ALTO Working Group Q. Wu Internet-Draft Huawei Intended status: Standards Track Y. Yang Expires: January 13, 2021 Yale University Y. Lee Samsung D. Dhody Huawei S. Randriamasy Nokia Bell Labs L. Contreras Telefonica July 12, 2020 ALTO Performance Cost Metrics draft-ietf-alto-performance-metrics-12 Abstract Cost metric is a basic concept in Application-Layer Traffic Optimization (ALTO), and is used in basic ALTO services including both the cost map service and the endpoint cost service. Different applications may use different cost metrics, but the ALTO base protocol [RFC7285] defines only a single cost metric, i.e., the generic "routingcost" metric; see Sec. 14.2 of [RFC7285]. Hence, if the ALTO client of an application wants to issue a cost map or an endpoint cost request to determine the resource provider that offers better delay performance (i.e., low-delay) to a resource consumer, the base protocol does not define the cost metric to be used. This document addresses the issue by introducing network performance metrics, including network delay, jitter, packet loss rate, hop count, and bandwidth. The ALTO server may derive and aggregate such performance metrics from routing protocols such as BGP-LS, OSPF-TE and ISIS-TE, or from end-to-end traffic management tools, and then expose the information to allow applications to determine "where" to connect based on network performance criteria. There are multiple sources to derive the performance metrics. For example, whether the metric reported is an estimation based on measurements or it is a service-level agreement (SLA) can define the meaning of the performance metric. Hence, an application may need additional contextual information beyond the metric value. This document introduces an additional "cost-context" field to the ALTO "cost-type" field to convey such information. Wu, et al. Expires January 13, 2021 [Page 1] Internet-Draft ALTO Performance Cost Metrics July 2020 Requirements Language The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in [RFC2119]. Status of This Memo This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79. Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet- Drafts is at http://datatracker.ietf.org/drafts/current/. Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress." This Internet-Draft will expire on January 13, 2021. Copyright Notice Copyright (c) 2020 IETF Trust and the persons identified as the document authors. All rights reserved. This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (http://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License. Table of Contents 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 4 2. Performance Metric Attributes . . . . . . . . . . . . . . . . 5 2.1. Performance Metric Context: cost-context . . . . . . . . 6 2.2. Performance Metric Statistics . . . . . . . . . . . . . . 8 3. Packet Performance Metrics . . . . . . . . . . . . . . . . . 9 3.1. Cost Metric: One-Way Delay (delay-ow) . . . . . . . . . . 9 3.1.1. Base Identifier . . . . . . . . . . . . . . . . . . . 10 3.1.2. Value Representation . . . . . . . . . . . . . . . . 10 3.1.3. Intended Semantics and Use . . . . . . . . . . . . . 10 Wu, et al. Expires January 13, 2021 [Page 2] Internet-Draft ALTO Performance Cost Metrics July 2020 3.1.4. Cost-Context Specification Considerations . . . . . . 11 3.2. Cost Metric: Round-trip Delay (delay-rt) . . . . . . . . 12 3.2.1. Base Identifier . . . . . . . . . . . . . . . . . . . 12 3.2.2. Value Representation . . . . . . . . . . . . . . . . 12 3.2.3. Intended Semantics and Use . . . . . . . . . . . . . 12 3.2.4. Cost-Context Specification Considerations . . . . . . 13 3.3. Cost Metric: Delay Variation (delay-variation) . . . . . 14 3.3.1. Base Identifier . . . . . . . . . . . . . . . . . . . 14 3.3.2. Value Representation . . . . . . . . . . . . . . . . 14 3.3.3. Intended Semantics and Use . . . . . . . . . . . . . 14 3.3.4. Cost-Context Specification Considerations . . . . . . 15 3.4. Cost Metric: Hop Count (hopcount) . . . . . . . . . . . . 16 3.4.1. Base Identifier . . . . . . . . . . . . . . . . . . . 16 3.4.2. Value Representation . . . . . . . . . . . . . . . . 16 3.4.3. Intended Semantics and Use . . . . . . . . . . . . . 16 3.4.4. Cost-Context Specification Considerations . . . . . . 17 3.5. Cost Metric: Loss Rate (lossrate) . . . . . . . . . . . . 18 3.5.1. Base Identifier . . . . . . . . . . . . . . . . . . . 18 3.5.2. Value Representation . . . . . . . . . . . . . . . . 18 3.5.3. Intended Semantics and Use . . . . . . . . . . . . . 18 3.5.4. Cost-Context Specification Considerations . . . . . . 19 4. Bandwidth Performance Metrics . . . . . . . . . . . . . . . . 20 4.1. Cost Metric: TCP Throughput (tput) . . . . . . . . . . . 20 4.1.1. Base Identifier . . . . . . . . . . . . . . . . . . . 20 4.1.2. Value Representation . . . . . . . . . . . . . . . . 20 4.1.3. Intended Semantics and Use . . . . . . . . . . . . . 20 4.1.4. Cost-Context Specification Considerations . . . . . . 21 4.2. Cost Metric: Residue Bandwidth (bw-residue) . . . . . . . 22 4.2.1. Base Identifier . . . . . . . . . . . . . . . . . . . 22 4.2.2. Value Representation . . . . . . . . . . . . . . . . 22 4.2.3. Intended Semantics and Use . . . . . . . . . . . . . 22 4.2.4. Cost-Context Specification Considerations . . . . . . 23 4.3. Cost Metric: Maximum Reservable Bandwidth (bw-maxres) . . 24 4.3.1. Base Identifier . . . . . . . . . . . . . . . . . . . 24 4.3.2. Value Representation . . . . . . . . . . . . . . . . 24 4.3.3. Intended Semantics and Use . . . . . . . . . . . . . 24 4.3.4. Cost-Context Specification Considerations . . . . . . 25 5. Operational Considerations . . . . . . . . . . . . . . . . . 26 5.1. Source Considerations . . . . . . . . . . . . . . . . . . 26 5.2. Metric Timestamp Consideration . . . . . . . . . . . . . 27 5.3. Backward Compatibility Considerations . . . . . . . . . . 27 5.4. Computation Considerations . . . . . . . . . . . . . . . 27 5.4.1. Configuration Parameters Considerations . . . . . . . 27 5.4.2. Availability Considerations . . . . . . . . . . . . . 28 6. Security Considerations . . . . . . . . . . . . . . . . . . . 28 7. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 28 8. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 29 9. References . . . . . . . . . . . . . . . . . . . . . . . . . 29 Wu, et al. Expires January 13, 2021 [Page 3] Internet-Draft ALTO Performance Cost Metrics July 2020 9.1. Normative References . . . . . . . . . . . . . . . . . . 