Network Working Group E. Stephan Internet-Draft France Telecom Intended status: Informational L. Liang Expires:~~April 25,~~September 2,2007 University of Surrey A. Morton AT&T Labs~~October 22, 2006~~March 1, 2007IP Performance Metrics (IPPM) for spatial and multicast~~draft-ietf-ippm-multimetrics-02~~draft-ietf-ippm-multimetrics-03Status of this Memo By submitting this Internet-Draft, each author represents that any applicable patent or other IPR claims of which he or she is aware have been or will be disclosed, and any of which he or she becomes aware will be disclosed, in accordance with Section 6 of BCP 79. Internet-Drafts are working documents of the Internet Engineering Task Force (IETF), its areas, and its working groups. Note that other groups may also distribute working documents as Internet- Drafts. 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." The list of current Internet-Drafts can be accessed at http://www.ietf.org/ietf/1id-abstracts.txt. The list of Internet-Draft Shadow Directories can be accessed at http://www.ietf.org/shadow.html. This Internet-Draft will expire on~~April 25,~~September 2,2007. Copyright Notice Copyright (C) The~~Internet Society (2006).~~IETF Trust (2007).Abstract The IETF IP Performance Metrics (IPPM) working group has standardized metrics for measuring end-to-end performance between 2 points. This memo defines 2 sets of metrics to extend these end-to-end ones. It defines spatial metrics for measuring the performance of segments along a path and metrics for measuring the performance of a group of users in multiparty communications. Table of Contents 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . .~~3~~42. Terminology . . . . . . . . . . . . . . . . . . . . . . . . .~~5~~62.1. Multiparty metric . . . . . . . . . . . . . . . . . . . .~~5~~62.2. Spatial metric . . . . . . . . . . . . . . . . . . . . . .~~5~~62.3. Spatial metric points of interest . . . . . . . . . . . .~~5~~62.4. One-to-group metric . . . . . . . . . . . . . . . . . . .~~5~~62.5. One-to-group metric points of interest . . . . . . . . . .~~5~~62.6. Reference point . . . . . . . . . . . . . . . . . . . . .~~5~~62.7. Vector . . . . . . . . . . . . . . . . . . . . . . . . . .~~6~~72.8. Matrix . . . . . . . . . . . . . . . . . . . . . . . . . .~~6~~73. Motivations for spatial and one-to-group metrics . . . . . . .~~7~~83.1. spatial metrics . . . . . . . . . . . . . . . . . . . . .~~7~~83.2. One-to-group metrics . . . . . . . . . . . . . . . . . . .~~8~~93.3. Discussion on Group-to-one and Group-to-group metrics . .~~9~~104. Spatial metrics definitions . . . . . . . . . . . . . . . . .~~9~~104.1. A Definition for Spatial One-way Delay Vector . . . . . . 10 4.2. A Definition of a sample of One-way Delay of a sub path .~~12~~134.3. A Definition for Spatial One-way Packet Loss Vector . . .~~15~~164.4. A Definition for Spatial One-way Jitter Vector . . . . . .~~16~~174.5. Pure Passive Metrics . . . . . . . . . . . . . . . . . . .~~18~~194.6. Discussion on spatial statistics . . . . . . . . . . . . .~~20~~215. One-to-group metrics definitions . . . . . . . . . . . . . . .~~20~~215.1. A Definition for one-to-group One-way Delay . . . . . . .~~20~~215.2. A Definition for one-to-group One-way Packet Loss . . . .~~21~~225.3. A Definition for one-to-group One-way Jitter . . . . . . .~~21 5.4.~~22 6. One-to-Group Sample Statistics . . . . . . . . . . . . . . . . 24 6.1.Discussion on~~one-to-group~~the Impact of packet loss onstatistics . .26 6.2. General Metric Parameters. . . . . . . .~~23 6. Extension from one-to-one to one-to-many measurement~~. . . . .~~26~~. . . 27 6.3. One-to-Group one-way Delay Statistics . . . . . . . . . . 28 6.4. One-to-Group one-way Loss Statistics . . . . . . . . . . . 31 6.5. One-to-Group one-way Delay Variation Statistics . . . . . 337.~~Open issues~~Measurement Methods: Scaleability and Reporting. . . . . . .33 7.1. Computation methods. . . . . . . . . . . . . . . . . .~~27 8. Security Considerations~~. 34 7.2. Measurement. . . . . . . . . . . . . . . . . . .~~27 9. Acknowledgments~~. . . . 35 7.3. effect of Time and Space Aggregation Order on Group Stats. . . . . . . . . . . . . . . . . . . . . . .~~27 10. IANA~~. . . 35 7.4. effect of Time and Space Aggregation Order on Spatial Stats . . . . . . . . . . . . . . . . . . . . . . . . . . 37 8. Open issues . . . . . . . . . . . . . . . . . . . . . . . . . 37 9. SecurityConsiderations . . . . . . . . . . . . . . . . . . .37 9.1. passive measurement. .~~27 11. References~~. . . . . . . . . . . . . . . . .37 9.2. one-to-group metric. . . . . . . . .~~28 11.1. Normative References~~. . . . . . . . . .37 10. Acknowledgments. . . . . . . . .