): <>. This is true for every packet 360 flow in each marking period. 362 The Monitored Network Packet Loss with n input nodes and m output 363 nodes is given by: 365 PL = (PI1 + PI2 +...+ PIn) - (PO1 + PO2 +...+ POm) 367 where: 369 PL is the Network Packet Loss (number of lost packets) 371 PIi is the Number of packets flowed through the i-th Input node in 372 this period 374 POj is the Number of packets flowed through the j-th Output node in 375 this period 377 The equation is applied on a per-time-interval basis. 379 6. Network Clustering 381 The previous Equation can determine the number of packets lost 382 globally in the monitored network, exploiting only the data provided 383 by the counters in the input and output nodes. 385 In addition it is also possible to leverage the data provided by the 386 other counters in the network to converge on the smallest 387 identifiable subnetworks where the losses occur. These subnetworks 388 are named Clusters. 390 A Cluster graph is a subnetwork of the entire Monitoring Network 391 graph that still satisfies the packet loss equation where PL in this 392 case is the number of packets lost in the Cluster. 394 For this reason a Cluster should contain all the arcs emanating from 395 its input nodes and all the arcs terminating at its output nodes. 396 This ensures that we can count all the packets (and only those) 397 exiting an input node again at the output node, whatever path they 398 follow. 400 In a completely monitored network (a network where every network 401 interface is monitored), each network device corresponds to a Cluster 402 and each physical link corresponds to two Clusters (one for each 403 direction). 405 Clusters can have different sizes depending on flow filtering 406 criteria adopted. 408 Moreover, sometimes Clusters can be optionally simplified. For 409 example when two monitored interfaces are divided by a single router 410 (one is the input interface and the other is the output interface and 411 the router has only these two interfaces), instead of counting 412 exactly twice, upon entering and leaving, it is possible to consider 413 a single measurement point (in this case we do not care of the 414 internal packet loss of the router). 416 6.1. Algorithm for Cluster partition 418 A simple algorithm can be applied in order to split our monitoring 419 network into Clusters. It is a two-step algorithm: 421 o Group the links where there is the same starting node; 423 o Join the grouped links with at least one ending node in common. 425 In our monitoring network graph example it is possible to identify 426 the Clusters partition by applying this two-step algorithm. 428 The first step identifies the following groups: 430 1. Group 1: (R1-R2), (R1-R3), (R1-R10) 432 2. Group 2: (R2-R4), (R2-R5) 434 3. Group 3: (R3-R5), (R3-R9) 436 4. Group 4: (R4-R6), (R4-R7) 438 5. Group 5: (R5-R8) 440 And then, the second step builds the Clusters partition (in 441 particular we can underline that Group 2 and Group 3 connect 442 together, since R5 is in common): 444 1. Cluster 1: (R1-R2), (R1-R3), (R1-R10) 446 2. Cluster 2: (R2-R4), (R2-R5), (R3-R5), (R3-R9) 448 3. Cluster 3: (R4-R6), (R4-R7) 450 4. Cluster 4: (R5-R8) 452 In the end the following 4 Clusters are obtained: 454 Cluster 1 455 +------+ 456 <> R2 <>--- 457 / +------+ 458 / 459 +------+ / +------+ 460 ---<> R1 <>---<> R3 <>--- 461 +------+ \ +------+ 462 \ 463 \ 464 \ 465 \ 466 \ 467 \ 468 \ 469 \ 470 \ +------+ 471 <> R10 <>--- 472 +------+ 474 Cluster 2 475 +------+ +------+ 476 ---<> R2 <>---<> R4 <>--- 477 +------+ \ +------+ 478 \ 479 +------+ \ +------+ 480 ---<> R3 <>---<> R5 <>--- 481 +------+ \ +------+ 482 \ 483 \ 484 \ 485 \ 486 \ +------+ 487 <> R9 <>--- 488 +------+ 490 Cluster 3 491 +------+ 492 <> R6 <>--- 493 / +------+ 494 +------+ / 495 ---<> R4 <> 496 +------+ \ 497 \ +------+ 498 <> R7 <>--- 499 +------+ 501 Cluster 4 502 +------+ 503 ---<> R5 <> 504 +------+ \ 505 \ +------+ 506 <> R8 <>--- 507 +------+ 509 Figure 3: Clusters example 511 There are Clusters with more than 2 nodes and two-nodes Clusters. In 512 the two-nodes Clusters the loss is on the link (Cluster 4). In more- 513 than-2-nodes Clusters the loss is on the Cluster but we cannot know 514 in which link (Cluster 1, 2, 3). 516 In this way the calculation of packet loss can be made on Cluster 517 basis. Note that CIR(Committed Information Rate) and EIR(Excess 518 Information Rate) can also be deduced on Cluster basis. 520 Obviously, by combining some Clusters in a new connected subnetwork 521 (called Super Cluster) the Packet Loss Rule is still true. 523 In this way in a very large network there is no need to configure 524 detailed filter criteria to inspect the traffic. You can check 525 multipoint network and only in case of problems you can go deep with 526 a step-by-step cluster analysis, but only for the cluster or 527 combination of clusters where the problem happens. 529 7. Timing Aspects 531 The mark switching approach based on a fixed timer is considered in 532 this document. 