29 9.2. Informative References . . . . . . . . . . . . . . . . . 30 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 31 1. Introduction Cost Metric is a basic concept in Application-Layer Traffic Optimization (ALTO). It is used in both the ALTO cost map service and the ALTO endpoint cost service in the ALTO base protocol [RFC7285]. Since different applications may use different cost metrics, the ALTO base protocol introduces an ALTO Cost Metric Registry (Section 14.2 of [RFC7285]), as a systematic mechanism to allow different metrics to be specified. For example, a delay-sensitive application may want to use latency related metrics, and a bandwidth-sensitive application may want to use bandwidth related metrics. The ALTO base protocol, however, has registered only a single cost metric, i.e., the generic "routingcost" metric; no latency or bandwidth related metrics are defined. This document registers a set of new cost metrics specified in Table 1, to allow applications to better determine "where" to connect based on network performance criteria. This document follows the guideline defined in Section 14.2 of the ALTO base protocol [RFC7285]) on registering ALTO cost metrics. Hence it specifies the identifier, the intended semantics, and the security considerations of each one of the metrics defined in Table 1. +--------------------------+-------------+-------------------+ | Metric | Definition | Origin Example | +--------------------------+-------------+-------------------+ | One-way Delay | Section 3.1 | [RFC7679] | | Round-trip Delay | Section 3.2 | [RFC2681] | | Delay Variation | Section 3.3 | [RFC3393] | | Hop Count | Section 3.4 | [RFC7285] | | Loss Rate | Section 3.5 | [RFC7680] | | | | | | TCP Throughput | Section 4.1 | [RFC6349] | | Residue Bandwidth | Section 4.2 | [RFC7810] | | Max Reservable Bandwidth | Section 4.3 | [RFC5305] | +------------+-----------------------------------------------+ Table 1. Cost Metrics Defined in this Document. The purpose of this document is to ensure proper usage of the performance metrics defined in Table 1; it does not claim novelty of the metrics. The Origin column of Table 1 gives the RFC which defines each metric. Wu, et al. Expires January 13, 2021 [Page 4] Internet-Draft ALTO Performance Cost Metrics July 2020 We can rough classify the performance metrics into two categories: those derived from the performance of individual packets (i.e., one- way delay, round-trip delay, delay variation, hop count, and loss rate), and those related with bandwidth (TCP throughput, residue bandwidth and max reservable bandwidth). These two categories are defined in Section 3 and Section 4 respectively. Note that all metrics except round trip delay are unidirectional. Hence, a client will need to query both directions if needed. An ALTO server may provide only a subset of the metrics described in this document. For example, those that are subject to privacy concerns should not be provided to unauthorized ALTO clients. Hence, all cost metrics defined in this document are optional and not all of them need to be exposed to a given application. When an ALTO server supports a cost metric defined in this document, it should announce this metric in its information resource directory (IRD). An ALTO server introducing these metrics should consider security issues. As a generic security consideration on the reliability and trust in the exposed metric values, applications SHOULD rapidly give up using ALTO-based guidance if they detect that the exposed information does not preserve their performance level or even degrades it. This document discusses security considerations in more details in Section 6. Following the ALTO base protocol, this document uses JSON to specify the value type of each defined metric. See [RFC8259] for JSON data type specification. 2. Performance Metric Attributes When defining the metrics in Table 1, this document considers the guidelines specified in [RFC6390], which requires fine-grained specification of (i) Metric Name, (ii) Metric Description, (iii) Method of Measurement or Calculation, (iv) Units of Measurement, (v) Measurement Points, and (vi) Measurement Timing. In particular, for each metric, this document defines (i) Metric Name, (ii) Metric Description, and (iv) Units of Measurement. The Measurement Points are always specified by the specific ALTO services; for example, endpoint cost service is between the two end points. On the other hand, to be able to use coarse-grained information such as routing system information (e.g., [RFC8571]), which may not provide fine-grained information such as (iii) Method of Measurement or Calculation and (vi) Measurement Timing, this document provides context information to indicate the source of information and hence available metric details. Wu, et al. Expires January 13, 2021 [Page 5] Internet-Draft ALTO Performance Cost Metrics July 2020 2.1. Performance Metric Context: cost-context The details of a performance metric depend on the source of the information. Specifically, this document defines four types of information sources: "nominal", and "sla" (service level agreement), "import", and "estimation". For a given type of source, precise interpretation of a performance metric value can depend on particular measurement and computation parameters. For example, see Section 3.8 of [RFC7679] on items which a more complete measurement-based report should include. To make it possible to specify the source and the aforementioned parameters, this document introduces an optional "cost-context" field to the "cost-type" field defined by the ALTO base protocol (Section 10.7 of [RFC7285]) as the following: object { CostMetric cost-metric; CostMode cost-mode; [CostContext cost-context;] [JSONString description;] } CostType; object { JSONString cost-source; [JSONValue parameters;] } CostContext; The "cost-source" field of the "cost-context" field MUST be one of four category values: "nominal", "sla", "import", and "estimation". "cost-context" will not be used as a key to distinguish among performance metrics. Hence, an ALTO information resource SHOULD NOT announce multiple CostType with the same "cost-metric" and "cost- mode". They can be placed into different information resources. The "nominal" category indicates that the value of the metric is statically configured by the underlying devices. Not all metrics have reasonable "nominal" values. For example, throughput can have