~~28 11.2. Informative References~~. . . . . . . . . . . . . .37 11. IANA Considerations. . . .~~28 Authors' Addresses~~. . . . . . . . . . . . . . . . .38 12. References. . . . . . .~~29 Intellectual Property and Copyright Statements~~. . . . . . . . . .~~30 1. Introduction The metrics specified in this memo are built on notions introduced and discussed in the IPPM Framework document, RFC 2330~~. . . . . . . . . 42 12.1. Normative References . . . . . . . . . . . . . . . . . . . 42 12.2. Informative References . . . . . . . . . . . . . . . . . . 43 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 43 Intellectual Property and Copyright Statements . . . . . . . . . . 45 1. Introduction The metrics specified in this memo are built on notions introduced and discussed in the IPPM Framework document, RFC 2330[RFC2330]. The reader should be familiar with these documents. This memo makes use of definitions of end-to-end One-way Delay Metrics defined in the RFC 2679 [RFC2679] to define metrics for decomposition of end-to-end one-way delays measurements. This memo makes use of definitions of end-to-end One-way Packet loss Metrics defined in the RFC 2680 [RFC2680] to define metrics for decomposition of end-to-end one-way packet loss measurements. The IPPM WG defined a framework for metric definitions and end-to-end measurements: o A general framework for defining performance metrics, described in the Framework for IP Performance~~Metrics, RFC 2330~~Metrics[RFC2330]; o A One-way Active Measurement Protocol~~Requirements, RFC 3763~~Requirements[RFC3763]; o A One-way Active Measurement Protocol (OWAMP)~~[work in progress];~~[RFC4656];o An IP Performance Metrics Registry~~, RFC 4148~~[RFC4148]; It specified a set of end-to-end metrics, which conform to this framework: o The IPPM Metrics for Measuring~~Connectivity, RFC 2678~~Connectivity[RFC2678]; o The One-way Delay Metric for~~IPPM, RFC 2679~~IPPM[RFC2679]; o The One-way Packet Loss Metric for~~IPPM, RFC 2680~~IPPM[RFC2680]; o The Round-trip Delay Metric for~~IPPM, RFC 2681~~IPPM[RFC2681]; o A Framework for Defining Empirical Bulk Transfer Capacity Metrics~~RFC 3148~~[RFC3148]; o One-way Loss Pattern Sample~~Metrics, RFC 3357~~Metrics[RFC3357]; o IP Packet Delay Variation Metric for~~IPPM, RFC 3393~~IPPM[RFC3393]; o Network performance measurement for periodic~~streams, RFC 3432~~streams[RFC3432]; o Packet Reordering Metric for IPPM~~[Work~~[RFC4737][Workin progress]; Based on these works, this memo defines 2 kinds of multi party metrics. Firstly it defines spatial metrics: o A 'sample', called Type-P-Spatial-One-way-Delay-Vector, will be introduced to divide an end-to-end Type-P-One-way-Delay in a spatial sequence of one-way delays. o A 'sample', called Type-P-Spatial-One-way-Packet-Loss-Vector, will be introduced to divide an end-to-end Type-P-One-way-Packet-Loss in a spatial sequence of packet loss. o Using the Type-P-Spatial-One-way-Delay-Vector metric, a 'sample', called Type-P-Spatial-One-way-Jitter-Vector, will be introduced to divide an end-to-end Type-P-One-way-ipdv in a spatial sequence of jitter. o Using the Type-P-Spatial-One-way-Delay-Vector metric, a 'sample', called Type-P-subpath-One-way-Delay-Stream, will be introduced to define the one-way-delay between a pair of host of the path. This metric is similar to Type-P-One-way-Delay-Stream. o Using Type-P-subpath-One-way-Delay-Stream, a 'sample' Type-P- Passive-One-way-Delay-Stream will be introduced to define passive metrics. These metrics are designed for pure passive measurement methodology as introduced by PSAMP WG. Then it defines one-to-group metrics. o Using one test packet sent from one sender to a group of receivers, a 'sample', called Type-P-one-to-group-One-way-Delay- Vector, will be introduced to define the list of Type-P-one-way- delay between this sender and the group of receivers. o Using one test packet sent from one sender to a group of receivers, a 'sample', called Type-P-one-to-group-One-way-Packet- Loss-Vector, will be introduced to define the list of Type-P-One- way-Packet-Loss between this sender and the group of receivers o Using one test packet sent from one sender to a group of receivers, a 'sample', called Type-P-one-to-group-One-way-Jitter- Vector, will be introduced to define the list of Type-P-One-way- ipdv between this sender and the group of receivers o Then a discussion section presents the set of statistics that may be computed on the top of these metrics to present the QoS in a view of a group of users as well as the requirements of relative QoS on multiparty communications. 