idnits 2.17.1 draft-ietf-alto-performance-metrics-08.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 ([RFC2119], [RFC7285]), which it shouldn't. Please replace those with straight textual mentions of the documents in question. -- The document has examples using IPv4 documentation addresses according to RFC6890, but does not use any IPv6 documentation addresses. Maybe there should be IPv6 examples, too? Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (November 04, 2019) is 1634 days in the past. Is this intentional? Checking references for intended status: Proposed Standard ---------------------------------------------------------------------------- (See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) == Missing Reference: 'RFC8259' is mentioned on line 709, but not defined == Missing Reference: 'RFC3630' is mentioned on line 988, but not defined == Missing Reference: 'RFC3784' is mentioned on line 988, but not defined ** Obsolete undefined reference: RFC 3784 (Obsoleted by RFC 5305) == Unused Reference: 'RFC5234' is defined on line 1135, but no explicit reference was found in the text == Unused Reference: 'RFC6390' is defined on line 1183, but no explicit reference was found in the text == Outdated reference: A later version (-16) exists of draft-ietf-ippm-initial-registry-12 ** Obsolete normative reference: RFC 2679 (Obsoleted by RFC 7679) ** Obsolete normative reference: RFC 4627 (Obsoleted by RFC 7158, RFC 7159) ** Downref: Normative reference to an Informational RFC: RFC 6349 ** Obsolete normative reference: RFC 7752 (Obsoleted by RFC 9552) ** Obsolete normative reference: RFC 7810 (Obsoleted by RFC 8570) Summary: 7 errors (**), 0 flaws (~~), 7 warnings (==), 2 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 ALTO Working Group Q. Wu 3 Internet-Draft Huawei 4 Intended status: Standards Track Y. Yang 5 Expires: May 7, 2020 Yale University 6 Y. Lee 7 D. Dhody 8 Huawei 9 S. Randriamasy 10 Nokia Bell Labs 11 November 04, 2019 13 ALTO Performance Cost Metrics 14 draft-ietf-alto-performance-metrics-08 16 Abstract 18 Cost metric is a basic concept in Application-Layer Traffic 19 Optimization (ALTO), and is used in basic services including both the 20 cost map service and the endpoint cost service. 22 Different applications may use different cost metrics, but the ALTO 23 base protocol documents only one single cost metric, i.e., the 24 generic "routingcost" metric; see Sec. 14.2 of ALTO base 25 specification [RFC7285]. Hence, if the resource consumer of an 26 application prefers a resource provider that offers low-delay 27 delivery to the resource consumer, the base protocol does not define 28 the cost metric to be used. 30 ALTO cost metrics can be generic metrics and this document focuses on 31 network performance metrics, including network delay, jitter, packet 32 loss, hop count, and bandwidth. Additional cost metrics may be 33 documented in other documents. 35 When using an ALTO performance metric, applications need additional 36 information beyond the metric. In particular, its "cost-source", 37 such as it being an estimation or an SLA, is key to define the 38 meaning of a performance metric. Hence, each ALTO performance metric 39 should include the "cost-source" of the metric. To report an 40 estimated value of a performance metric, the ALTO server may derive 41 and aggregate from routing protocols with different granularity and 42 scope, such as BGP-LS, OSPF-TE and ISIS-TE, or from end-to-end 43 traffic management tools. These metrics may then be exposed by an 44 ALTO Server to allow applications to determine "where" to connect 45 based on network performance criteria. 47 Requirements Language The key words "MUST", "MUST NOT", "REQUIRED", 48 "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", 49 and "OPTIONAL" in this document are to be interpreted as described in 50 RFC 2119 [RFC2119]. 52 Status of This Memo 54 This Internet-Draft is submitted in full conformance with the 55 provisions of BCP 78 and BCP 79. 57 Internet-Drafts are working documents of the Internet Engineering 58 Task Force (IETF). Note that other groups may also distribute 59 working documents as Internet-Drafts. The list of current Internet- 60 Drafts is at http://datatracker.ietf.org/drafts/current/. 62 Internet-Drafts are draft documents valid for a maximum of six months 63 and may be updated, replaced, or obsoleted by other documents at any 64 time. It is inappropriate to use Internet-Drafts as reference 65 material or to cite them other than as "work in progress." 67 This Internet-Draft will expire on May 7, 2020. 69 Copyright Notice 71 Copyright (c) 2019 IETF Trust and the persons identified as the 72 document authors. All rights reserved. 74 This document is subject to BCP 78 and the IETF Trust's Legal 75 Provisions Relating to IETF Documents 76 (http://trustee.ietf.org/license-info) in effect on the date of 77 publication of this document. Please review these documents 78 carefully, as they describe your rights and restrictions with respect 79 to this document. Code Components extracted from this document must 80 include Simplified BSD License text as described in Section 4.e of 81 the Trust Legal Provisions and are provided without warranty as 82 described in the Simplified BSD License. 84 Table of Contents 86 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 87 2. Network Performance Cost Metrics . . . . . . . . . . . . . . 