idnits 2.17.1 draft-ietf-ippm-multipoint-alt-mark-03.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 (November 4, 2019) is 1635 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** Obsolete normative reference: RFC 8321 (Obsoleted by RFC 9341) == Outdated reference: A later version (-21) exists of draft-song-opsawg-ifit-framework-06 == Outdated reference: A later version (-14) exists of draft-zhou-ippm-enhanced-alternate-marking-04 Summary: 2 errors (**), 0 flaws (~~), 4 warnings (==), 1 comment (--).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 Huawei Technologies 4 Intended status: Experimental M. Cociglio 5 Expires: May 7, 2020 Telecom Italia 6 A. Sapio 7 R. Sisto 8 Politecnico di Torino 9 November 4, 2019 11 Multipoint Alternate Marking method for passive and hybrid performance 12 monitoring 13 draft-ietf-ippm-multipoint-alt-mark-03 15 Abstract 17 The Alternate Marking method, as presented in RFC 8321 [RFC8321], can 18 be applied only to point-to-point flows because it assumes that all 19 the packets of the flow measured on one node are measured again by a 20 single second node. This document aims to generalize and expand this 21 methodology to measure any kind of unicast flows, whose packets can 22 follow several different paths in the network, in wider terms a 23 multipoint-to-multipoint network. For this reason the technique here 24 described is called Multipoint Alternate Marking. Some definitions 25 here introduced extend the scope of RFC 5644 [RFC5644] in the context 26 of alternate marking schema. 28 Requirements Language 30 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 31 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 32 document are to be interpreted as described in RFC 2119 [RFC2119]. 34 Status of This Memo 36 This Internet-Draft is submitted in full conformance with the 37 provisions of BCP 78 and BCP 79. 39 Internet-Drafts are working documents of the Internet Engineering 40 Task Force (IETF). Note that other groups may also distribute 41 working documents as Internet-Drafts. The list of current Internet- 42 Drafts is at https://datatracker.ietf.org/drafts/current/. 44 Internet-Drafts are draft documents valid for a maximum of six months 45 and may be updated, replaced, or obsoleted by other documents at any 46 time. It is inappropriate to use Internet-Drafts as reference 47 material or to cite them other than as "work in progress." 48 This Internet-Draft will expire on May 7, 2020. 50 Copyright Notice 52 Copyright (c) 2019 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 . . . . . . . . . . . . . . . . . . . . . . . . 3 68 2. Correlation with RFC5644 . . . . . . . . . . . . . . . . . . 4 69 3. Flow classification . . . . . . . . . . . . . . . . . . . . . 5 70 4. Multipoint Performance Measurement . . . . . . . . . . . . . 7 71 4.1. Monitoring Network . . . . . . . . . . . . . . . . . . . 7 72 5. Multipoint Packet Loss . . . . . . . . . . . . . . . . . . . 9 73 6. Network Clustering . . . . . . . . . . . . . . . . . . . . . 9 74 6.1. Algorithm for Cluster partition . . . . . . . . . . . . . 10 75 7. Timing Aspects . . . . . . . . . . . . . . . . . . . . . . . 13 76 8. Multipoint Delay and Delay Variation . . . . . . . . . . . . 14 77 8.1. Delay measurements on multipoint paths basis . . . . . . 15 78 8.1.1. Single Marking measurement . . . . . . . . . . . . . 15 79 8.2. Delay measurements on single packets basis . . . . . . . 15 80 8.2.1. Single and Double Marking measurement . . . . . . . . 15 81 8.2.2. Hashing selection method . . . . . . . . . . . . . . 16 82 9. An Intelligent Performance Management approach . . . . . . . 17 83 10. Examples of application . . . . . . . . . . . . . . . . . . . 18 84 11. Security Considerations . . . . . . . . . . . . . . . . . . . 19 85 12. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 19 86 13. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 19 87 14. References . . . . . . . . . . . . . . . . . . . . . . . . . 20 88 14.1. Normative References . . . . . . . . . . . . . . . . . . 20 89 14.2. Informative References . . . . . . . . . . . . . . . . . 20 90 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 21 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. Note that additional details about the 106 Alternate Marking methodology are described in the paper 107 [IEEE-Network-PNPM] 109 There are some applications of the alternate marking method where 110 there are a lot of monitored flows and nodes. Multipoint Alternate 111 Marking aims to reduce these values and makes the performance 112 monitoring more flexible in case a detailed analysis is not needed. 113 For instance, by considering n measurement points and m monitored 114 flows,the order of magnitude of the packet counters for each time 115 interval is n*m*2 (1 per color). If both n and m are high values the 116 packet counters increase a lot and Multipoint Alternate Marking 117 offers a tool to control these parameters. 