a nominal value, which indicates the configured transmission rate of the devices; latency typically do not have a nominal value. The "sla" category indicates that the value of the metric is derived from some commitment which this document refers to as service-level agreement (SLA). Some operators also use terms such as "target" or Wu, et al. Expires January 13, 2021 [Page 6] Internet-Draft ALTO Performance Cost Metrics July 2020 "committed" values. For a "sla" metric, it is RECOMMENDED that the "parameters" field provides a link to the SLA definition. The "estimation" category indicates that the value of the metric is computed through an estimation process. An ALTO server may compute "estimation" values by retrieving and/or aggregating information from routing protocols (e.g., [RFC8571]) and traffic measurement management tools (e.g., TWAMP [RFC5357]), with corresponding operational issues. A potential architecture on estimating these metrics is shown in Figure 1 below. Section 5 will discuss in more detail the operational issues and how a network may address them. +--------+ +--------+ +--------+ | Client | | Client | | Client | +----^---+ +---^----+ +---^----+ | | | +-----------|-----------+ NBI |ALTO protocol | | +--+-----+ retrieval +-----------+ | ALTO |<----------------| Routing | | Server | and aggregation| | | |<-------------+ | Protocols | +--------+ | +----------+ | | +-----------+ | |Management | ---| | | Tool | +-----------+ Figure 1. Potential framework to compute estimation to performance metrics A particular type of "estimation is direct "import", which indicates that the value of the metric is imported directly from a specific existing protocol or system. Specifying "import" as source instead of the more generic "estimation" may allow better tracing of information flow. For an "import" metric, it is RECOMMENDED that the "parameters" field provides details to the system from which raw data is imported. In particular, one may notice that the set of end-to- end metrics defined in Table 1 has large overlap with the set defined in [RFC8571], in the setting of IGP traffic engineering performance metrics for each link (i.e., unidirectional link delay, min/max unidirectional link delay, unidirectional delay variation, unidirectional link loss, unidirectional residual bandwidth, unidirectional available bandwidth, unidirectional utilized bandwidth). Hence, an ALTO server may use "import" to indicate that Wu, et al. Expires January 13, 2021 [Page 7] Internet-Draft ALTO Performance Cost Metrics July 2020 its end-to-end metrics are computed from link metrics imported from [RFC8571]. There can be overlap in deciding the cost-source category. It is the operator of an ALTO server who chooses the category. If a metric does not include a "cost-source" value, the application MUST assume that the value of "cost-source" is the most generic "estimation". 2.2. Performance Metric Statistics The measurement of a performance metric often yields a set of samples from an observation distribution ([Prometheus], instead of a single value. This document considers that the samples are aggregated as a statistic when reported. Hence, each performance metric's identifier should indicate the statistic (i.e., an aggregation operation), to become -, where MUST be one of the following: percentile, with letter p followed by a number p: gives the p percentile. Specifically, consider the samples coming from a random variable X. The metric returns x, relative to 100, such that the probability of X is less than or equal to x, i.e., Prob(X <= x) = p/100. The number p MUST be a non-negative JSON number in the range [0, 100] (i.e., greater than or equal to 0 and less than or equal to 100). To avoid complex identifiers, the number MUST NOT include the minus or the exp component (Section 6 of [RFC8259]). For example, delay-ow-p75 gives the 75% percentile of observed one-way delay; delay-ow-p99.9 gives the 99.9% percentile of delay. Note that some systems use quantile, which is in the range [0, 1]. This document uses percentile to make the identifier easier to read. min: the minimal value of the observation distribution. max: the maximal value of the observations. median: Wu, et al. Expires January 13, 2021 [Page 8] Internet-Draft ALTO Performance Cost Metrics July 2020 the mid point (i.e., p50) of the observation distribution. mean: the arithmetic mean value of the observations. stddev: the standard deviation of the observations. stdvar: the standard variance of the observations. If a metric has no (and hence no - as well), the metric MUST be considered as the 50 percentile (median). Since this scheme is common for all metrics defined in this document, below we only specify the base identifier. 3. Packet Performance Metrics This section introduces ALTO network performance metrics including one way delay, round trip delay, delay variation, hop count, and packet loss rate. They measure the "quality of experience" of the stream of packets sent from a resource provider to a resource consumer. The measures of each individual packet (pkt) can include the delay from the time that the packet enters the network to the time that the packet leaves the network (pkt.delay); the number of network hops that the packet traverses (pkt.hopcount); and whether the packet is dropped before reaching destination (pkt.dropped). The semantics of the performance metrics defined in this section is that they are statistics (percentiles) computed from these measures; for example, the x-percentile of the one-way delay is the x-percentile of the set of delays {pkt.delay} for the packets in the stream. 3.1. Cost Metric: One-Way Delay (delay-ow) Wu, et al. Expires January 13, 2021 [Page 9] Internet-Draft ALTO Performance Cost Metrics July 2020 3.1.1. Base Identifier The base identifier for this performance metric is "delay-ow". 3.1.2. Value Representation The metric value type is a single 'JSONNumber' type value conforming to the number specification of [RFC8259] Section 6. The unit is expressed in milliseconds. Hence, the number can be a floating point number to express delay that is smaller than milliseconds. The number MUST be non-negative. 