6 88 2.1. Cost Metric: One Way Delay (owdelay) . . . . . . . . . . 6 89 2.1.1. Intended Semantics . . . . . . . . . . . . . . . . . 6 90 2.1.2. Use and Example . . . . . . . . . . . . . . . . . . . 6 91 2.1.3. Measurement Considerations . . . . . . . . . . . . . 7 92 2.2. Cost Metric: RoundTrip Time (rtt) . . . . . . . . . . . . 8 93 2.2.1. Intended Semantics . . . . . . . . . . . . . . . . . 8 94 2.2.2. Use and Example . . . . . . . . . . . . . . . . . . . 8 95 2.2.3. Measurement Considerations . . . . . . . . . . . . . 9 96 2.3. Cost Metric: Packet Delay Variation (pdv) . . . . . . . . 10 97 2.3.1. Intended Semantics . . . . . . . . . . . . . . . . . 10 98 2.3.2. Use and Example . . . . . . . . . . . . . . . . . . . 10 99 2.3.3. Measurement Considerations . . . . . . . . . . . . . 11 100 2.4. Cost Metric: Hop Count . . . . . . . . . . . . . . . . . 12 101 2.4.1. Intended Semantics . . . . . . . . . . . . . . . . . 12 102 2.4.2. Use and Example . . . . . . . . . . . . . . . . . . . 13 103 2.4.3. Measurement Considerations . . . . . . . . . . . . . 14 104 2.5. Cost Metric: Packet Loss . . . . . . . . . . . . . . . . 14 105 2.5.1. Intended Semantics . . . . . . . . . . . . . . . . . 14 106 2.5.2. Use and Example . . . . . . . . . . . . . . . . . . . 15 107 2.5.3. Measurement Considerations . . . . . . . . . . . . . 16 108 2.6. Cost Metric: Throughput . . . . . . . . . . . . . . . . . 16 109 2.6.1. Intended Semantics . . . . . . . . . . . . . . . . . 17 110 2.6.2. Use and Example . . . . . . . . . . . . . . . . . . . 17 111 2.6.3. Measurement Considerations . . . . . . . . . . . . . 18 112 3. Traffic Engineering Performance Cost Metrics . . . . . . . . 18 113 3.1. Cost Metric: Link Maximum Reservable Bandwidth . . . . . 19 114 3.1.1. Intended Semantics . . . . . . . . . . . . . . . . . 19 115 3.1.2. Use and Example . . . . . . . . . . . . . . . . . . . 19 116 3.1.3. Measurement Considerations . . . . . . . . . . . . . 20 117 3.2. Cost Metric: Link Residue Bandwidth . . . . . . . . . . . 21 118 3.2.1. Intended Semantics . . . . . . . . . . . . . . . . . 21 119 3.2.2. Use and Example . . . . . . . . . . . . . . . . . . . 21 120 3.2.3. Measurement Considerations . . . . . . . . . . . . . 22 121 4. Operational Considerations . . . . . . . . . . . . . . . . . 23 122 4.1. Source Considerations . . . . . . . . . . . . . . . . . . 23 123 4.2. Backward Compatibility Considerations . . . . . . . . . . 24 124 4.3. Computation Considerations . . . . . . . . . . . . . . . 24 125 4.3.1. Configuration Parameters Considerations . . . . . . . 24 126 4.3.2. Availability Considerations . . . . . . . . . . . . . 24 127 5. Security Considerations . . . . . . . . . . . . . . . . . . . 25 128 6. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 25 129 7. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 26 130 8. References . . . . . . . . . . . . . . . . . . . . . . . . . 26 131 8.1. Normative References . . . . . . . . . . . . . . . . . . 26 132 8.2. Informative References . . . . . . . . . . . . . . . . . 28 133 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 28 135 1. Introduction 137 Cost Metric is a basic concept in Application-Layer Traffic 138 Optimization (ALTO). It is used in both the ALTO cost map service 139 and the ALTO endpoint cost service, to allow applications to request 140 network cost metrics. 142 Different applications may use different cost metrics. Hence, the 143 ALTO base protocol [RFC7285] introduces an ALTO Cost Metric Registry 144 (Section 14.2 of [RFC7285]) as a systematic mechanism to allow 145 different metrics to be specified. For example, a more delay- 146 sensitive application may want to use latency related metrics, and a 147 more bandwidth-sensitive application may want to use bandwidth 148 related metrics. The ALTO base protocol [RFC7285], however, has 149 registered only one single cost metric, i.e., the generic 150 "routingcost" metric; no latency or bandwidth related metrics are 151 defined. 153 This document registers a set of new cost metrics specified in 154 Table 1, to support the aforementioned applications, to allow them to 155 determine "where" to connect based on network performance criteria. 156 This document follows the guideline (Section 14.2 of [RFC7285]) of 157 the ALTO base protocol on registering ALTO cost metrics. Hence it 158 specifies the identifier, the intended semantics, and the security 159 considerations of each one of the metrics defined in Table 1. 161 +--------------------------+-------------+-----------------------+ 162 | Metric | Definition | Origin | 163 +--------------------------+-------------+-----------------------+ 164 | One Way Delay | Section 2.1 | [RFC2679] Section 3.6 | 165 | Round Trip Delay | Section 2.2 | [RFC2681] Section 2.6 | 166 | Packet Delay Variation | Section 2.3 | [RFC3393] Section 2.6 | 167 | Hop Count | Section 2.4 | [RFC7285] | 168 | Packet Loss | Section 2.5 | [RFC7680] Section 2.6 | 169 | Throughput | Section 2.6 | [RFC6349] Section 3.3 | 170 | Max Reservable Bandwidth | Section 3.1 | [RFC5305] Section 3.5 | 171 | Residue Bandwidth | Section 3.2 | [RFC7810] Section 4.5 | 172 +------------+---------------------------------------------------+ 173 Table 1. Cost Metrics Defined in this Document 175 The purpose of this document is to ensure proper usage of the metrics 176 by ALTO clients. It does not claim novelty of the metrics. Some of 177 these metrics are already specified by standards such as IPPM; some 178 are ISP dependent such as those registered in ISIS or OSPF-TE. This 179 document will refer to the relevant specifications. 