119 The approach presented in this document is applied only to unicast 120 flows and not to multicast. BUM (Broadcast Unknown Unicast 121 Multicast) traffic is not considered here, because traffic 122 replication is not covered by the Multipoint Alternate Marking 123 method. Furthermore it can be applicable to anycast flows and ECMP 124 (Equal-Cost Multi-Path) paths can also be easily monitored with this 125 technique. 127 In short, RFC 8321 [RFC8321] applies to point-to-point unicast flows 128 and BUM traffic and the Multipoint alternate marking and its 129 Clustering approach is valid for multipoint-to-multipoint unicast 130 flows, anycast and ECMP flows. 132 The Alternate Marking method can therefore be extended to any kind of 133 multipoint to multipoint paths, and the network clustering approach 134 presented in this document is the formalization of how to implement 135 this property and allow a flexible and optimized performance 136 measurement support for network management in every situation. 138 Without network clustering, it is possible to apply alternate marking 139 only for all the network or per single flow. Instead, with network 140 clustering, it is possible to use the network clusters partition at 141 different levels to perform the needed degree of detail. In some 142 circumstances it is possible to monitor a Multipoint Network by 143 analysing the Network Clustering, without examining in depth. In 144 case of problems (packet loss is measured or the delay is too high) 145 the filtering criteria could be specified more in order to perform a 146 detailed analysis by using a different combination of clusters up to 147 a per-flow measurement as described in RFC 8321 [RFC8321]. 149 This approach fits very well with the Intelligent Network and 150 Software Defined Network (SDN) paradigm where the SDN Orchestrator 151 and the SDN Controllers are the brains of the network and can manage 152 flow control to the switches and routers and, in the same way, can 153 calibrate the performance measurements depending on the necessity. 154 An SDN Controller Application can orchestrate how deep the network 155 performance monitoring is setup by applying the Multipoint Alternate 156 Marking as described in this document. 158 It is important to underline that, as extension of RFC 8321 159 [RFC8321], this is a methodology draft, so the mechanism that can be 160 used to transmit the counters and the timestamps is out of scope here 161 and the implementation is open. Several options are possible, e.g. 162 [I-D.zhou-ippm-enhanced-alternate-marking]. 164 2. Correlation with RFC5644 166 RFC 5644 [RFC5644] is limited to active measurements using a single 167 source packet or stream, and observations of corresponding packets 168 along the path (spatial), at one or more destinations (one-to-group), 169 or both. 171 Instead, the scope of this memo is to define multiparty metrics for 172 passive and hybrid measurements in a group-to-group topology with 173 multiple sources and destinations. 175 RFC 5644 [RFC5644] introduces metric names that can be reused also 176 here but have to be extended and rephrased to be applied to the 177 alternate marking schema: 179 a. the multiparty metrics are not only one-to-group metrics but can 180 be also group-to-group metrics; 182 b. the spatial metrics, used for measuring the performance of 183 segments of a source to destination path, are applied here to 184 group-to-group segments (called Clusters). 186 3. Flow classification 188 An unicast flow is identified by all the packets having a set of 189 common characteristics. This definition is inspired by RFC 7011 190 [RFC7011]. 192 As an example, by considering a flow as all the packets sharing the 193 same source IP address or the same destination IP address, it is easy 194 to understand that the resulting pattern will not be a point-to-point 195 connection, but a point-to-multipoint or multipoint-to-point 196 connection. 198 In general a flow can be defined by a set of selection rules used to 199 match a subset of the packets processed by the network device. These 200 rules specify a set of headers fields (Identification Fields) and the 201 relative values that must be found in matching packets. 203 The choice of the identification fields directly affects the type of 204 paths that the flow would follow in the network. In fact, it is 205 possible to relate a set of identification fields with the pattern of 206 the resulting graphs, as listed in Figure 1. 208 A TCP 5-tuple usually identifies flows following either a single path 209 or a point-to-point multipath (in case of load balancing). On the 210 contrary, a single source address selects flows following a point-to- 211 multipoint, while a multipoint-to-point can be the result of a 212 matching on a single destination address. In case a selection rule 213 and its reverse are used for bidirectional measurements, they can 214 correspond to a point-to-multipoint in one direction and a 215 multipoint-to-point in the opposite direction. 