3.1.3. Intended Semantics and Use Intended Semantics: To specify spatial and temporal aggregated delay of a stream of packets from the specified source and the specified destination. The spatial aggregation level is specified in the query context (e.g., PID to PID, or endpoint to endpoint). Use: This metric could be used as a cost metric constraint attribute or as a returned cost metric in the response. Example 1: Delay value on source-destination endpoint pairs POST /endpointcost/lookup HTTP/1.1 Host: alto.example.com Content-Length: TBA Content-Type: application/alto-endpointcostparams+json Accept: application/alto-endpointcost+json,application/alto-error+json { "cost-type": {"cost-mode" : "numerical", "cost-metric" : "delay-ow"}, "endpoints" : { "srcs": [ "ipv4:192.0.2.2" ], "dsts": [ "ipv4:192.0.2.89", "ipv4:198.51.100.34", "ipv6:2000::1:2345:6789:abcd" ] } } Wu, et al. Expires January 13, 2021 [Page 10] Internet-Draft ALTO Performance Cost Metrics July 2020 HTTP/1.1 200 OK Content-Length: TBA Content-Type: application/alto-endpointcost+json { "meta" :{ "cost-type": {"cost-mode" : "numerical", "cost-metric" : "delay-ow" } }, "endpoint-cost-map" : { "ipv4:192.0.2.2": { "ipv4:192.0.2.89" : 10, "ipv4:198.51.100.34" : 20, "ipv6:2000::1:2345:6789:abcd" : 30, } } } Comment: Since the "cost-type" does not include the "cost-source" field, the values are based on "estimation". Since the identifier does not include the - component, the values will represent median values. 3.1.4. Cost-Context Specification Considerations "nominal": Typically network one-way delay does not have a nominal value. "sla": Many networks provide delay in their application-level service level agreements. It is RECOMMENDED that the "parameters" field of an "sla" one-way delay metric provides a link ("link") to the SLA definition. "import": There can be multiple sources to import one-way delay. For example, if the import is from [RFC8571] (by using unidirectional link delay, min/max unidirectional link delay), it is RECOMMENDED that "parameters" provides "protocol" as a field and "RFC8571" as the value. During import, the server should be cognizant of potential issues when computing an end-to-end summary statistics from a link statistics. Another example import source is the IPPM framework. For IPPM, it is recommended that "parameters" provides "protocol" as a field and "ippm" as the value; see Section 4 of [I-D.ietf-ippm- initial-registry] for additional fields which can be specified for "ippm" in "parameters". "estimation": The exact estimation method is out of the scope of this document. It is RECOMMENDED that the "parameters" field of an Wu, et al. Expires January 13, 2021 [Page 11] Internet-Draft ALTO Performance Cost Metrics July 2020 "estimation" one-way delay metric provides a link ("link") to a description of the "estimation" method. 3.2. Cost Metric: Round-trip Delay (delay-rt) 3.2.1. Base Identifier The base identifier for this performance metric is "delay-rt". 3.2.2. Value Representation The metric value type is a single 'JSONNumber' type value conforming to the number specification of [RFC8259] Section 6. The number MUST be non-negative. The unit is expressed in milliseconds. 3.2.3. Intended Semantics and Use Intended Semantics: To specify spatial and temporal aggregated round- trip delay between the specified source and specified destination. The spatial aggregation level is specified in the query context (e.g., PID to PID, or endpoint to endpoint). Note that it is possible for a client to query two one-way delays and then compute the round-trip delay. The server should be cognizant of the consistency of values. Use: This metric could be used either as a cost metric constraint attribute or as a returned cost metric in the response. Wu, et al. Expires January 13, 2021 [Page 12] Internet-Draft ALTO Performance Cost Metrics July 2020 Example 2: Round-trip Delay value on source-destination endpoint pairs POST /endpointcost/lookup HTTP/1.1 Host: alto.example.com Content-Length: TBA Content-Type: application/alto-endpointcostparams+json Accept: application/alto-endpointcost+json,application/alto-error+json { "cost-type": {"cost-mode" : "numerical", "cost-metric" : "delay-rt"}, "endpoints" : { "srcs": [ "ipv4:192.0.2.2" ], "dsts": [ "ipv4:192.0.2.89", "ipv4:198.51.100.34", "ipv6:2000::1:2345:6789:abcd" ] } } HTTP/1.1 200 OK Content-Length: TBA Content-Type: application/alto-endpointcost+json { "meta" :{ "cost-type": {"cost-mode" : "numerical", "cost-metric" : "delay-rt" } }, "endpoint-cost-map" : { "ipv4:192.0.2.2": { "ipv4:192.0.2.89" : 4, "ipv4:198.51.100.34" : 3, "ipv6:2000::1:2345:6789:abcd" : 2, } } } 3.2.4. Cost-Context Specification Considerations "nominal": Typically network round-trip delay does not have a nominal value. Wu, et al. Expires January 13, 2021 [Page 13] Internet-Draft ALTO Performance Cost Metrics July 2020 "sla": It is RECOMMENDED that the "parameters" field of an "sla" round-trip delay metric provides a link ("link") to the SLA definition. "import": There can be multiple sources to import round-trip delay. If the import is from [RFC8571] (by using unidirectional link delay, min/max unidirectional link delay), it is RECOMMENDED that "parameters" provides "protocol" as a field and "RFC8571" as the value; see Section 3.1.4 for discussions on summing up link metrics to obtain end-to-end metrics. If the import is from the IPPM framework, it is recommended that "parameters" provides "protocol" as a field and "ippm" as the value; see Section 4 of [I-D.ietf-ippm- initial-registry] for additional fields which can be specified for "ippm" in "parameters". "estimation": The exact estimation method is out of the scope of this document. It is RECOMMENDED that the "parameters" field of an "estimation" round-trip delay metric provides a link ("link") to a description of the "estimation" method. 3.3. Cost Metric: Delay Variation (delay-variation) 3.3.1. Base Identifier The base identifier for this performance metric is "delay-variation". 3.3.2. Value Representation The metric value type is a single 'JSONNumber' type value conforming to the number specification of [RFC8259] Section 6. The number MUST be non-negative. The unit is expressed in milliseconds. 