2. Terminology 2.1. Multiparty metric A metric is said to be multiparty if the definition involved more than two sources or destinations in the measurements. All multiparty metrics define a set of hosts called "points of interest", where one host is the source and other hosts are the measurement collection points. For example, if the set of points of interest is < ha, hb, hc, ..., hn >, where ha is the source and < hb, hc, ..., hn > are the destinations, then measurements may be conducted between < ha, hb>, < ha, hc>, ..., <ha, hn >. 2.2. Spatial metric A metric is said to be spatial if one of the hosts involved is neither the source nor the destination of the metered packet. 2.3. Spatial metric points of interest Points of interest of a spatial metric are the routers or sibling in the path between source and destination (in addition to the source and the destination~~themself).~~themselves).2.4. One-to-group metric A metric is said to be one-to-group if the measured packet is sent by one source and (potentially) received by several destinations. Thus, the topology of the communication group can be viewed as a centre- distributed or server-client topology with the source as the centre/ server in the topology. 2.5. One-to-group metric points of interest Points of interest of One-to-group metrics are the set of host destinations receiving packets from the source (in addition to the source itself). 2.6. Reference point The centre/server in the one-to-group measurement that is controlled by network operators can be a very good reference point where measurement data can be collected for further processing although the actual measurements have to be carried out at all points of interest. I.e., the measurement points will be all clients/receivers while the reference point acts as source for the one-to-group metric. Thus, we can define the reference point as the host while the statistic calculation will be carried out. 2.7. Vector A group of singletons is the set of results of the observation of the behaviour of the same packet at different places of a network. A Vector is a set of singletons, which are a set of results of the observation of the behaviour of the same packet at different places of a network at different time. For instance, if One-way delay singletons~~abserved~~observedat N receivers for Packet P sent by the source Src are dT1, dT2,..., dTN, it can be say that a vector V with N elements can be~~orgnized~~organizedas {dT1, dT2,..., dTN}. The elements in one vector are singletons distinct with each other in terms of both measurement point and time. Given the vector V as an example, the element dT1 is distinct from the rest by measured at receiver 1 at time T1. Additional to a singleton, Vector gives information over a space dimension. 2.8. Matrix Several vectors can~~orgnize~~organizeform up a Matrix, which contains results observed in a sampling interval at different place of a network at different time. For instance, given One-way delay vectors V1={dT11, dT12,..., dT1N}, V2={dT21, dT22,..., dT2N},..., Vm={dTm1, dTm2,..., dTmN} for Packet P1, P2,...,Pm, we can have a One-way delay Matrix {V1, V2,...,Vm}. Additional to the information given by a Vector, a Matrix is more powerful to present network performance in both space and time dimensions. It normally~~corresponding~~correspondsto a sample. The relation among Singleton, Vector and Matrix can be shown in the following~~Fig~~Figure1.~~one to group~~one-to-groupSingleton / Sample Src~~Rcvr~~Recv..............................~~..................R1dT1~~.................... 1 R1dT1R1dT2 R1dT3 R1dT4 `:=-.._ T `._ ``-..__ `.~~`-~~`-... 2R2dT1 R2dT2 R2dT3 R2dT4 `-. `-.~~`._R3dT1~~`.... N R3dT1R3dT2 R3dT3 R3dT4 Vector Matrix (space) (time) Figure~~1.~~1: Relation beween Singletons, vectors and matrix3. Motivations for spatial and one-to-group metrics All IPPM metrics are defined for end-to-end measurement. These metrics provide very good guides for measurement in the pair communications. However, further efforts should be put to define metrics for multiparty measurements such as one to one trajectory metrics and one to multipoint metrics. 3.1. spatial metrics Decomposition of instantaneous end-to-end measures is needed: o~~The PCE WG is extending existing protocols to permit remote path computation and path computation quality, including inter domain. One may say that in intra domain the decomposing~~Decomposingthe performance of~~a~~interdomainpath is~~not whished. However such decomposition is~~desirable in interdomain to qualify~~each~~perAS~~computation with~~contribution tothe~~initial request.