181 An ALTO server may provide only a subset of the cost metrics 182 described in this document. Hence, all cost metrics defined in this 183 document are optional and not all them need to be exposed to 184 applications. For example, those that are subject to privacy 185 concerns should not be provided to unauthorized ALTO clients. 187 When an ALTO server supports a cost metric defined in this document, 188 it MUST announce this metric in its information resource directory 189 (IRD). 191 To make clear how the value of an ALTO performance metric is derived, 192 this document defines an optional field named "cost-source" to extend 193 the ALTO "cost-type". The "cost-source" indicates how the metric is 194 derived, and currently it can be either "estimation" or "sla". If a 195 "cost-type" does not include the optional "cost-source" field, the 196 application MUST assume that the value of "cost-source" is 197 "estimation". 199 An ALTO server may compute "estimation" values by retrieving and/or 200 aggregating information from routing protocols or other traffic 201 measurement management tools, with corresponding operational issues. 202 A potential architecture on estimating these metrics is shown in 203 Figure 1 below. In Section 4, we discuss in more detail the 204 operations issues and how to address them. 206 +--------+ +--------+ +--------+ 207 | Client | | Client | | Client | 208 +----^---+ +---^----+ +---^----+ 209 | | | 210 +-----------|-----------+ 211 NBI |ALTO protocol 212 | 213 | 214 +--+-----+ retrieval +---------+ 215 | ALTO |<----------------| Routing | 216 | Server | and aggregation| | 217 | |<-------------+ | Protocol| 218 +--------+ | +---------+ 219 | 220 | +---------+ 221 | |Management 222 ---| | 223 | Tool | 224 +---------+ 225 Figure 1. Potential framework to compute performance cost metrics 227 An ALTO server introducing these metrics should also consider 228 security issues. As a generic security consideration on the 229 reliability and trust in the exposed metric values, applications 230 SHOULD rapidly give up using ALTO-based guidance if they feel the 231 exposed information does not preserve their performance level or even 232 degrades it. We discuss security considerations in more details in 233 Section 5. 235 Following the ALTO base protocol, this document uses JSON to specify 236 the value type of each defined metric. See [RFC4627] for JSON data 237 type specification. 239 2. Network Performance Cost Metrics 241 This section introduces generic ALTO network performance metrics such 242 as one way delay, round trip delay, hop count, packet loss, and 243 throughput. 245 2.1. Cost Metric: One Way Delay (owdelay) 247 Metric name: 249 One Way Delay 251 Metric Identifier: 253 owdelay 255 2.1.1. Intended Semantics 257 Metric Description: To specify spatial and temporal aggregated delay 258 of a stream of packets exchanged between the specified source and 259 destination or the time that the packet spends to travel from source 260 to destination. The spatial aggregation level is specified in the 261 query context (e.g., PID to PID, or endpoint to endpoint). 263 Metric Representation: The metric value type is a single 'JSONNumber' 264 type value conforming to the number specification of [RFC8259] 265 Section 6. The number MUST be non-negative. The unit is expressed 266 in milliseconds in this document. 268 2.1.2. Use and Example 270 This metric could be used as a cost metric constraint attribute used 271 either together with cost metric attribute 'routingcost' or on its 272 own or as a returned cost metric in the response. 274 Example 1: Delay value on source-destination endpoint pairs 276 POST /endpointcost/lookup HTTP/1.1 277 Host: alto.example.com 278 Content-Length: TBA 279 Content-Type: application/alto-endpointcostparams+json 280 Accept: 281 application/alto-endpointcost+json,application/alto-error+json 283 { 284 "cost-type": {"cost-mode" : "numerical", 285 "cost-metric" : "owdelay"}, 286 "endpoints" : { 287 "srcs": [ "ipv4:192.0.2.2" ], 288 "dsts": [ 289 "ipv4:192.0.2.89", 290 "ipv4:198.51.100.34", 291 "ipv6:2000::1:2345:6789:abcd" 292 ] 293 } 294 } 296 HTTP/1.1 200 OK 297 Content-Length: TBA 298 Content-Type: application/alto-endpointcost+json 299 { 300 "meta" :{ 301 "cost-type": {"cost-mode" : "numerical", 302 "cost-metric" : "owdelay" 303 } 304 }, 305 "endpoint-cost-map" : { 306 "ipv4:192.0.2.2": { 307 "ipv4:192.0.2.89" : 10, 308 "ipv4:198.51.100.34" : 20, 309 "ipv6:2000::1:2345:6789:abcd" : 30, 310 } 311 } 312 } 314 Comment: Since the "cost-type" does not include the "cost-source" 315 field, the values are based on "estimation". 317 2.1.3. Measurement Considerations 319 Method of Measurement or Calculation: 321 See section 8.3 of [I-D.ietf-ippm-initial-registry] for potential 322 measurement method. 324 Measurement Point(s) with Potential Measurement Domain: 326 See Section 4.1, Data sources for potential data sources. 328 Measurement Timing: 330 See section 8.3.5 of [I-D.ietf-ippm-initial-registry] for 331 potential measurement timing considerations. 