3.3.3. Intended Semantics and Use Intended Semantics: To specify spatial and temporal aggregated delay variation (also called delay jitter)) with respect to the minimum delay observed on the stream over the specified source and destination. The spatial aggregation level is specified in the query context (e.g., PID to PID, or endpoint to endpoint). Note that in statistics, variations are typically evaluated by the distance from samples relative to the mean. In networking context, it is more commonly defined from samples relative to the min. This definition follows the networking convention. Use: This metric could be used either as a cost metric constraint attribute or as a returned cost metric in the response. Wu, et al. Expires January 13, 2021 [Page 14] Internet-Draft ALTO Performance Cost Metrics July 2020 Example 3: Delay variation value on source-destination endpoint pairs POST /endpointcost/lookup HTTP/1.1 Host: alto.example.com Content-Length: TBA Content-Type: application/alto-endpointcostparams+json Accept: application/alto-endpointcost+json,application/alto-error+json { "cost-type": {"cost-mode" : "numerical", "cost-metric" : "delay-var"}, "endpoints" : { "srcs": [ "ipv4:192.0.2.2" ], "dsts": [ "ipv4:192.0.2.89", "ipv4:198.51.100.34", "ipv6:2000::1:2345:6789:abcd" ] } } HTTP/1.1 200 OK Content-Length: TBA Content-Type: application/alto-endpointcost+json { "meta": { "cost type": { "cost-mode": "numerical", "cost-metric":"delay-var" } }, "endpoint-cost-map": { "ipv4:192.0.2.2": { "ipv4:192.0.2.89" : 0 "ipv4:198.51.100.34" : 1 "ipv6:2000::1:2345:6789:abcd" : 5 } } } 3.3.4. Cost-Context Specification Considerations "nominal": Typically network delay variation does not have a nominal value. "sla": It is RECOMMENDED that the "parameters" field of an "sla" delay variation metric provides a link ("link") to the SLA definition. Wu, et al. Expires January 13, 2021 [Page 15] Internet-Draft ALTO Performance Cost Metrics July 2020 "import": There can be multiple sources to import delay variation. If the import is from [RFC8571] (by using unidirectional delay variation), it is RECOMMENDED that "parameters" provides "protocol" as a field and "RFC8571" as the value; see Section 3.1.4 for discussions on summing up link metrics to obtain end-to-end metrics. If the import is from the IPPM framework, it is recommended that "parameters" provides "protocol" as a field and "ippm" as the value; see Section 4 of [I-D.ietf-ippm-initial-registry] for additional fields which can be specified for "ippm" in "parameters". "estimation": The exact estimation method is out of the scope of this document. It is RECOMMENDED that the "parameters" field of an "estimation" delay variation metric provides a link ("link") to a description of the "estimation" method. 3.4. Cost Metric: Hop Count (hopcount) The metric hopcount is mentioned in [RFC7285] Section 9.2.3 as an example. This section further clarifies its properties. 3.4.1. Base Identifier The base identifier for this performance metric is "hopcount". 3.4.2. Value Representation The metric value type is a single 'JSONNumber' type value conforming to the number specification of [RFC8259] Section 6. The number MUST be a non-negative integer (greater than or equal to 0). The value represents the number of hops. 3.4.3. Intended Semantics and Use Intended Semantics: To specify the number of hops in the path from the specified source to the specified destination. The hop count is a basic measurement of distance in a network and can be exposed as router hops, in direct relation to the routing protocols originating this information. The spatial aggregation level is specified in the query context (e.g., PID to PID, or endpoint to endpoint). Use: This metric could be used as a cost metric constraint attribute or as a returned cost metric in the response. Wu, et al. Expires January 13, 2021 [Page 16] Internet-Draft ALTO Performance Cost Metrics July 2020 Example 4: hopcount value on source-destination endpoint pairs POST /endpointcost/lookup HTTP/1.1 Host: alto.example.com Content-Length: TBA Content-Type: application/alto-endpointcostparams+json Accept: application/alto-endpointcost+json,application/alto-error+json { "cost-type": {"cost-mode" : "numerical", "cost-metric" : "hopcount"}, "endpoints" : { "srcs": [ "ipv4:192.0.2.2" ], "dsts": [ "ipv4:192.0.2.89", "ipv4:198.51.100.34", "ipv6:2000::1:2345:6789:abcd" ] } } HTTP/1.1 200 OK Content-Length: TBA Content-Type: application/alto-endpointcost+json { "meta": { "cost type": { "cost-mode": "numerical", "cost-metric":"hopcount"} } }, "endpoint-cost-map": { "ipv4:192.0.2.2": { "ipv4:192.0.2.89" : 5, "ipv4:198.51.100.34": 3, "ipv6:2000::1:2345:6789:abcd" : 2, } } } 3.4.4. Cost-Context Specification Considerations "nominal": Typically hop count does not have a nominal value. "sla": Typically hop count does not have an SLA value. Wu, et al. Expires January 13, 2021 [Page 17] Internet-Draft ALTO Performance Cost Metrics July 2020 "import": There can be multiple sources to import hop count such as IGP routing protocols. "estimation": The exact estimation method is out of the scope of this document. It is RECOMMENDED that the "parameters" field of an "estimation" hop count metric provides a link ("link") to a description of the "estimation" method. 3.5. Cost Metric: Loss Rate (lossrate) 3.5.1. Base Identifier The base identifier for this performance metric is "lossrate". 3.5.2. Value Representation The metric value type is a single 'JSONNumber' type value conforming to the number specification of [RFC8259] Section 6. The number MUST be non-negative. The value represents the percentage of packet losses. 3.5.3. Intended Semantics and Use Intended Semantics: To specify spatial and temporal aggregated packet loss rate from the specified source and the specified destination. The spatial aggregation level is specified in the query context (e.g., PID to PID, or endpoint to endpoint). Use: This metric could be used as a cost metric constraint attribute or as a returned cost metric in the response. Wu, et al. Expires January 13, 2021 [Page 18] Internet-Draft ALTO Performance Cost Metrics July 2020 Example 5: Loss rate value on source-destination endpoint pairs POST /endpointcost/lookup HTTP/1.1 Host: alto.example.com Content-Length: TBA Content-Type: application/alto-endpointcostparams+json Accept: application/alto-endpointcost+json,application/alto-error+json { "cost-type": {"cost-mode" : "numerical", "cost-metric" : "lossrate" }, "endpoints" : { "srcs": [ "ipv4:192.0.2.2" ], "dsts": [ "ipv4:192.0.2.89", "ipv4:198.51.100.34", "ipv6:2000::1:2345:6789:abcd" ] } } HTTP/1.1 200 OK Content-Length: TBA Content-Type: application/alto-endpointcost+json { "meta": { "cost-type": { "cost-mode": "numerical", "cost-metric":"lossrate" } }, "endpoint-cost-map": { "ipv4:192.0.2.2": { "ipv4:192.0.2.89" : 0, "ipv4:198.51.100.34": 0, "ipv6:2000::1:2345:6789:abcd" : 0, } } } 3.5.4. Cost-Context Specification Considerations "nominal": Typically packet loss rate does not have a nominal