217 In this way the flows to be monitored are selected into the 218 monitoring points using packet selection rules, that can also change 219 the pattern of the monitored network. 221 The alternate marking method is applicable only to a single path (and 222 partially to a one-to-one multipath), so the extension proposed in 223 this document is suitable also for the most general case of 224 multipoint-to-multipoint, which embraces all the other patterns of 225 Figure 1. 227 point-to-point single path 228 +------+ +------+ +------+ 229 ---<> R1 <>----<> R2 <>----<> R3 <>--- 230 +------+ +------+ +------+ 232 point-to-point multipath 233 +------+ 234 <> R2 <> 235 / +------+ \ 236 / \ 237 +------+ / \ +------+ 238 ---<> R1 <> <> R4 <>--- 239 +------+ \ / +------+ 240 \ / 241 \ +------+ / 242 <> R3 <> 243 +------+ 245 point-to-multipoint 246 +------+ 247 <> R4 <>--- 248 / +------+ 249 +------+ / 250 <> R2 <> 251 / +------+ \ 252 +------+ / \ +------+ 253 ---<> R1 <> <> R5 <>--- 254 +------+ \ +------+ 255 \ +------+ 256 <> R3 <> 257 +------+ \ 258 \ +------+ 259 <> R6 <>--- 260 +------+ 262 multipoint-to-point 263 +------+ 264 ---<> R1 <> 265 +------+ \ 266 \ +------+ 267 <> R4 <> 268 / +------+ \ 269 +------+ / \ +------+ 270 ---<> R2 <> <> R4 <>--- 271 +------+ / +------+ 272 +------+ / 273 <> R5 <> 274 / +------+ 275 +------+ / 276 ---<> R3 <> 277 +------+ 279 multipoint-to-multipoint 280 +------+ +------+ 281 ---<> R1 <> <> R6 <>--- 282 +------+ \ / +------+ 283 \ +------+ / 284 <> R4 <> 285 +------+ \ 286 +------+ \ +------+ 287 ---<> R2 <> <> R7 <>--- 288 +------+ \ / +------+ 289 \ +------+ / 290 <> R5 <> 291 / +------+ \ 292 +------+ / \ +------+ 293 ---<> R3 <> <> R8 <>--- 294 +------+ +------+ 296 Figure 1: Flow classification 298 The case of unicast flow is considered in the previous figure. 299 Anyway the anycast flow is also in scope because there is no 300 replication and only a single node from the anycast group receives 301 the traffic, so it can be viewed as a special case of unicast flow. 302 While ECMP flow is in scope by definition, since it is a point-to- 303 multipoint unicast flow. 305 4. Multipoint Performance Measurement 307 By Using the "traditional" alternate marking method only point-to- 308 point paths can be monitored. To have an IP (TCP/UDP) flow that 309 follows a point-to-point path we have to define, with a specific 310 value, 5 identification fields (IP Source, IP Destination, Transport 311 Protocol, Source Port, Destination Port). 313 Multipoint Alternate Marking enables the performance measurement for 314 multipoint flows selected by identification fields without any 315 constraints (even the entire network production traffic). It is also 316 possible to use multiple marking points for the same monitored flow. 318 4.1. Monitoring Network 320 The Monitoring Network is deduced from the Production Network, by 321 identifying the nodes of the graph that are the measurement points, 322 and the links that are the connections between measurement points. 324 There are some techniques that can help with the building of the 325 monitoring network (as an example it is possible to mention 327 [I-D.amf-ippm-route]). In general there are different options: the 328 monitoring network can be obtained by considering all the possible 329 paths for the traffic or also by checking the traffic sometimes and 330 update the graph consequently. 332 So a graph model of the monitoring network can be built according to 333 the alternate marking method: the monitored interfaces and links are 334 identified. Only the measurement points and links where the traffic 335 has flowed have to be represented in the graph. 337 The following figure shows a simple example of a Monitoring Network 338 graph: 340 +------+ 341 <> R6 <>--- 342 / +------+ 343 +------+ +------+ / 344 <> R2 <>---<> R4 <> 345 / +------+ \ +------+ \ 346 / \ \ +------+ 347 +------+ / +------+ \ +------+ <> R7 <>--- 348 ---<> R1 <>---<> R3 <>---<> R5 <> +------+ 349 +------+ \ +------+ \ +------+ \ 350 \ \ \ +------+ 351 \ \ <> R8 <>--- 352 \ \ +------+ 353 \ \ 354 \ \ +------+ 355 \ <> R9 <>--- 356 \ +------+ 357 \ 358 \ +------+ 359 <> R10 <>--- 360 +------+ 362 Figure 2: Monitoring Network Graph 364 Each monitoring point is characterized by the packet counter that 365 refers only to a marking period of the monitored flow. 367 The same is applicable also for the delay but it will be described in 368 the following sections. 370 5. Multipoint Packet Loss 372 Since all the packets of the considered flow leaving the network have 373 previously entered the network, the number of packets counted by all 374 the input nodes is always greater or equal than the number of packets 375 counted by all the output nodes. 