~~performance.So it is necessary to define standard spatial metrics before going further in the computation of inter domain path with QoS constraint. o Traffic engineering and troubleshooting applications require spatial views of the one-way delay consumption, identification of the location of the lost of packets and the decomposition of the jitter over the path. o Monitoring the QoS of a multicast tree, of MPLS point-to- multipoint and inter-domain communication require spatial decomposition of the one-way delay, of the packet loss and of the jitter. o Composition of metrics is a need to scale in the measurement plane.~~The definition of composition metrics is a work in progress [I-D.ietf-ippm-spatial-composition]; .~~Spatial measure give typically the individual performance of an intra domain segment. It is the elementary piece of information to exchange for measuring interdomain performance based on composition of metrics. o The PSAMP WG defines capabilities to sample packets in a way to to support~~measurement. [I-D.boschi-ipfix-reducing-redundancy]; defines a method to collect packets information to measure the~~instantaneous~~spatial performance without injecting test traffic.~~measurement respecful of the IPPM framework [RFC2330].Consequently it is~~urgent~~necessaryto define a set of~~common~~spatial metrics for passive and active~~techniques which respect the IPPM framework [RFC2330]. This need is emphases by the fact that end- to-end spatial measurement involves the 2 techniques;~~techniques.3.2. One-to-group metrics While the node-to-node based spatial measures can provide very useful data in the view of each connection, we also need measures to present the performance of a multiparty communication in the view of a group with consideration that it involves a group of people rather than two. As a consequence a simple one-way metric cannot describe the multi-connection situation. We need some new metrics to collect performance of all the connections for further statistics analysis. A group of metrics are proposed in this stage named one-to-group performance metrics based on the unicast metrics defined in IPPM WG. One-to-group metrics are trying to composite one-way metrics from one source to a group of destinations to make up new metrics. The compositions are necessary for judging the network performance of multiparty communications and can also be used to describe the difference of the QoS served among a group of users. One-to-group performance metrics are needed for several reasons: o For designing and engineering multicast trees and MPLS point-to- multipoint LSP; o For evaluating and controlling of the quality of the multicast services; o For controlling the performance of the inter domain multicast services; o For presenting and evaluating the relative QoS requirements for the multiparty communications. To understand the connection situation between one source and any one receiver in the multiparty communication group, we need the collection of instantaneous end-to-end measures. It will give us very detailed insight into each branch of the multicast tree in terms of end-to-end absolute QoS. It can provide clear and helpful information for engineers to identify the connection with problems in a complex multiparty routing tree.~~3.3. Discussion on Group-to-one and Group-to-group~~The one-to-groupmetrics~~We note that points of interest can also be selected~~described in this memo introduce one-to-many concerns to the IPPM working group to measure the performance of a group of users who receiving data from the same source. The concept extends the "path" in the one-way measurement to "path tree" to cover both one-to-one and one-to-many communications. Nevertheless, applied to one-to-one communications they provide exactly the same results as the corresponding one-to-one metrics. 3.3. Discussion on Group-to-one and Group-to-group metrics We note that points of interest can also be selectedto define measurements on Group-to-one and Group-to-group topologies. These topologies are currently beyond the scope of this memo, because they would involve multiple packets launched from different sources. However, we can give some clues here on these two cases. The measurements for group-to-one topology can be easily derived from the one-to-group measurement. The measurement point is the reference point that is acting as a receiver while all of clients/receivers defined for one-to-group measurement act as sources in this case. For the group-to-group connection topology, we can hardly define the reference point and, therefore, have difficulty to define the measurement points. However, we can always avoid this confusion by treating the connections as one-to-group or group-to-one in our measurements without consideration on how the real communication will be carried out. For example, if one group of hosts < ha, hb, hc, ..., hn > are acting as sources to send data to another group of hosts < Ha, Hb, Hc, ..., Hm >, we can always decompose them into n one-to-group communications as < ha, Ha, Hb, Hc, ..., Hm >, < hb, Ha, Hb, Hc, ..., Hm >, <hc, Ha, Hb, Hc, ..., Hm >, ..., < hn, Ha, Hb, Hc, ..., Hm >. 