value, although some networks may specify zero losses. Wu, et al. Expires January 13, 2021 [Page 19] Internet-Draft ALTO Performance Cost Metrics July 2020 "sla": It is RECOMMENDED that the "parameters" field of an "sla" packet loss rate provides a link ("link") to the SLA definition. "import": There can be multiple sources to import packet loss rate. If the import is from [RFC8571] (by using unidirectional link loss), it is RECOMMENDED that "parameters" provides "protocol" as a field and "RFC8571" as the value; see Section 3.1.4 for discussions on summing up link metrics to obtain end-to-end metrics. If the import is from the IPPM framework, it is recommended that "parameters" provides "protocol" as a field and "ippm" as the value; see Section 4 of [I-D.ietf-ippm-initial-registry] for additional fields which can be specified for "ippm" in "parameters". "estimation": The exact estimation method is out of the scope of this document. It is RECOMMENDED that the "parameters" field of an "estimation" packet loss rate metric provides a link ("link") to a description of the "estimation" method. 4. Bandwidth Performance Metrics This section introduces three bandwidth related metrics. Given a specified source to a specified destination, these metrics reflect the volume of traffic that the network can carry from the source to the destination. 4.1. Cost Metric: TCP Throughput (tput) 4.1.1. Base Identifier The base identifier for this performance metric is "tput". 4.1.2. Value Representation The metric value type is a single 'JSONNumber' type value conforming to the number specification of [RFC8259] Section 6. The number MUST be non-negative. The unit is bytes per second. 4.1.3. Intended Semantics and Use Intended Semantics: To give the throughput of a TCP flow from the specified source to the specified destination. The spatial aggregation level is specified in the query context (e.g., PID to PID, or endpoint to endpoint). Use: This metric could be used as a cost metric constraint attribute or as a returned cost metric in the response. Wu, et al. Expires January 13, 2021 [Page 20] Internet-Draft ALTO Performance Cost Metrics July 2020 Example 5: TCP throughput value on source-destination endpoint pairs POST /endpointcost/lookup HTTP/1.1 Host: alto.example.com Content-Length: TBA Content-Type: application/alto-endpointcostparams+json Accept: application/alto-endpointcost+json,application/alto-error+json { "cost-type": {"cost-mode" : "numerical", "cost-metric" : "tput"}, "endpoints" : { "srcs": [ "ipv4:192.0.2.2" ], "dsts": [ "ipv4:192.0.2.89", "ipv4:198.51.100.34", "ipv6:2000::1:2345:6789:abcd" ] } } HTTP/1.1 200 OK Content-Length: TBA Content-Type: application/alto-endpointcost+json { "meta": { "cost type": { "cost-mode": "numerical", "cost-metric":"tput" } } "endpoint-cost-map": { "ipv4:192.0.2.2": { "ipv4:192.0.2.89" : 256000, "ipv4:198.51.100.34": 128000, "ipv6:2000::1:2345:6789:abcd" : 428000, } } 4.1.4. Cost-Context Specification Considerations "nominal": Typically TCP throughput does not have a nominal value. "sla": Typically TCP throughput does not have an SLA value. "import": Typically there is not a routing protocol through which one can import TCP throughput. If the import is from the IPPM framework, Wu, et al. Expires January 13, 2021 [Page 21] Internet-Draft ALTO Performance Cost Metrics July 2020 it is recommended that "parameters" provides "protocol" as a field and "ippm" as the value; see Section 4 of [I-D.ietf-ippm-initial- registry] for additional fields which can be specified for "ippm" in "parameters". "estimation": The exact estimation method is out of the scope of this document. See [Prophet] for a method to estimate TCP throughput. It is RECOMMENDED that the "parameters" field of an "estimation" TCP throughput metric provides a link ("link") to a description of the "estimation" method. 4.2. Cost Metric: Residue Bandwidth (bw-residue) 4.2.1. Base Identifier The base identifier for this performance metric is "bw-residue". 4.2.2. Value Representation The metric value type is a single 'JSONNumber' type value that is non-negative. The unit of measurement is bytes per second. 4.2.3. Intended Semantics and Use Intended Semantics: To specify spatial and temporal residual bandwidth from the specified source and the specified destination. The value is calculated by subtracting tunnel reservations from Maximum Bandwidth (motivated from [RFC7810], Section 4.5). The spatial aggregation unit is specified in the query context (e.g., PID to PID, or endpoint to endpoint). Use: This metric could be used either as a cost metric constraint attribute or as a returned cost metric in the response. Wu, et al. Expires January 13, 2021 [Page 22] Internet-Draft ALTO Performance Cost Metrics July 2020 Example 7: bw-residue value on source-destination endpoint pairs POST/ endpointcost/lookup HTTP/1.1 Host: alto.example.com Content-Length: TBA Content-Type: application/alto-endpointcostparams+json Accept: application/alto-endpointcost+json,application/alto-error+json { "cost-type": { "cost-mode": "numerical", "cost-metric": "bw-residue"}, "endpoints": { "srcs": [ "ipv4 : 192.0.2.2" ], "dsts": [ "ipv4:192.0.2.89", "ipv4:198.51.100.34", "ipv6:2000::1:2345:6789:abcd" ] } } HTTP/1.1 200 OK Content-Length: TBA Content-Type: application/alto-endpointcost+json { "meta": { "cost-type" { "cost-mode": "numerical", "cost-metric": "bw-residue" } }, "endpoint-cost-map" { "ipv4:192.0.2.2" { "ipv4:192.0.2.89" : 0, "ipv4:198.51.100.34": 2000, "ipv6:2000::1:2345:6789:abcd": 5000, } } } 4.2.4. Cost-Context Specification Considerations "nominal": Typically residue bandwidth does not have a nominal value. "sla": Typically residue bandwidth does not have an "sla" value. Wu, et al. Expires January 13, 2021 [Page 23] Internet-Draft ALTO Performance Cost Metrics July 2020 "import": There can be multiple sources to import residue bandwidth. If the import is from [RFC8571] (by using unidirectional residue bandwidth), it is RECOMMENDED that "parameters" provides "protocol" as a field and "RFC8571" as the value. The server should be cognizant of issues when computing end-to-end summary statistics from link statistics. For example, the min of the end-to-end path residue bandwidth is the min of all links on the path. "estimation": The exact estimation method is out of the scope of this document. It is RECOMMENDED that the "parameters" field of an "estimation" residue bandwidth metric provides a link ("link") to a description of the "estimation" method. 