377 And in case of no packet loss occurring in the marking period, if all 378 the input and output points of the network domain to be monitored are 379 measurement points, the sum of the number of packets on all the 380 ingress interfaces and on all the egress interfaces is the same. In 381 this circumstance, if no packet loss occurs, the intermediate 382 measurement points have only the task to split the measurement. 384 It is possible to define the Network Packet Loss (for 1 flow, for 1 385 period): <>. This is true for every packet 388 flow in each marking period. 390 The Monitored Network Packet Loss with n input nodes and m output 391 nodes is given by: 393 PL = (PI1 + PI2 +...+ PIn) - (PO1 + PO2 +...+ POm) 395 where: 397 PL is the Network Packet Loss (number of lost packets) 399 PIi is the Number of packets flowed through the i-th Input node in 400 this period 402 POj is the Number of packets flowed through the j-th Output node in 403 this period 405 The equation is applied on a per-time-interval basis. 407 6. Network Clustering 409 The previous Equation can determine the number of packets lost 410 globally in the monitored network, exploiting only the data provided 411 by the counters in the input and output nodes. 413 In addition it is also possible to leverage the data provided by the 414 other counters in the network to converge on the smallest 415 identifiable subnetworks where the losses occur. These subnetworks 416 are named Clusters. 418 A Cluster graph is a subnetwork of the entire Monitoring Network 419 graph that still satisfies the packet loss equation where PL in this 420 case is the number of packets lost in the Cluster. 422 For this reason a Cluster should contain all the arcs emanating from 423 its input nodes and all the arcs terminating at its output nodes. 424 This ensures that we can count all the packets (and only those) 425 exiting an input node again at the output node, whatever path they 426 follow. 428 In a completely monitored network (a network where every network 429 interface is monitored), each network device corresponds to a Cluster 430 and each physical link corresponds to two Clusters (one for each 431 direction). 433 Clusters can have different sizes depending on flow filtering 434 criteria adopted. 436 Moreover, sometimes Clusters can be optionally simplified. For 437 example when two monitored interfaces are divided by a single router 438 (one is the input interface and the other is the output interface and 439 the router has only these two interfaces), instead of counting 440 exactly twice, upon entering and leaving, it is possible to consider 441 a single measurement point (in this case we do not care of the 442 internal packet loss of the router). 444 6.1. Algorithm for Cluster partition 446 A simple algorithm can be applied in order to split our monitoring 447 network into Clusters. It is a two-step algorithm: 449 o Group the links where there is the same starting node; 451 o Join the grouped links with at least one ending node in common. 453 In our monitoring network graph example it is possible to identify 454 the Clusters partition by applying this two-step algorithm. 456 The first step identifies the following groups: 458 1. Group 1: (R1-R2), (R1-R3), (R1-R10) 460 2. Group 2: (R2-R4), (R2-R5) 462 3. Group 3: (R3-R5), (R3-R9) 464 4. Group 4: (R4-R6), (R4-R7) 465 5. Group 5: (R5-R8) 467 And then, the second step builds the Clusters partition (in 468 particular we can underline that Group 2 and Group 3 connect 469 together, since R5 is in common): 471 1. Cluster 1: (R1-R2), (R1-R3), (R1-R10) 473 2. Cluster 2: (R2-R4), (R2-R5), (R3-R5), (R3-R9) 475 3. Cluster 3: (R4-R6), (R4-R7) 477 4. Cluster 4: (R5-R8) 479 In the end the following 4 Clusters are obtained: 481 Cluster 1 482 +------+ 483 <> R2 <>--- 484 / +------+ 485 / 486 +------+ / +------+ 487 ---<> R1 <>---<> R3 <>--- 488 +------+ \ +------+ 489 \ 490 \ 491 \ 492 \ 493 \ 494 \ 495 \ 496 \ 497 \ +------+ 498 <> R10 <>--- 499 +------+ 501 Cluster 2 502 +------+ +------+ 503 ---<> R2 <>---<> R4 <>--- 504 +------+ \ +------+ 505 \ 506 +------+ \ +------+ 507 ---<> R3 <>---<> R5 <>--- 508 +------+ \ +------+ 509 \ 510 \ 511 \ 512 \ 513 \ +------+ 514 <> R9 <>--- 515 +------+ 517 Cluster 3 518 +------+ 519 <> R6 <>--- 520 / +------+ 521 +------+ / 522 ---<> R4 <> 523 +------+ \ 524 \ +------+ 525 <> R7 <>--- 526 +------+ 528 Cluster 4 529 +------+ 530 ---<> R5 <> 531 +------+ \ 532 \ +------+ 533 <> R8 <>--- 534 +------+ 536 Figure 3: Clusters example 538 There are Clusters with more than 2 nodes and two-nodes Clusters. In 539 the two-nodes Clusters the loss is on the link (Cluster 4). In more- 540 than-2-nodes Clusters the loss is on the Cluster but we cannot know 541 in which link (Cluster 1, 2, 3). 543 In this way the calculation of packet loss can be made on Cluster 544 basis. Note that CIR(Committed Information Rate) and EIR(Excess 545 Information Rate) can also be deduced on Cluster basis. 