534 So, if we analyze a multipoint-to-multipoint path with more than one 535 marking node, it is important to recognize the reference measurement 536 interval. In general the measurement interval for describing the 537 results is the interval of the marking node that is more aligned with 538 the start of the measurement, as reported in the following figure. 540 time -> start stop 541 T(R1) |-------------| 542 T(R2) |-------------| 543 T(R3) |------------| 545 Figure 4: Measurement Interval 547 T(R1) is the measurement interval and this is essential in order to 548 be compatible and make comparison with other active/passive/hybrid 549 Packet Loss metrics. 551 That is why, when we expand to multipoint-to-multipoint flows, we 552 have to consider that all source nodes mark the traffic. 554 Regarding the timing aspects of the methodology, RFC 8321 [RFC8321] 555 already describes two contributions that are taken into account: the 556 clock error between network devices and the network delay between 557 measurement points. 559 But we should now consider an additional contribution. Since all 560 source nodes mark the traffic, the source measurement intervals can 561 be of different lengths and with different offsets and this mismatch 562 m can be added to d, as shown in figure. 564 ...BBBBBBBBB | AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA | BBBBBBBBB... 565 |<======================================>| 566 | L | 567 ...=========>|<==================><==================>|<==========... 568 | L/2 L/2 | 569 |<=><===>| |<===><=>| 570 m d | | d m 571 |<====================>| 572 available counting interval 574 Figure 5: Timing Aspects for Multipoint paths 576 So the misalignment between the marking source routers gives an 577 additional constraint and the value of m is added to d (that already 578 includes clock error and network delay). 580 In the end, the condition that must be satisfied to enable the method 581 to function properly is that the available counting interval must be 582 > 0, and that means: L - 2m - 2d > 0 for each measurement point on 583 the multipoint path. Therefore, the mismatch between measurement 584 intervals must satisfy this condition. 586 8. Multipoint Delay and Delay Variation 588 The same line of reasoning can be applied to Delay and Delay 589 Variation. It is important to highlight that both delay and delay 590 variation measurements make sense in a multipoint path. The Delay 591 Variation is calculated by considering the same packets selected for 592 measuring the Delay. 594 In general, it is possible to perform delay and delay variation 595 measurements on multipoint paths basis or on single packets basis: 597 o Delay measurements on multipoint paths basis means that the delay 598 value is representative of an entire multipoint path (e.g. whole 599 multipoint network, a cluster or a combination of clusters). 601 o Delay measurements on single packets basis means that you can use 602 multipoint path just to easily couple packets between inputs and 603 output nodes of a multipoint path, as it is described in the 604 following sections. 606 8.1. Delay measurements on multipoint paths basis 608 8.1.1. Single Marking measurement 610 Mean delay and mean delay variation measurements can also be 611 generalized to the case of multipoint flows. It is possible to 612 compute the average one-way delay of packets, in one block, in a 613 cluster or in the entire monitored network. 615 The average latency can be measured as the difference between the 616 weighted averages of the mean timestamps of the sets of output and 617 input nodes. 619 8.2. Delay measurements on single packets basis 621 8.2.1. Single and Double Marking measurement 623 Delay and delay variation measurements relative to only one picked 624 packet per period (both single and double marked) can be performed in 625 the Multipoint scenario with some limitations: 627 Single marking based on the first/last packet of the interval 628 would not work, because it would not be possible to agree on the 629 first packet of the interval. 631 Double marking or multiplexed marking would work, but each 632 measurement would only give information about the delay of a 633 single path. However, by repeating the measurement multiple 634 times, it is possible to get information about all the paths in 635 the multipoint flow. This can be done in case of point-to- 636 multipoint path but it is more difficult to achieve in case of 637 multipoint-to-multipoint path because of the multiple source 638 routers. 640 if we would perform a delay measurement for more than one picked 641 packet in the same marking period and, especially, if we want to get 642 delay mesurements on multipoint-to-multipoint basis, both single and 643 double marking method are not useful in the Multipoint scenario, 644 since they would not be representative of the entire flow. The 645 packets can follow different paths with various delays and in general 646 it can be very difficult to recognize marked packets in a multipoint- 647 to-multipoint path especially in case they are more than one per 648 period. 