4. Spatial metrics definitions Spatial decomposition metrics are based on standard end-to-end metrics. The definition of a spatial metric is coupled with the corresponding end-to-end metric. The~~methodoly~~methodologyis based on the measure of the same test packet and parameters of the corresponding end-to-end metric. 4.1. A Definition for Spatial One-way Delay Vector This section is coupled with the definition of Type-P-One-way-Delay. When a parameter from section 3 of [RFC2679] is first used in this section, it will be tagged with a trailing asterisk. Sections 3.5 to 3.8 of [RFC2679] give requirements and applicability statements for end-to-end one-way-delay measurements. They are applicable to each point of interest Hi involved in the measure. Spatial one-way-delay measurement SHOULD be respectful of them, especially those related to methodology, clock, uncertainties and reporting. Following we adapt some of them and introduce points specific to spatial measurement. 4.1.1. Metric Name Type-P-Spatial-One-way-Delay-Vector 4.1.2. Metric Parameters + Src*, the IP address of the sender. + Dst*, the IP address of the receiver. + i, An integer which ordered the hosts in the path. + Hi, exchange points of the path digest. + T*, a time, the sending (or initial observation) time for a measured packet. + dT* a delay, the one-way delay for a measured packet. + dT1,..., dTn a list of delay. + P*, the specification of the packet type. + <Src, H1, H2,..., Hn, Dst>, a path digest. 4.1.3. Metric Units A sequence of times. 4.1.4. Definition Given a Type-P packet sent by the sender Src at wire-time (first bit) T to the receiver Dst in the path <H1, H2,..., Hn>. Given the sequence of values <T+dT1,T+dT2,...,T+dTn,T+dT> such that dT is the Type-P-One-way-Delay from Src to Dst and such that for each Hi of the path, T+dTi is either a real number corresponding to the wire-time the packet passes (last bit received) Hi, or undefined if the packet never passes Hi. Type-P-Spatial-One-way-Delay-Vector metric is defined for the path <Src, H1, H2,..., Hn, Dst> as the sequence of values <T,dT1,dT2,...,dTn,dT>. 4.1.5. Discussion Following are specific issues which may occur: o the delay looks to decrease: dTi > DTi+1. this seem typically du to some clock synchronisation issue. this point is discussed in the section 3.7.1. "Errors or uncertainties related to Clocks" of of [RFC2679]; o The location of the point of interest in the device influences the~~result (see [I-D.quittek-ipfix-middlebox]).~~result.If the packet is not observed on the input interface the delay includes buffering time and consequently an uncertainty due to the difference between 'wire time' and 'host time'; 4.1.6. Interference with other test packet To avoid packet collision it is preferable to include a sequence number in the packet. 4.1.7. loss threshold To determine if a dTi is defined or undefined it is necessary to define a period of time after which a packet is considered loss. 4.1.8. Methodologies Section 3.6 of [RFC2679] gives methodologies for end-to-end one-way- delay measurements. Most of them apply to each points interest Hi and are relevant to this section. Generally, for a given Type-P, in a given Hi, the methodology would proceed as follows: o At each Hi, prepare to capture the packet sent a time T, take a timestamp Ti', determine the internal delay correction dTi', extract the timestamp T from the packet, then compute the one-way- delay from Src to Hi: dTi = Ti' - dTi' - T. The one-way delay is undefined (infinite) if the packet is not detected after the 'loss threshold' duration; o Gather the set of dTi of each Hi and order them according to the path to build the Type-P-Spatial-One-way-Delay-Vector metric <T,dT1,dT2,...,dTn,dT> over the path <H1, H2,..., Hn>. It is out of the scope of this document to define how each Hi detects the packet. 4.1.9. Reporting the metric Section 3.6 of [RFC2679] indicates the items to report. 4.1.10. Path It is clear that a end-to-end Type-P-One-way-Delay can't determine the list of hosts the packet passes~~throught.