547 Obviously, by combining some Clusters in a new connected subnetwork 548 (called Super Cluster) the Packet Loss Rule is still true. 550 In this way in a very large network there is no need to configure 551 detailed filter criteria to inspect the traffic. You can check 552 multipoint network and only in case of problems you can go deep with 553 a step-by-step cluster analysis, but only for the cluster or 554 combination of clusters where the problem happens. 556 The complete and mathematical analysis of the possible Algorithms for 557 Cluster partition, including the considerations in terms of 558 efficiency and a comparison between the different methods, is in the 559 paper [IEEE-ACM-ToN-MPNPM]. 561 7. Timing Aspects 563 It is important to consider the timing aspects, since out of order 564 packets happen and have to be handled as well as described in RFC 565 8321 [RFC8321]. But, in a multi-source situation an additional issue 566 has to be considered. 568 So, if we analyse a multipoint-to-multipoint path with more than one 569 marking node, it is important to recognize the reference measurement 570 interval. In general the measurement interval for describing the 571 results is the interval of the marking node that is more aligned with 572 the start of the measurement, as reported in the following figure. 574 Note that the mark switching approach based on a fixed timer is 575 considered in this document. 577 time -> start stop 578 T(R1) |-------------| 579 T(R2) |-------------| 580 T(R3) |------------| 582 Figure 4: Measurement Interval 584 T(R1) is the measurement interval and this is essential in order to 585 be compatible and make comparison with other active/passive/hybrid 586 Packet Loss metrics. 588 That is why, when we expand to multipoint-to-multipoint flows, we 589 have to consider that all source nodes mark the traffic. 591 Regarding the timing aspects of the methodology, RFC 8321 [RFC8321] 592 already describes two contributions that are taken into account: the 593 clock error between network devices and the network delay between 594 measurement points. 596 But we should now consider an additional contribution. Since all 597 source nodes mark the traffic, the source measurement intervals can 598 be of different lengths and with different offsets and this mismatch 599 m can be added to d, as shown in figure. 601 ...BBBBBBBBB | AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA | BBBBBBBBB... 602 |<======================================>| 603 | L | 604 ...=========>|<==================><==================>|<==========... 605 | L/2 L/2 | 606 |<=><===>| |<===><=>| 607 m d | | d m 608 |<====================>| 609 available counting interval 611 Figure 5: Timing Aspects for Multipoint paths 613 So the misalignment between the marking source routers gives an 614 additional constraint and the value of m is added to d (that already 615 includes clock error and network delay). 617 Therefore, three different possible constraints are considered: clock 618 error between network devices, network delay between measurement 619 points and the misalignment between the marking source routers. 621 In the end, the condition that must be satisfied to enable the method 622 to function properly is that the available counting interval must be 623 > 0, and that means: L - 2m - 2d > 0 for each measurement point on 624 the multipoint path. Therefore, the mismatch between measurement 625 intervals must satisfy this condition. 627 The timing considerations are valid for both packet loss and delay 628 measurements. 630 8. Multipoint Delay and Delay Variation 632 The same line of reasoning can be applied to Delay and Delay 633 Variation. Similarly to the delay measurements defined in RFC 8321 634 [RFC8321], the marking batches anchor the samples to a particular 635 period and this is the time reference that can be used. It is 636 important to highlight that both delay and delay variation 637 measurements make sense in a multipoint path. The Delay Variation is 638 calculated by considering the same packets selected for measuring the 639 Delay. 641 In general, it is possible to perform delay and delay variation 642 measurements on multipoint paths basis or on single packets basis: 644 o Delay measurements on multipoint paths basis means that the delay 645 value is representative of an entire multipoint path (e.g. whole 646 multipoint network, a cluster or a combination of clusters). 648 o Delay measurements on single packets basis means that you can use 649 multipoint path just to easily couple packets between inputs and 650 output nodes of a multipoint path, as it is described in the 651 following sections. 653 8.1. Delay measurements on multipoint paths basis 655 8.1.1. Single Marking measurement 657 Mean delay and mean delay variation measurements can also be 658 generalized to the case of multipoint flows. It is possible to 659 compute the average one-way delay of packets, in one block, in a 660 cluster or in the entire monitored network. 