650 A desirable option is to monitor simultaneously all the paths of a 651 multipoint path in the same marking period and, for this purpose, 652 hashing can be used as reported in the next Section. 654 8.2.2. Hashing selection method 656 RFC 5474 [RFC5474] and RFC 5475 [RFC5475] introduce sampling and 657 filtering techniques for IP Packet Selection. 659 The hash-based selection methodologies for delay measurement can work 660 in a multipoint-to-multipoint path and can be used both coupled to 661 mean delay or stand alone. 663 [I-D.mizrahi-ippm-compact-alternate-marking] introduces how to use 664 the Hash method combined with alternate marking method for point-to- 665 point flows. It is also called Mixed Hashed Marking: the coupling of 666 marking method and hashing technique is very useful because the 667 marking batches anchor the samples selected with hashing and this 668 simplifies the correlation of the hashing packets along the path. 670 It is possible to use a basic hash or a dynamic hash method. One of 671 the challenges of the basic approach is that the frequency of the 672 sampled packets may vary considerably. For this reason the dynamic 673 approach has been introduced for point-to-point flow in order to have 674 the desired and almost fixed number of samples for each measurement 675 period. In the hash-based sampling, alternate marking is used to 676 create periods, so that hash-based samples are divided into batches, 677 allowing to anchor the selected samples to their period. Moreover in 678 the dynamic hash-based sampling, by dynamically adapting the length 679 of the hash value, the number of samples is bounded in each marking 680 period. This can be realized by choosing the maximum number of 681 samples (NMAX) to be catched in a marking period. The algorithm 682 starts with only few hash bits, that permit to select a greater 683 percentage of packets (e.g. with 0 bit of hash all the packets are 684 sampled, with 1 bit of hash half of the packets are sampled, and so 685 on). When the number of selected packets reaches NMAX, a hashing bit 686 is added. As a consequence, the sampling proceeds at half of the 687 original rate and also the packets already selected that don't match 688 the new hash are discarded. This step can be repeated iteratively. 689 It is assumed that each sample includes the timestamp (used for delay 690 measurement) and the hash value, allowing the management system to 691 match the samples received from the two measurement points. The 692 dynamic process statistically converges at the end of a marking 693 period and the final number of selected samples is between NMAX/2 and 694 NMAX. Therefore, the dynamic approach paces the sampling rate, 695 allowing to bound the number of sampled packets per sampling period. 697 In a multipoint environment the behaviour is similar to point-to 698 point flow. In particular, in the context of multipoint-to- 699 multipoint flow, the dynamic hash could be the solution to perform 700 delay measurements on specific packets and to overcome the single and 701 double marking limitations. 703 The management system receives the samples including the timestamps 704 and the hash value from all the MPs, and this happens both for point- 705 to-point and for multipoint-to-multipoint flow. Then the longest 706 hash used by MPs is deduced and it is applied to couple timestamps of 707 same packets of 2 MPs of a point-to-point path or of input and output 708 MPs of a Cluster (or a Super Cluster or the entire network). But 709 some considerations are needed: if there isn't packet loss the set of 710 input samples is always equal to the set of output samples. In case 711 of packet loss the set of output samples can be a subset of input 712 samples but the method still works because, at the end, it is easy to 713 couple the input and output timestamps of each catched packet using 714 the hash (in particular the "unused part of the hash" that should be 715 different for each packet). 717 In summary, the basic hash is logically similar to the double marking 718 method, and in case of point-to-point path double marking and basic 719 hash selection are equivalent. The dynamic approach scales the 720 number of measurements per interval, and it would seem that double 721 marking would also work well if we reduced the interval length, but 722 this can be done only for point-to-point path and not for multipoint 723 path, where we cannot couple the picked packets in a multipoint 724 paths. So, in general, if we want to get delay mesurements on 725 multipoint-to-multipoint path basis and want to select more than one 726 packet per period, double marking cannot be used because we could not 727 be able to couple the picked packets between input and output nodes. 728 On the other hand we can do that by using hashing selection. 730 9. An SDN enabled Performance Management 732 The Multipoint Alternate Marking framework that is introduced in this 733 document adds flexibility to PM because it can reduce the order of 734 magnitude of the packet counters. This allows an SDN Orchestrator to 735 supervise, control and manage PM in large networks. 