333 2.2. Cost Metric: RoundTrip Time (rtt) 335 Metric name: 337 Round Trip Time 339 Metric Identifier: 341 rtt 343 2.2.1. Intended Semantics 345 Metric Description: To specify spatial and temporal aggregated round 346 trip delay between the specified source and destination or the time 347 that the packet spends to travel from source to destination and then 348 from destination to source. The spatial aggregation level is 349 specified in the query context (e.g., PID to PID, or endpoint to 350 endpoint). 352 Metric Representation: The metric value type is a single 'JSONNumber' 353 type value conforming to the number specification of [RFC8259] 354 Section 6. The number MUST be non-negative. The unit is expressed 355 in milliseconds in this document. 357 2.2.2. Use and Example 359 This metric could be used as a cost metric constraint attribute used 360 either together with cost metric attribute 'routingcost' or on its 361 own or as a returned cost metric in the response. 363 Example 2: Roundtrip Delay value on source-destination endpoint pairs 365 POST /endpointcost/lookup HTTP/1.1 366 Host: alto.example.com 367 Content-Length: TBA 368 Content-Type: application/alto-endpointcostparams+json 369 Accept: 370 application/alto-endpointcost+json,application/alto-error+json 372 { 373 "cost-type": {"cost-mode" : "numerical", 374 "cost-metric" : "rtt"}, 375 "endpoints" : { 376 "srcs": [ "ipv4:192.0.2.2" ], 377 "dsts": [ 378 "ipv4:192.0.2.89", 379 "ipv4:198.51.100.34", 380 "ipv6:2000::1:2345:6789:abcd" 381 ] 382 } 383 } 385 HTTP/1.1 200 OK 386 Content-Length: TBA 387 Content-Type: application/alto-endpointcost+json 388 { 389 "meta" :{ 390 "cost-type": {"cost-mode" : "numerical", 391 "cost-metric" : "rtt" 392 } 393 }, 394 "endpoint-cost-map" : { 395 "ipv4:192.0.2.2": { 396 "ipv4:192.0.2.89" : 4, 397 "ipv4:198.51.100.34" : 3, 398 "ipv6:2000::1:2345:6789:abcd" : 2, 399 } 400 } 401 } 403 2.2.3. Measurement Considerations 405 Method of Measurement or Calculation: 407 See section 4.3 of [I-D.ietf-ippm-initial-registry] for potential 408 measurement method. 410 Measurement Point(s) with Potential Measurement Domain: 412 See section 4.1, Data sources. 414 Measurement Timing: 416 See section 4.3.5 of [I-D.ietf-ippm-initial-registry] for 417 Measurement Timing. 419 2.3. Cost Metric: Packet Delay Variation (pdv) 421 Metric name: 423 Packet Delay Variation 425 Metric Identifier: 427 pdv 429 2.3.1. Intended Semantics 431 Metric Description: To specify spatial and temporal aggregated jitter 432 (packet delay variation) with respect to the minimum delay observed 433 on the stream over the specified source and destination. The spatial 434 aggregation level is specified in the query context (e.g., PID to 435 PID, or endpoint to endpoint). 437 Metric Representation: The metric value type is a single 'JSONNumber' 438 type value conforming to the number specification of [RFC8259] 439 Section 6. The number MUST be non-negative. The unit is expressed 440 in milliseconds in this document. 442 2.3.2. Use and Example 444 This metric could be used as a cost metric constraint attribute used 445 either together with cost metric attribute 'routingcost' or on its 446 own or as a returned cost metric in the response. 448 Example 3: PDV value on source-destination endpoint pairs 450 POST /endpointcost/lookup HTTP/1.1 451 Host: alto.example.com 452 Content-Length: TBA 453 Content-Type: application/alto-endpointcostparams+json 454 Accept: 455 application/alto-endpointcost+json,application/alto-error+json 457 { 458 "cost-type": {"cost-mode" : "numerical", 459 "cost-metric" : "pdv"}, 460 "endpoints" : { 461 "srcs": [ "ipv4:192.0.2.2" ], 462 "dsts": [ 463 "ipv4:192.0.2.89", 464 "ipv4:198.51.100.34", 465 "ipv6:2000::1:2345:6789:abcd" 466 ] 467 } 468 } 469 HTTP/1.1 200 OK 470 Content-Length: TBA 471 Content-Type: application/alto-endpointcost+json 472 { 473 "meta": { 474 "cost type": { 475 "cost-mode": "numerical", 476 "cost-metric":"delayjitter" 477 } 478 }, 479 "endpoint-cost-map": { 480 "ipv4:192.0.2.2": { 481 "ipv4:192.0.2.89" : 0 482 "ipv4:198.51.100.34" : 1 483 "ipv6:2000::1:2345:6789:abcd" : 5 484 } 485 } 486 } 488 2.3.3. Measurement Considerations 490 Method of Measurement or Calculation: 492 See Section 5.3 of [I-D.ietf-ippm-initial-registry] for potential 493 measurement method. 495 Measurement Point(s) with Potential Measurement Domain: 497 See Section 4.1, Data sources for potential data sources. 499 Measurement Timing: 501 See Section 5.3.5 of [I-D.ietf-ippm-initial-registry] for 502 Measurement Timing. 504 2.4. Cost Metric: Hop Count 506 The metric hopcount is mentioned in [RFC7285] Section 9.2.3 as an 507 example. This section further clarifies its properties. 509 Metric name: 511 Hop count 513 Metric Identifier: 515 hopcount 517 2.4.1. Intended Semantics 519 Metric Description: 521 To specify the number of hops in the path between the source 522 endpoint and the destination endpoint. The hop count is a basic 523 measurement of distance in a network and can be exposed as Router 524 Hops, in direct relation to the routing protocols originating this 525 information. 527 Metric Representation: 529 The metric value type is a single 'JSONNumber' type value 530 conforming to the number specification [RFC8259], Section 6. The 531 number MUST be an integer and non-negative. The value represents 532 the number of hops. 534 2.4.2. Use and Example 536 This metric could be used as a cost metric constraint attribute used 537 either together with cost metric attribute 'routingcost' or on its 538 own or as a returned cost metric in the response. 