662 The average latency can be measured as the difference between the 663 weighted averages of the mean timestamps of the sets of output and 664 input nodes. 666 8.2. Delay measurements on single packets basis 668 8.2.1. Single and Double Marking measurement 670 Delay and delay variation measurements relative to only one picked 671 packet per period (both single and double marked) can be performed in 672 the Multipoint scenario with some limitations: 674 Single marking based on the first/last packet of the interval 675 would not work, because it would not be possible to agree on the 676 first packet of the interval. 678 Double marking or multiplexed marking would work, but each 679 measurement would only give information about the delay of a 680 single path. However, by repeating the measurement multiple 681 times, it is possible to get information about all the paths in 682 the multipoint flow. This can be done in case of point-to- 683 multipoint path but it is more difficult to achieve in case of 684 multipoint-to-multipoint path because of the multiple source 685 routers. 687 if we would perform a delay measurement for more than one picked 688 packet in the same marking period and, especially, if we want to get 689 delay measurements on multipoint-to-multipoint basis, both single and 690 double marking method are not useful in the Multipoint scenario, 691 since they would not be representative of the entire flow. The 692 packets can follow different paths with various delays and in general 693 it can be very difficult to recognize marked packets in a multipoint- 694 to-multipoint path especially in case they are more than one per 695 period. 697 A desirable option is to monitor simultaneously all the paths of a 698 multipoint path in the same marking period and, for this purpose, 699 hashing can be used as reported in the next Section. 701 8.2.2. Hashing selection method 703 RFC 5474 [RFC5474] and RFC 5475 [RFC5475] introduce sampling and 704 filtering techniques for IP Packet Selection. 706 The hash-based selection methodologies for delay measurement can work 707 in a multipoint-to-multipoint path and can be used both coupled to 708 mean delay or stand alone. 710 [I-D.mizrahi-ippm-compact-alternate-marking] introduces how to use 711 the Hash method combined with alternate marking method for point-to- 712 point flows. It is also called Mixed Hashed Marking: the coupling of 713 marking method and hashing technique is very useful because the 714 marking batches anchor the samples selected with hashing and this 715 simplifies the correlation of the hashing packets along the path. 717 It is possible to use a basic hash or a dynamic hash method. One of 718 the challenges of the basic approach is that the frequency of the 719 sampled packets may vary considerably. For this reason the dynamic 720 approach has been introduced for point-to-point flow in order to have 721 the desired and almost fixed number of samples for each measurement 722 period. In the hash-based sampling, alternate marking is used to 723 create periods, so that hash-based samples are divided into batches, 724 allowing to anchor the selected samples to their period. Moreover in 725 the dynamic hash-based sampling, by dynamically adapting the length 726 of the hash value, the number of samples is bounded in each marking 727 period. This can be realized by choosing the maximum number of 728 samples (NMAX) to be caught in a marking period. The algorithm 729 starts with only few hash bits, that permit to select a greater 730 percentage of packets (e.g. with 0 bit of hash all the packets are 731 sampled, with 1 bit of hash half of the packets are sampled, and so 732 on). When the number of selected packets reaches NMAX, a hashing bit 733 is added. As a consequence, the sampling proceeds at half of the 734 original rate and also the packets already selected that don't match 735 the new hash are discarded. This step can be repeated iteratively. 736 It is assumed that each sample includes the timestamp (used for delay 737 measurement) and the hash value, allowing the management system to 738 match the samples received from the two measurement points. The 739 dynamic process statistically converges at the end of a marking 740 period and the final number of selected samples is between NMAX/2 and 741 NMAX. Therefore, the dynamic approach paces the sampling rate, 742 allowing to bound the number of sampled packets per sampling period. 744 In a multipoint environment the behaviour is similar to point-to 745 point flow. In particular, in the context of multipoint-to- 746 multipoint flow, the dynamic hash could be the solution to perform 747 delay measurements on specific packets and to overcome the single and 748 double marking limitations. 