4.3. Cost Metric: Maximum Reservable Bandwidth (bw-maxres) 4.3.1. Base Identifier The base identifier for this performance metric is "bw-maxres". 4.3.2. Value Representation The metric value type is a single 'JSONNumber' type value that is non-negative. The unit of measurement is bytes per second. 4.3.3. Intended Semantics and Use Intended Semantics: To specify spatial and temporal maximum reservable bandwidth from the specified source to the specified destination. The value is corresponding to the maximum bandwidth that can be reserved (motivated from [RFC3630] Section 2.5.7). The spatial aggregation unit is specified in the query context (e.g., PID to PID, or endpoint to endpoint). Use: This metric could be used either as a cost metric constraint attribute or as a returned cost metric in the response. Wu, et al. Expires January 13, 2021 [Page 24] Internet-Draft ALTO Performance Cost Metrics July 2020 Example 6: bw-maxres value on source-destination endpoint pairs POST/ endpointcost/lookup HTTP/1.1 Host: alto.example.com Content-Length: TBA Content-Type: application/alto-endpointcostparams+json Accept: application/alto-endpointcost+json,application/alto-error+json { "cost-type" { "cost-mode": "numerical", "cost-metric": "bw-maxres"}, "endpoints": { "srcs": [ "ipv4 : 192.0.2.2" ], "dsts": [ "ipv4:192.0.2.89", "ipv4:198.51.100.34", "ipv6:2000::1:2345:6789:abcd" ] } } HTTP/1.1 200 OK Content-Length: TBA Content-Type: application/alto-endpointcost+json { "meta": { "cost-type": { "cost-mode": "numerical", "cost-metric": "bw-maxres" } }, "endpoint-cost-map": { "ipv4:192.0.2.2" { "ipv4:192.0.2.89" : 0, "ipv4:198.51.100.34": 2000, "ipv6:2000::1:2345:6789:abcd": 5000, } } } 4.3.4. Cost-Context Specification Considerations "nominal": Typically maximum reservable bandwidth does not have a nominal value. "sla": Typically maximum reservable bandwidth does not have an "sla" value. Wu, et al. Expires January 13, 2021 [Page 25] Internet-Draft ALTO Performance Cost Metrics July 2020 "import": There can be multiple sources to import maximum reservable bandwidth. For example, Maximum reservable bandwidth is defined by IS-IS/OSPF TE, and measures the reservable bandwidth between two directly connected IS-IS neighbors or OSPF neighbors; see Section 3.5 of [RFC5305]. If the import is from [RFC8571] (by using unidirectional maximum reservable bandwidth), it is RECOMMENDED that "parameters" provides "protocol" as a field and "RFC8571" as the value. "estimation": The exact estimation method is out of the scope of this document. It is RECOMMENDED that the "parameters" field of an "estimation" maximum reservable bandwidth metric provides a link ("link") to a description of the "estimation" method. 5. Operational Considerations The exact measurement infrastructure, measurement condition and computation algorithms can vary from different networks, and are outside the scope of this document. Both the ALTO server and the ALTO clients, however, need to be cognizant of the operational issues discussed below. Also, the performance metrics specified in this document are similar, in that they may use similar data sources and have similar issues in their calculation. Hence, we specify common issues unless one metric has its unique challenges. 5.1. Source Considerations The addition of the "cost-source" field is to solve a key issue: An ALTO server needs data sources to compute the cost metrics described in this document and an ALTO client needs to know the data sources to better interpret the values. To avoid too fine-grained information, this document introduces "cost-source" to indicate only the high-level type of data sources: "estimation" or "sla", where "estimation" is a type of measurement data source and "sla" is a type that is more based on policy. For estimation, for example, the ALTO server may use log servers or the OAM system as its data source as recommended by [RFC7971]. In particular, the cost metrics defined in this document can be computed using routing systems as the data sources. Wu, et al. Expires January 13, 2021 [Page 26] Internet-Draft ALTO Performance Cost Metrics July 2020 5.2. Metric Timestamp Consideration Despite the introduction of the additional cost-context information, there is not a built-in field to indicate the timestamps of the data used to compute a metric. To indicate this attribute, the ALTO server SHOULD return HTTP "Last-Modified", to indicate the freshness of the data used to compute the performance metrics. If the ALTO client obtains updates through an incremental update mechanism (e.g., RFC editor: Fix the RFC number when available. [ALTO SSE]), the client SHOULD assume that the metric is computed using a snapshot at the time that is approximated by the receiving time. 5.3. Backward Compatibility Considerations One potential issue introduced by the optional "cost-source" field is backward compatibility. Consider that an IRD which defines two cost- types with the same "cost-mode" and "cost-metric", but one with "cost-source" being "estimation" and the other being "sla". Then an ALTO client that is not aware of the extension will not be able to distinguish between these two types. A similar issue can arise even with a single cost-type which has "cost-source" being "sla", but the backward client will ignore this field and consider the metric estimation. To address this issue, the only defined "routingcost" metric can be ONLY "estimation". 5.4. Computation Considerations The metric values exposed by an ALTO server may result from additional processing on measurements from data sources to compute exposed metrics. This may involve data processing tasks such as aggregating the results across multiple systems, removing outliers, and creating additional statistics. There are two challenges on the computation of ALTO performance metrics. 5.4.1. Configuration Parameters Considerations Performance metrics often depend on configuration parameters. For example, the value of packet loss rate depends on the measurement interval and varies over time. To handle this issue, an ALTO server may collect data on time periods covering the previous and current time or only collect data on present time. The ALTO server may further aggregate these data to provide an abstract and unified view that can be more useful to applications. To make the ALTO client better understand how to use these performance data, the ALTO server Wu, et al. Expires January 13, 2021 [Page 27] Internet-Draft ALTO Performance Cost Metrics July 2020 may provide the client with the validity period of the exposed metric values. 