540 Example 4: hopcount value on source-destination endpoint pairs 542 POST /endpointcost/lookup HTTP/1.1 543 Host: alto.example.com 544 Content-Length: TBA 545 Content-Type: application/alto-endpointcostparams+json 546 Accept: 547 application/alto-endpointcost+json,application/alto-error+json 549 { 550 "cost-type": {"cost-mode" : "numerical", 551 "cost-metric" : "hopcount"}, 552 "endpoints" : { 553 "srcs": [ "ipv4:192.0.2.2" ], 554 "dsts": [ 555 "ipv4:192.0.2.89", 556 "ipv4:198.51.100.34", 557 "ipv6:2000::1:2345:6789:abcd" 558 ] 559 } 560 } 562 HTTP/1.1 200 OK 563 Content-Length: TBA 564 Content-Type: application/alto-endpointcost+json 565 { 566 "meta": { 567 "cost type": { 568 "cost-mode": "numerical", 569 "cost-metric":"hopcount"} 570 } 571 }, 572 "endpoint-cost-map": { 573 "ipv4:192.0.2.2": { 574 "ipv4:192.0.2.89" : 5, 575 "ipv4:198.51.100.34": 3, 576 "ipv6:2000::1:2345:6789:abcd" : 2, 577 } 578 } 579 } 581 2.4.3. Measurement Considerations 583 Method of Measurement or Calculation: 585 The hop count can be calculated based on the number of routers 586 from the source endpoint through which data must pass to reach the 587 destination endpoint. 589 Measurement Point(s) with Potential Measurement Domain: 591 The hop count can be measured at the source endpoint by 592 traceroute. 594 Measurement Timing: 596 Upon need, the traceroute can use UDP probe message or other 597 implementations that use ICMP and TCP to discover the hop counts 598 along the path from source endpoint to destination endpoint. 600 2.5. Cost Metric: Packet Loss 602 Metric name: 604 Packet loss 606 Metric Identifier: 608 pktloss 610 2.5.1. Intended Semantics 612 Metric Description: 614 To specify spatial and temporal aggregated packet loss over the 615 specified source and destination. The spatial aggregation level 616 is specified in the query context (e.g., PID to PID, or endpoint 617 to endpoint). 619 Metric Representation: 621 The metric value type is a single 'JSONNumber' type value 622 conforming to the number specification of [RFC8259] Section 6. 623 The number MUST be non-negative. The value represents the 624 percentage of packet loss. 626 2.5.2. Use and Example 628 This metric could be used as a cost metric constraint attribute used 629 either together with cost metric attribute 'routingcost' or on its 630 own or as a returned cost metric in the response. 632 Example 5: pktloss value on source-destination endpoint pairs 634 POST /endpointcost/lookup HTTP/1.1 635 Host: alto.example.com 636 Content-Length: TBA 637 Content-Type: application/alto-endpointcostparams+json 638 Accept: 639 application/alto-endpointcost+json,application/alto-error+json 641 { 642 "cost-type": {"cost-mode" : "numerical", 643 "cost-metric" : "pktloss"}, 644 "endpoints" : { 645 "srcs": [ "ipv4:192.0.2.2" ], 646 "dsts": [ 647 "ipv4:192.0.2.89", 648 "ipv4:198.51.100.34", 649 "ipv6:2000::1:2345:6789:abcd" 650 ] 651 } 652 } 654 HTTP/1.1 200 OK 655 Content-Length: TBA 656 Content-Type: application/alto-endpointcost+json 657 { 658 "meta": { 659 "cost type": { 660 "cost-mode": "numerical", 661 "cost-metric":"pktloss"} 662 } 663 }, 664 "endpoint-cost-map": { 665 "ipv4:192.0.2.2": { 666 "ipv4:192.0.2.89" : 0, 667 "ipv4:198.51.100.34": 0, 668 "ipv6:2000::1:2345:6789:abcd" : 0, 669 } 670 } 671 } 673 2.5.3. Measurement Considerations 675 Method of Measurement or Calculation: 677 See Section 2.6 of [RFC7680] for Measurement Method. 679 Measurement Point(s) with Potential Measurement Domain: 681 See Section 4.1 this document, Data sources. 683 Measurement Timing: 685 See Section 2 and Section 3 of [RFC7680] for Measurement Timing. 687 2.6. Cost Metric: Throughput 689 Metric name: 691 Throughput 693 Metric Identifier: 695 throughput 697 2.6.1. Intended Semantics 699 Metric Description: 701 To specify spatial and temporal throughput over the specified 702 source and destination. The spatial aggregation level is 703 specified in the query context (e.g., PID to PID, or endpoint to 704 endpoint). 706 Metric Representation: 708 The metric value type is a single 'JSONNumber' type value 709 conforming to the number specification of [RFC8259] Section 6. 710 The number MUST be non-negative. The unit is Mbps. 712 2.6.2. Use and Example 714 This metric could be used as a cost metric constraint attribute used 715 either together with cost metric attribute 'routingcost' or on its 716 own or as a returned cost metric in the response. 718 Example 5: throughtput value on source-destination endpoint pairs 720 POST /endpointcost/lookup HTTP/1.1 721 Host: alto.example.com 722 Content-Length: TBA 723 Content-Type: application/alto-endpointcostparams+json 724 Accept: 725 application/alto-endpointcost+json,application/alto-error+json 727 { 728 "cost-type": {"cost-mode" : "numerical", 729 "cost-metric" : "throughput"}, 730 "endpoints" : { 731 "srcs": [ "ipv4:192.0.2.2" ], 732 "dsts": [ 733 "ipv4:192.0.2.89", 734 "ipv4:198.51.100.34", 735 "ipv6:2000::1:2345:6789:abcd" 736 ] 737 } 738 } 739 HTTP/1.1 200 OK 740 Content-Length: TBA 741 Content-Type: application/alto-endpointcost+json 742 { 743 "meta": { 744 "cost type": { 745 "cost-mode": "numerical", 746 "cost-metric":"throughput" 747 } 748 } 749 "endpoint-cost-map": { 750 "ipv4:192.0.2.2": { 751 "ipv4:192.0.2.89" : 25.6, 752 "ipv4:198.51.100.34": 12.8, 753 "ipv6:2000::1:2345:6789:abcd" : 42.8, 754 } 755 } 757 2.6.3. Measurement Considerations 759 Method of Measurement or Calculation: 761 See Section 3.3 of [RFC6349] for Measurement Method. 