750 The management system receives the samples including the timestamps 751 and the hash value from all the MPs, and this happens both for point- 752 to-point and for multipoint-to-multipoint flow. Then the longest 753 hash used by MPs is deduced and it is applied to couple timestamps of 754 same packets of 2 MPs of a point-to-point path or of input and output 755 MPs of a Cluster (or a Super Cluster or the entire network). But 756 some considerations are needed: if there isn't packet loss the set of 757 input samples is always equal to the set of output samples. In case 758 of packet loss the set of output samples can be a subset of input 759 samples but the method still works because, at the end, it is easy to 760 couple the input and output timestamps of each caught packet using 761 the hash (in particular the "unused part of the hash" that should be 762 different for each packet). 764 In summary, the basic hash is logically similar to the double marking 765 method, and in case of point-to-point path double marking and basic 766 hash selection are equivalent. The dynamic approach scales the 767 number of measurements per interval, and it would seem that double 768 marking would also work well if we reduced the interval length, but 769 this can be done only for point-to-point path and not for multipoint 770 path, where we cannot couple the picked packets in a multipoint 771 paths. So, in general, if we want to get delay measurements on 772 multipoint-to-multipoint path basis and want to select more than one 773 packet per period, double marking cannot be used because we could not 774 be able to couple the picked packets between input and output nodes. 775 On the other hand we can do that by using hashing selection. 777 9. An Intelligent Performance Management approach 779 The Multipoint Alternate Marking framework that is introduced in this 780 document adds flexibility to PM because it can reduce the order of 781 magnitude of the packet counters. This allows an SDN Orchestrator to 782 supervise, control and manage PM in large networks. 784 The monitoring network can be considered as a whole or can be split 785 in Clusters, that are the smallest subnetworks (group-to-group 786 segments), maintaining the packet loss property for each subnetwork. 787 They can also be combined in new connected subnetworks at different 788 levels depending on the detail we want to achieve. 790 An SDN Controller can calibrate Performance Measurements since it is 791 aware of the network topology. It can start without examining in 792 depth. In case of necessity (packet loss is measured or the delay is 793 too high), the filtering criteria could be immediately specified more 794 in order to perform a partition of the network by using Clusters and/ 795 or different combinations of Clusters. In this way the problem can 796 be localized in a specific Cluster or in a single combination of 797 Clusters and a more detailed analysis can be performed step-by-step 798 by successive approximation up to a point-to-point flow detailed 799 analysis. 801 This approach can be called Network Zooming and can be performed in 802 two different ways: 804 1) change the traffic filter and select more detailed flows; 806 2) activate new measurement points by defining more specified 807 clusters. 809 The Network Zooming approach implies that the some filters or rules 810 are changed and there is a transient time to wait once the new 811 network configuration takes effect and it can be determined by the 812 Network Orchestrator/Controller, based on the network conditions. 814 [I-D.song-opsawg-ifit-framework] defines an architecture where the 815 centralized Data Collector and Network Management can apply the 816 intelligent and flexible Alternate Marking algorithm as previously 817 described. 819 As for RFC 8321 [RFC8321], it is possible to classify the traffic and 820 mark a portion of the total traffic. For each period the packet rate 821 and bandwidth are calculated from the number of packets. In this way 822 the Network Orchestrator becomes aware if the traffic rate overcomes 823 limits. In addition more precision can be obtained by reducing the 824 marking period, indeed some implementations use a marking period of 1 825 sec and less. 827 In addition an SDN Controller could also collect the measurement 828 history. 