5.4.2. Availability Considerations Applications value information relating to bandwidth availability whereas bandwidth related metrics can often be only measured at the link level. This document specifies a set of link-level bandwidth related values that may be exposed as such by an ALTO server. The server may also expose other metrics derived from their aggregation and having different levels of endpoint granularity, e.g., link endpoints or session endpoints. The metric specifications may also expose the utilized aggregation laws. 6. Security Considerations The properties defined in this document present no security considerations beyond those in Section 15 of the base ALTO specification [RFC7285]. However concerns addressed in Sections "15.1 Authenticity and Integrity of ALTO Information", "15.2 Potential Undesirable Guidance from Authenticated ALTO Information" and "15.3 Confidentiality of ALTO Information" remain of utmost importance. Indeed, TE performance is a highly sensitive ISP information, therefore, sharing TE metric values in numerical mode requires full mutual confidence between the entities managing the ALTO Server and Client. Numerical TE performance information will most likely be distributed by ALTO Servers to Clients under strict and formal mutual trust agreements. On the other hand, ALTO Clients must be cognizant on the risks attached to such information that they would have acquired outside formal conditions of mutual trust. 7. IANA Considerations IANA has created and now maintains the "ALTO Cost Metric Registry", listed in Section 14.2, Table 3 of [RFC7285]. This registry is located at . This document requests to add the following entries to "ALTO Cost Metric Registry". Wu, et al. Expires January 13, 2021 [Page 28] Internet-Draft ALTO Performance Cost Metrics July 2020 +-----------------+--------------------+ | Identifier | Intended Semantics | +-----------------+--------------------+ | delay-ow | See Section 3.1 | | delay-rt | See Section 3.2 | | delay-variation | See Section 3.3 | | hopcount | See Section 3.4 | | lossrate | See Section 3.5 | | tput | See Section 4.1 | | bw-residue | See Section 4.2 | | bw-maxres | See Section 4.3 | +------------+--------------------+ Since he This document requests the creation of the "ALTO Cost Source Registry" with the following currently defined values: This document requests the creation of the "ALTO Cost Source Registry" with the following currently defined values: +------------+-----------------------------+ | Identifier | Intended Semantics | +------------+-----------------------------+ | nominal | Values in nominal cases | | sla | Values reflecting service | | | level agreement | | import | Values from a given protocol| | estimation | Values by estimation | +------------+-----------------------------+ 8. Acknowledgments The authors of this document would also like to thank Brian Trammell, Haizhou Du, Kai Gao, Lili Liu, Geng Li, Danny Alex Lachos Perez for the reviews and comments. 9. References 9.1. Normative References [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, DOI 10.17487/ RFC2119, March 1997, . [RFC6390] Clark, A. and B. Claise, "Guidelines for Considering New Performance Metric Development", BCP 170, RFC 6390, DOI 10.17487/RFC6390, October 2011, . Wu, et al. Expires January 13, 2021 [Page 29] Internet-Draft ALTO Performance Cost Metrics July 2020 [RFC7285] Alimi, R., Ed., Penno, R., Ed., Yang, Y., Ed., Kiesel, S., Previdi, S., Roome, W., Shalunov, S., and R. Woundy, "Application-Layer Traffic Optimization (ALTO) Protocol", RFC 7285, DOI 10.17487/RFC7285, September 2014, . [RFC7971] Stiemerling, M., Kiesel, S., Scharf, M., Seidel, H., and S. Previdi, "Application-Layer Traffic Optimization (ALTO) Deployment Considerations", RFC 7971, DOI 10.17487/ RFC7971, October 2016, . [RFC8259] Bray, T., Ed., "The JavaScript Object Notation (JSON) Data Interchange Format", STD 90, RFC 8259, DOI 10.17487/ RFC8259, December 2017, . 9.2. Informative References [Prometheus] Volz, J. and B. Rabenstein, "Prometheus: A Next-Generation Monitoring System", 2015. [Prophet] Gao, K., Zhang, J., and YR. Yang, "Prophet: Fast, Accurate Throughput Prediction with Reactive Flows", ACM/IEEE Transactions on Networking July, 2020. [RFC2679] Almes, G., Kalidindi, S., and M. Zekauskas, "A One-way Delay Metric for IPPM", RFC 2679, DOI 10.17487/RFC2679, September 1999, . [RFC2681] Almes, G., Kalidindi, S., and M. Zekauskas, "A Round-trip Delay Metric for IPPM", RFC 2681, DOI 10.17487/RFC2681, September 1999, . [RFC3393] Demichelis, C. and P. Chimento, "IP Packet Delay Variation Metric for IP Performance Metrics (IPPM)", RFC 3393, DOI 10.17487/RFC3393, November 2002, . [RFC5305] Li, T. and H. Smit, "IS-IS Extensions for Traffic Engineering", RFC 5305, DOI 10.17487/RFC5305, October 2008, . [RFC5357] Hedayat, K., Krzanowski, R., Morton, A., Yum, K., and J. Babiarz, "A Two-Way Active Measurement Protocol (TWAMP)", RFC 5357, DOI 10.17487/RFC5357, October 2008, . Wu, et al. Expires January 13, 2021 [Page 30] Internet-Draft ALTO Performance Cost Metrics July 2020 [RFC6349] Constantine, B., Forget, G., Geib, R., and R. Schrage, "Framework for TCP Throughput Testing", RFC 6349, DOI 10.17487/RFC6349, August 2011, . [RFC7680] Almes, G., Kalidindi, S., Zekauskas, M., and A. Morton, Ed., "A One-Way Loss Metric for IP Performance Metrics (IPPM)", STD 82, RFC 7680, DOI 10.17487/RFC7680, January 2016, . [RFC7810] Previdi, S., Ed., Giacalone, S., Ward, D., Drake, J., and Q. Wu, "IS-IS Traffic Engineering (TE) Metric Extensions", RFC 7810, DOI 10.17487/RFC7810, May 2016, . Authors' Addresses Qin Wu Huawei 101 Software Avenue, Yuhua District Nanjing, Jiangsu 210012 CHINA Email: bill.wu@huawei.com Y. Richard Yang Yale University 51 Prospect St New Haven, CT 06520 USA Email: yry@cs.yale.edu Young Lee Samsung 1700 Alma Drive, Suite 500 Plano, TX 75075 USA Email: young.lee@gmail.com Wu, et al. Expires January 13, 2021 [Page 31] Internet-Draft ALTO Performance Cost Metrics July 2020 Dhruv Dhody Huawei Leela Palace Bangalore, Karnataka 560008 INDIA Email: dhruv.ietf@gmail.com Sabine Randriamasy Nokia Bell Labs Route de Villejust Nozay 91460 FRANCE Email: sabine.randriamasy@nokia-bell-labs.com Luis Miguel Contreras Murillo Telefonica Madrid SPAIN Email: luismiguel.contrerasmurillo@telefonica.com Wu, et al. Expires January 13, 2021 [Page 32]