763 Measurement Point(s) with Potential Measurement Domain: 765 See Section 4.1 of this document. 767 Measurement Timing: 769 Similar to RTT. See Section 4.3.5 of [I-D.ietf-ippm-initial- 770 registry] for Measurement Timing. 772 3. Traffic Engineering Performance Cost Metrics 774 This section introduces ALTO network performance metrics that may be 775 aggregated from network metrics measured on links and specified in 776 other documents. In particular, the bandwidth related metrics 777 specified in this section are only available through link level 778 measurements. For some of these metrics, the ALTO Server may further 779 expose aggregated values while specifying the aggregation laws. 781 3.1. Cost Metric: Link Maximum Reservable Bandwidth 783 Metric name: 785 Maximum Reservable Bandwidth 787 Metric Identifier: 789 maxresbw 791 3.1.1. Intended Semantics 793 Metric Description: 795 To specify spatial and temporal maximum reservable bandwidth over 796 the specified source and destination. The value is corresponding 797 to the maximum bandwidth that can be reserved (motivated from RFC 798 3630 Sec. 2.5.7.). The spatial aggregation unit is specified in 799 the query context (e.g., PID to PID, or endpoint to endpoint). 801 Metric Representation: 803 The metric value type is a single 'JSONNumber' type value that is 804 non-negative. The unit of measurement is Mbps. 806 3.1.2. Use and Example 808 This metric could be used as a cost metric constraint attribute used 809 either together with cost metric attribute 'routingcost' or on its 810 own or as a returned cost metric in the response. 812 Example 6: maxresbw value on source-destination endpoint pairs 814 POST/ endpointcost/lookup HTTP/1.1 815 Host: alto.example.com 816 Content-Length: TBA 817 Content-Type: application/alto-endpointcostparams+json 818 Accept: 819 application/alto-endpointcost+json,application/alto-error+json 821 { 822 "cost-type" { "cost-mode": "numerical", 823 "cost-metric": "maxresbw"}, 824 "endpoints": { 825 "srcs": [ "ipv4 : 192.0.2.2" ], 826 "dsts": [ 827 "ipv4:192.0.2.89", 828 "ipv4:198.51.100.34", 829 "ipv6:2000::1:2345:6789:abcd" 830 ] 831 } 832 } 834 HTTP/1.1 200 OK 835 Content-Length: TBA 836 Content-Type: application/alto-endpointcost+json 837 { 838 "meta": { 839 "cost-type": { 840 "cost-mode": "numerical", 841 "cost-metric": "maxresbw" 842 } 843 }, 844 " endpoint-cost-map": { 845 "ipv4:192.0.2.2" { 846 "ipv4:192.0.2.89" : 0, 847 "ipv4:198.51.100.34": 2000, 848 "ipv6:2000::1:2345:6789:abcd": 5000, 849 } 850 } 851 } 853 3.1.3. Measurement Considerations 855 Method of Measurement or Calculation: 857 Maximum Reservable Bandwidth is the bandwidth measured between two 858 directly connected IS-IS neighbors or OSPF neighbors. See 859 Section 3.5 of [RFC5305] for Measurement Method. 861 Measurement Point(s) with Potential Measurement Domain: 863 See Section 4.1 this document for discussions. 865 Measurement Timing: 867 See Section 3.5 of [RFC5305] and Section 5 of [RFC7810] for 868 Measurement Timing. 870 3.2. Cost Metric: Link Residue Bandwidth 872 Metric name: 874 Residue Bandwidth 876 Metric Identifier: 878 residuebw 880 3.2.1. Intended Semantics 882 Metric Description: 884 To specify spatial and temporal residual bandwidth over the 885 specified source and destination. The value is calculated by 886 subtracting tunnel reservations from Maximum Bandwidth (motivated 887 from [RFC7810], Section 4.5.). The spatial aggregation unit is 888 specified in the query context (e.g., PID to PID, or endpoint to 889 endpoint). 891 Metric Representation: 893 The metric value type is a single 'JSONNumber' type value that is 894 non-negative. The unit of measurement is Mbps. 896 3.2.2. Use and Example 898 This metric could be used as a cost metric constraint attribute used 899 either together with cost metric attribute 'routingcost' or on its 900 own or as a returned cost metric in the response. 902 Example 7: residuebw value on source-destination endpoint pairs 904 POST/ endpointcost/lookup HTTP/1.1 905 Host: alto.example.com 906 Content-Length: TBA 907 Content-Type: application/alto-endpointcostparams+json 908 Accept: 909 application/alto-endpointcost+json,application/alto-error+json 911 { 912 "cost-type": { "cost-mode": "numerical", 913 "cost-metric": "residuebw"}, 914 "endpoints": { 915 "srcs": [ "ipv4 : 192.0.2.2" ], 916 "dsts": [ 917 "ipv4:192.0.2.89", 918 "ipv4:198.51.100.34", 919 "ipv6:2000::1:2345:6789:abcd" 920 ] 921 } 922 } 924 HTTP/1.1 200 OK 925 Content-Length: TBA 926 Content-Type: application/alto-endpointcost+json 927 { 928 "meta": { 929 "cost-type" { 930 "cost-mode": "numerical", 931 "cost-metric": "residuebw" 932 } 933 }, 934 "endpoint-cost-map" { 935 "ipv4:192.0.2.2" { 936 "ipv4:192.0.2.89" : 0, 937 "ipv4:198.51.100.34": 2000, 938 "ipv6:2000::1:2345:6789:abcd": 5000, 939 } 940 } 941 } 943 3.2.3. Measurement Considerations 945 Method of Measurement or Calculation: 947 Residue Bandwidth is the Unidirectional Residue bandwidth measured 948 between two directly connected IS-IS neighbors or OSPF neighbors. 949 See Section 4.5 of [RFC7810] for Measurement Method. 