830 It is important to mention that the Multipoint Alternate Marking 831 framework also helps Traffic Visualization. Indeed this methodology 832 is very useful to identify which path or which cluster is crossed by 833 the flow. 835 10. Examples of application 837 There are application fields where it may be useful to take into 838 consideration the Multipoint Alternate Marking: 840 o VPN: The IP traffic is selected on IP source basis in both 841 directions. At the end point WAN interface all the output traffic 842 is counted in a single flow. The input traffic is composed by all 843 the other flows aggregated for source address. So, by considering 844 n end-points, the monitored flows are n (each flow with 1 ingress 845 point and (n-1) egress points) instead of n*(n-1) flows (each 846 flow, with 1 ingress point and 1 egress point); 848 o Mobile Backhaul: LTE traffic is selected, in the Up direction, by 849 the EnodeB source address and, in Down direction, by the EnodeB 850 destination address because the packets are sent from the Mobile 851 Packet Core to the EnodeB. So the monitored flow is only one per 852 EnodeB in both directions; 854 o OTT(Over The Top) services: The traffic is selected, in the Down 855 direction by the source addresses of the packets sent by OTT 856 Servers. In the opposite direction (Up) by the destination IP 857 addresses of the same Servers. So the monitoring is based on a 858 single flow per OTT Servers in both directions. 860 o Enterprise SD-WAN: SD-WAN allows to connect remote branch offices 861 to Data Centers and build higher-performance WANs. A centralized 862 controller is used to set policies and prioritize traffic. The 863 SD-WAN takes into account these policies and the availability of 864 network bandwidth to route traffic. This helps ensure that 865 application performance meets service level agreements (SLAs). 866 This methodology can also help the path selection for the WAN 867 connection based on per Cluster and per flow performance. 869 11. Security Considerations 871 This document specifies a method to perform measurements that does 872 not directly affect Internet security nor applications that run on 873 the Internet. However, implementation of this method must be mindful 874 of security and privacy concerns, as explained in RFC 8321 [RFC8321]. 876 12. Acknowledgements 878 The authors would like to thank Al Morton, Tal Mizrahi, Rachel Huang 879 for the precious contribution. 881 13. IANA Considerations 883 tbc 885 14. References 887 14.1. Normative References 889 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 890 Requirement Levels", BCP 14, RFC 2119, 891 DOI 10.17487/RFC2119, March 1997, 892 . 894 [RFC5644] Stephan, E., Liang, L., and A. Morton, "IP Performance 895 Metrics (IPPM): Spatial and Multicast", RFC 5644, 896 DOI 10.17487/RFC5644, October 2009, 897 . 899 [RFC8321] Fioccola, G., Ed., Capello, A., Cociglio, M., Castaldelli, 900 L., Chen, M., Zheng, L., Mirsky, G., and T. Mizrahi, 901 "Alternate-Marking Method for Passive and Hybrid 902 Performance Monitoring", RFC 8321, DOI 10.17487/RFC8321, 903 January 2018, . 905 14.2. Informative References 907 [I-D.amf-ippm-route] 908 Alvarez-Hamelin, J., Morton, A., and J. Fabini, "Advanced 909 Unidirectional Route Assessment", draft-amf-ippm-route-01 910 (work in progress), October 2017. 912 [I-D.mizrahi-ippm-compact-alternate-marking] 913 Mizrahi, T., Arad, C., Fioccola, G., Cociglio, M., Chen, 914 M., Zheng, L., and G. Mirsky, "Compact Alternate Marking 915 Methods for Passive and Hybrid Performance Monitoring", 916 draft-mizrahi-ippm-compact-alternate-marking-05 (work in 917 progress), July 2019. 919 [I-D.song-opsawg-ifit-framework] 920 Song, H., Li, Z., Zhou, T., Qin, F., Chen, H., Jin, J., 921 and J. Shin, "In-situ Flow Information Telemetry", draft- 922 song-opsawg-ifit-framework-06 (work in progress), October 923 2019. 925 [I-D.zhou-ippm-enhanced-alternate-marking] 926 Zhou, T., Fioccola, G., Li, Z., Lee, S., and M. Cociglio, 927 "Enhanced Alternate Marking Method", draft-zhou-ippm- 928 enhanced-alternate-marking-04 (work in progress), October 929 2019. 931 [IEEE-ACM-ToN-MPNPM] 932 IEEE/ACM TRANSACTION ON NETWORKING, "Multipoint Passive 933 Monitoring in Packet Networks", DOI to appear, 2019. 935 [IEEE-Network-PNPM] 936 IEEE Network, "AM-PM: Efficient Network Telemetry using 937 Alternate Marking", DOI 10.1109/MNET.2019.1800152, 2019. 939 [RFC5474] Duffield, N., Ed., Chiou, D., Claise, B., Greenberg, A., 940 Grossglauser, M., and J. Rexford, "A Framework for Packet 941 Selection and Reporting", RFC 5474, DOI 10.17487/RFC5474, 942 March 2009, . 944 [RFC5475] Zseby, T., Molina, M., Duffield, N., Niccolini, S., and F. 945 Raspall, "Sampling and Filtering Techniques for IP Packet 946 Selection", RFC 5475, DOI 10.17487/RFC5475, March 2009, 947 . 949 [RFC7011] Claise, B., Ed., Trammell, B., Ed., and P. Aitken, 950 "Specification of the IP Flow Information Export (IPFIX) 951 Protocol for the Exchange of Flow Information", STD 77, 952 RFC 7011, DOI 10.17487/RFC7011, September 2013, 953 . 955 Authors' Addresses 957 Giuseppe Fioccola (editor) 958 Huawei Technologies 959 Riesstrasse, 25 960 Munich 80992 961 Germany 963 Email: giuseppe.fioccola@huawei.com 965 Mauro Cociglio 966 Telecom Italia 967 Via Reiss Romoli, 274 968 Torino 10148 969 Italy 971 Email: mauro.cociglio@telecomitalia.it 972 Amedeo Sapio 973 Politecnico di Torino 974 Corso Duca degli Abruzzi, 24 975 Torino 10129 976 Italy 978 Email: amedeo.sapio@polito.it 980 Riccardo Sisto 981 Politecnico di Torino 982 Corso Duca degli Abruzzi, 24 983 Torino 10129 984 Italy 986 Email: riccardo.sisto@polito.it