737 The monitoring network can be considered as a whole or can be split 738 in Clusters, that are the smallest subnetworks (group-to-group 739 segments), maintaining the packet loss property for each subnetwork. 740 They can also be combined in new connected subnetworks at different 741 levels depending on the detail we want to achieve. 743 An SDN Controller can calibrate Performance Measurements. It can 744 start without examining in depth. In case of necessity (packet loss 745 is measured or the delay is too high), the filtering criteria could 746 be immediately specified more in order to perform a partition of the 747 network by using Clusters and/or different combinations of Clusters. 748 In this way the problem can be localized in a specific Cluster or in 749 a single combination of Clusters and a more detailed analysis can be 750 performed step-by-step by successive approximation up to a point-to- 751 point flow detailed analysis. 753 In addition an SDN Controller could also collect the measurement 754 history. 756 10. Examples of application 758 There are three application fields where it may be useful to take 759 into consideration the Multipoint Alternate Marking: 761 o VPN: The IP traffic is selected on IP source basis in both 762 directions. At the end point WAN interface all the output traffic 763 is counted in a single flow. The input traffic is composed by all 764 the other flows aggregated for source address. So, by considering 765 n end-points, the monitored flows are n (each flow with 1 ingress 766 point and (n-1) egress points) instead of n*(n-1) flows (each 767 flow, with 1 ingress point and 1 egress point); 769 o Mobile Backhaul: LTE traffic is selected, in the Up direction, by 770 the EnodeB source address and, in Down direction, by the EnodeB 771 destination address because the packets are sent from the Mobile 772 Packet Core to the EnodeB. So the monitored flow is only one per 773 EnodeB in both directions; 775 o OTT(Over The Top) services: The traffic is selected, in the Down 776 direction by the source addresses of the packets sent by OTT 777 Servers. In the opposite direction (Up) by the destination IP 778 addresses of the same Servers. So the monitoring is based on a 779 single flow per OTT Servers in both directions. 781 11. Security Considerations 783 This document specifies a method to perform measurements that does 784 not directly affect Internet security nor applications that run on 785 the Internet. However, implementation of this method must be mindful 786 of security and privacy concerns, as explained in RFC 8321 [RFC8321]. 788 12. Acknowledgements 790 The authors would like to thank Al Morton, Tal Mizrahi, Rachel Huang 791 for the precious contribution. 793 13. IANA Considerations 795 tbc 797 14. References 799 14.1. Normative References 801 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 802 Requirement Levels", BCP 14, RFC 2119, 803 DOI 10.17487/RFC2119, March 1997, 804 . 806 [RFC5644] Stephan, E., Liang, L., and A. Morton, "IP Performance 807 Metrics (IPPM): Spatial and Multicast", RFC 5644, 808 DOI 10.17487/RFC5644, October 2009, 809 . 811 [RFC8321] Fioccola, G., Ed., Capello, A., Cociglio, M., Castaldelli, 812 L., Chen, M., Zheng, L., Mirsky, G., and T. Mizrahi, 813 "Alternate-Marking Method for Passive and Hybrid 814 Performance Monitoring", RFC 8321, DOI 10.17487/RFC8321, 815 January 2018, . 817 14.2. Informative References 819 [I-D.amf-ippm-route] 820 Alvarez-Hamelin, J., Morton, A., and J. Fabini, "Advanced 821 Unidirectional Route Assessment", draft-amf-ippm-route-01 822 (work in progress), October 2017. 824 [I-D.mizrahi-ippm-compact-alternate-marking] 825 Mizrahi, T., Arad, C., Fioccola, G., Cociglio, M., Chen, 826 M., Zheng, L., and G. Mirsky, "Compact Alternate Marking 827 Methods for Passive and Hybrid Performance Monitoring", 828 draft-mizrahi-ippm-compact-alternate-marking-01 (work in 829 progress), March 2018. 831 [RFC5474] Duffield, N., Ed., Chiou, D., Claise, B., Greenberg, A., 832 Grossglauser, M., and J. Rexford, "A Framework for Packet 833 Selection and Reporting", RFC 5474, DOI 10.17487/RFC5474, 834 March 2009, . 836 [RFC5475] Zseby, T., Molina, M., Duffield, N., Niccolini, S., and F. 837 Raspall, "Sampling and Filtering Techniques for IP Packet 838 Selection", RFC 5475, DOI 10.17487/RFC5475, March 2009, 839 . 841 [RFC7011] Claise, B., Ed., Trammell, B., Ed., and P. Aitken, 842 "Specification of the IP Flow Information Export (IPFIX) 843 Protocol for the Exchange of Flow Information", STD 77, 844 RFC 7011, DOI 10.17487/RFC7011, September 2013, 845 . 847 Authors' Addresses 849 Giuseppe Fioccola (editor) 850 Telecom Italia 851 Via Reiss Romoli, 274 852 Torino 10148 853 Italy 855 Email: giuseppe.fioccola@telecomitalia.it 857 Mauro Cociglio 858 Telecom Italia 859 Via Reiss Romoli, 274 860 Torino 10148 861 Italy 863 Email: mauro.cociglio@telecomitalia.it 864 Amedeo Sapio 865 Politecnico di Torino 866 Corso Duca degli Abruzzi, 24 867 Torino 10129 868 Italy 870 Email: amedeo.sapio@polito.it 872 Riccardo Sisto 873 Politecnico di Torino 874 Corso Duca degli Abruzzi, 24 875 Torino 10129 876 Italy 878 Email: riccardo.sisto@polito.it