951 Measurement Point(s) with Potential Measurement Domain: 953 See Section 4.1 of this document. 955 Measurement Timing: 957 See Section 5 of [RFC7810] for Measurement Timing. 959 4. Operational Considerations 961 The exact measurement infrastructure, measurement condition and 962 computation algorithms can vary from different networks, and are 963 outside the scope of this document. Both the ALTO server and the 964 ALTO clients, however, need to be cognizant of the operational issues 965 discussed below. 967 Also, the performance metrics specified in this document are similar, 968 in that they may use similar data sources and have similar issues in 969 their calculation. Hence, we specify common issues unless one metric 970 has its unique challenges. 972 4.1. Source Considerations 974 The additiona of the "cost-source" field is to solve a key issue: An 975 ALTO server needs data sources to compute the cost metrics described 976 in this document and an ALTO client needs to know the data sources to 977 better interpret the values. 979 To avoid too fine-grained information, this document introduces 980 "cost-source" to indicate only the high-level type of data sources: 981 "estimation" or "sla", where "estimation" is a type of measurement 982 data source and "sla" is a type that is more based on policy. 984 For estimation, for example, the ALTO server may use log servers or 985 the OAM system as its data source [RFC7971]. In particular, the cost 986 metrics defined in this document can be computed using routing 987 systems as the data sources. Mechanisms defined in [RFC2681], 988 [RFC3393], [RFC7679], [RFC7680], [RFC3630], [RFC3784], [RFC7471], 989 [RFC7810], [RFC7752] and [I-D.ietf-idr-te-pm-bgp] that allow an ALTO 990 Server to retrieve and derive the necessary information to compute 991 the metrics that we describe in this document. 993 4.2. Backward Compatibility Considerations 995 One potential issue introduced by the optional "cost-source" field is 996 backward compatibility. Consider that an IRD which defines two cost- 997 types with the same "cost-mode" and "cost-metric", but one with 998 "cost-source" being "estimation" and the other being "sla". Then an 999 ALTO client that is not aware of the extension will not be able to 1000 distinguish between these two types. A similar issue can arise even 1001 with a single cost-type which has "cost-source" being "sla", but the 1002 backward client will ignore this field and consider the metric 1003 estimation. [RFC7285] 1005 To address this issue, the only defined "routingcost" metric can be 1006 ONLY "estimation". 1008 4.3. Computation Considerations 1010 The metric values exposed by an ALTO server may result from 1011 additional processing on measurements from data sources to compute 1012 exposed metrics. This may involve data processing tasks such as 1013 aggregating the results across multiple systems, removing outliers, 1014 and creating additional statistics. There are two challenges on the 1015 computation of ALTO performance metrics. 1017 4.3.1. Configuration Parameters Considerations 1019 Performance metrics often depend on configuration parameters. For 1020 example, the value of packet loss rate depends on the measurement 1021 interval and varies over time. To handle this issue, an ALTO server 1022 may collect data on time periods covering the previous and current 1023 time or only collect data on present time. The ALTO server may 1024 further aggregate these data to provide an abstract and unified view 1025 that can be more useful to applications. To make the ALTO client 1026 better understand how to use these performance data, the ALTO server 1027 may provide the client with the validity period of the exposed metric 1028 values. 1030 4.3.2. Availability Considerations 1032 Applications value information relating to bandwidth availability 1033 whereas bandwidth related metrics can often be only measured at the 1034 link level. This document specifies a set of link-level bandwidth 1035 related values that may be exposed as such by an ALTO server. The 1036 server may also expose other metrics derived from their aggregation 1037 and having different levels of endpoint granularity, e.g., link 1038 endpoints or session endpoints. The metric specifications may also 1039 expose the utilized aggregation laws. 1041 5. Security Considerations 1043 The properties defined in this document present no security 1044 considerations beyond those in Section 15 of the base ALTO 1045 specification [RFC7285]. 1047 However concerns addressed in Sections "15.1 Authenticity and 1048 Integrity of ALTO Information", "15.2 Potential Undesirable Guidance 1049 from Authenticated ALTO Information" and "15.3 Confidentiality of 1050 ALTO Information" remain of utmost importance. Indeed, TE 1051 performance is a highly sensitive ISP information, therefore, sharing 1052 TE metric values in numerical mode requires full mutual confidence 1053 between the entities managing the ALTO Server and Client. Numerical 1054 TE performance information will most likely be distributed by ALTO 1055 Servers to Clients under strict and formal mutual trust agreements. 1056 On the other hand, ALTO Clients must be cognizant on the risks 1057 attached to such information that they would have acquired outside 1058 formal conditions of mutual trust. 1060 6. IANA Considerations 1062 IANA has created and now maintains the "ALTO Cost Metric Registry", 1063 listed in Section 14.2, Table 3 of [RFC7285]. This registry is 1064 located at . This document requests to add the 1066 following entries to "ALTO Cost Metric Registry". 1068 +------------+--------------------+ 1069 | Identifier | Intended Semantics | 1070 +------------+--------------------+ 1071 | owdelay | See Section 2.1 | 1072 | rtt | See Section 2.2 | 1073 | pdv | See Section 2.3 | 1074 | hopcount | See Section 2.4 | 1075 | pktloss | See Section 2.5 | 1076 | throughput | See Section 2.6 | 1077 | maxresbw | See Section 3.1 | 1078 | residuebw | See Section 3.2 | 1079 +------------+--------------------+ 1081 This document requests the creation of the "ALTO Cost Source 1082 Registry" with the following currently defined values: 1084 +------------+------------------------+ 1085 | Identifier | Intended Semantics | 1086 +------------+------------------------+ 1087 | estimation | Values by estimation | 1088 | sla | Values reflect service | 1089 | | level agreement | 1090 +------------+------------------------+ 1092 7. Acknowledgments 1094 The authors of this document would also like to thank Brian Trammell, 1095 Haizhou Du, Kai Gao, Lili Liu, Li, Geng, Danny Alex Lachos Perez for 1096 the reviews and comments. 1098 8. References 1100 8.1. Normative References 1102 [I-D.ietf-idr-te-pm-bgp] 1103 Ginsberg, L., Previdi, S., Wu, Q., Tantsura, J., and C. 1104 Filsfils, "BGP-LS Advertisement of IGP Traffic Engineering 1105 Performance Metric Extensions", draft-ietf-idr-te-pm- 1106 bgp-18 (work in progress), December 2018. 1108 [I-D.ietf-ippm-initial-registry] 1109 Morton, A., Bagnulo, M., Eardley, P., and K. D'Souza, 1110 "Initial Performance Metrics Registry Entries", draft- 1111 ietf-ippm-initial-registry-12 (work in progress), 1112 September 2019. 1114 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 1115 Requirement Levels", March 1997. 1117 [RFC2679] Almes, G., Kalidindi, S., and M. Zekauskas, "A One-way 1118 Delay Metric for IPPM", RFC 2679, DOI 10.17487/RFC2679, 1119 September 1999, . 1121 [RFC2681] Almes, G., Kalidindi, S., and M. Zekauskas, "A Round-trip 1122 Delay Metric for IPPM", RFC 2681, DOI 10.17487/RFC2681, 1123 September 1999, . 1125 [RFC3393] Demichelis, C. and P. Chimento, "IP Packet Delay Variation 1126 Metric for IP Performance Metrics (IPPM)", RFC 3393, DOI 1127 10.17487/RFC3393, November 2002, . 1130 [RFC4627] Crockford, D., "The application/json Media Type for 1131 JavaScript Object Notation (JSON)", RFC 4627, DOI 1132 10.17487/RFC4627, July 2006, . 1135 [RFC5234] Crocker, D., Ed. and P. Overell, "Augmented BNF for Syntax 1136 Specifications: ABNF", STD 68, RFC 5234, DOI 10.17487/ 1137 RFC5234, January 2008, . 1140 [RFC5305] Li, T. and H. Smit, "IS-IS Extensions for Traffic 1141 Engineering", RFC 5305, DOI 10.17487/RFC5305, October 1142 2008, . 1144 [RFC6349] Constantine, B., Forget, G., Geib, R., and R. Schrage, 1145 "Framework for TCP Throughput Testing", RFC 6349, DOI 1146 10.17487/RFC6349, August 2011, . 1149 [RFC7285] Alimi, R., Ed., Penno, R., Ed., Yang, Y., Ed., Kiesel, S., 1150 Previdi, S., Roome, W., Shalunov, S., and R. Woundy, 1151 "Application-Layer Traffic Optimization (ALTO) Protocol", 1152 RFC 7285, DOI 10.17487/RFC7285, September 2014, 1153 . 1155 [RFC7471] Giacalone, S., Ward, D., Drake, J., Atlas, A., and S. 1156 Previdi, "OSPF Traffic Engineering (TE) Metric 1157 Extensions", RFC 7471, DOI 10.17487/RFC7471, March 2015, 1158 . 1160 [RFC7679] Almes, G., Kalidindi, S., Zekauskas, M., and A. Morton, 1161 Ed., "A One-Way Delay Metric for IP Performance Metrics 1162 (IPPM)", STD 81, RFC 7679, DOI 10.17487/RFC7679, January 1163 2016, . 1165 [RFC7680] Almes, G., Kalidindi, S., Zekauskas, M., and A. Morton, 1166 Ed., "A One-Way Loss Metric for IP Performance Metrics 1167 (IPPM)", STD 82, RFC 7680, DOI 10.17487/RFC7680, January 1168 2016, . 1170 [RFC7752] Gredler, H., Ed., Medved, J., Previdi, S., Farrel, A., and 1171 S. Ray, "North-Bound Distribution of Link-State and 1172 Traffic Engineering (TE) Information Using BGP", RFC 7752, 1173 DOI 10.17487/RFC7752, March 2016, . 1176 [RFC7810] Previdi, S., Ed., Giacalone, S., Ward, D., Drake, J., and 1177 Q. Wu, "IS-IS Traffic Engineering (TE) Metric Extensions", 1178 RFC 7810, DOI 10.17487/RFC7810, May 2016, 1179 . 1181 8.2. Informative References 1183 [RFC6390] Clark, A. and B. Claise, "Framework for Performance Metric 1184 Development", RFC 6390, July 2011. 1186 [RFC7971] Stiemerling, M., Kiesel, S., Scharf, M., Seidel, H., and 1187 S. Previdi, "Application-Layer Traffic Optimization (ALTO) 1188 Deployment Considerations", RFC 7971, DOI 10.17487/ 1189 RFC7971, October 2016, . 1192 Authors' Addresses 1194 Qin Wu 1195 Huawei 1196 101 Software Avenue, Yuhua District 1197 Nanjing, Jiangsu 210012 1198 China 1200 Email: bill.wu@huawei.com 1202 Y. Richard Yang 1203 Yale University 1204 51 Prospect St 1205 New Haven, CT 06520 1206 USA 1208 Email: yry@cs.yale.edu 1210 Young Lee 1211 Huawei 1212 1700 Alma Drive, Suite 500 1213 Plano, TX 75075 1214 USA 1216 Email: leeyoung@huawei.com 1217 Dhruv Dhody 1218 Huawei 1219 Leela Palace 1220 Bangalore, Karnataka 560008 1221 INDIA 1223 Email: dhruv.ietf@gmail.com 1225 Sabine Randriamasy 1226 Nokia Bell Labs 1227 Route de Villejust 1228 Nozay 91460 1229 FRANCE 1231 Email: sabine.randriamasy@nokia-bell-labs.com