~~through.Section 3.8.4 of [RFC2679] says that the path traversed by the packet SHOULD be reported but is practically impossible to determine. This part of the job is provide by Type-P-Spatial-One-way-Delay- Vector metric because each points of interest Hi which capture the packet is part of the path. 4.2. A Definition of a sample of One-way Delay of a sub path This metric is similar to the metric Type-P-One-way-Delay-Poisson- stream defined in [RFC2679] and to the metric Type-P-One-way-Delay- Periodic-Stream defined in [RFC3432]. Nevertheless its definition differs because it is based of the division of end-to-end One-way delay using the metric Type-P-Spatial- One-way-Delay-Vector defined above. It aims is to define a sample of One-way-Delay between a pair of hosts of a path usable by active and passive measurements. Sections 3.5 to 3.8 of [RFC2679] give requirements and applicability statements for end-to-end one-way-delay measurements. They are applicable to each point of interest Hi involved in the measure. Subpath one-way-delay measurement SHOULD be respectful of them, especially those related to methodology, clock, uncertainties and reporting. 4.2.1. Metric Name Type-P-subpath-One-way-Delay-Stream 4.2.2. Metric Parameters + Src*, the IP address of the sender. + Dst*, the IP address of the receiver. + i, An integer which orders exchange points in the path. + k, An integer which orders the packets sent. + <Src, H1, H2,..., Hn, Dst>, a path digest. + Ha, a host of the path digest different from Dst and Hb; + Hb, a host of the path digest different from Src and Ha. Hb order in the path must greater that Ha; + Hi, exchange points of the path digest. + dT1,..., dTn a list of delay. + P*, the specification of the packet type. 4.2.3. Metric Units A sequence of pairs <Tk,dt>. T is one of time of the sequence T1...Tn; dt is a delay. 4.2.4. Definition Given 2 hosts Ha and Hb of the path <Src, H1, H2,..., Hn, Dst>, given a flow of packets of Type-P sent from Src to Dst at the times T1, T2... Tn. At each of these times, we obtain a Type-P-Spatial-One- way-Delay-Vector <T1,dT1.1, dT1.2,..., dT1.n,dT1>. We define the value of the sample Type-P-subpath-One-way-Delay-Stream as the sequence made up of the couples <Tk,dTk.b - dTk.a>. dTk.a is the delay between Src and Ha. dTk.b is the delay between Src and Hb. 'dTk.b - dTk.a' is the one-way delay experienced by the packet sent at the time Tk by Src when going from Ha to Hb. 4.2.5. Discussion Following are specific issues which may occur: o When a is Src <Tk,dTk.b - dTk.a> is the measure of the first hop. o When b is Dst <Tk,dTk.b - dTk.a> is the measure of the last hop. o the delay looks to decrease: dTi > DTi+1: * This is typically du to clock synchronisation issue. this point is discussed in the section 3.7.1. "Errors or uncertainties related to Clocks" of of [RFC2679]; * This may occurs too when the clock resolution of one probe is bigger than the~~minimun~~minimumdelay of a path. As an example this happen when measuring the delay of a path which is 500 km long with one probe synchronized using NTP having a clock resolution of 8ms. o The location of the point of interest in the device influences the~~result (see [I-D.quittek-ipfix-middlebox]).~~result.If the packet is not observed on the input interface the delay includes buffering time and consequently an uncertainty due to the difference between 'wire time' and 'host time'; o dTk.b may be observed and not dTk.a. o Tk is unknown if the flow is made of end user packets, that is pure passive measure. In this case Tk may be forced to Tk+dTk.a. This motivate separate metrics names for pure passive measurement or specific reporting information. o Pure passive measure should consider packets of the same size and of the same Type-P. 4.2.6. Interference with other packet 4.2.7. loss threshold To determine if a dTi is defined or undefined it is necessary to define a period of time after which a packet is considered loss. 4.2.8. Methodologies Both active and passive method should discussed. 4.2.9. Reporting the metric Section 3.6 of [RFC2679] indicates the items to report. 4.2.10. Path 4.3. A Definition for Spatial One-way Packet Loss Vector This section is coupled with the definition of Type-P-One-way-Packet- Loss. Then when a parameter from the section 2 of [RFC2680] is first used in this section, it will be tagged with a trailing asterisk. Sections 2.5 to 2.8 of [RFC2680] give requirements and applicability statements for end-to-end one-way-Packet-Loss measurements. They are applicable to each point of interest Hi involved in the measure. Spatial packet loss measurement SHOULD be respectful of them, especially those related to methodology, clock,~~uncertainities~~uncertaintiesand reporting. Following we define the spatial metric, then we adapt some of the points above and introduce points specific to spatial measurement. 4.3.1. Metric Name Type-P-Spatial-One-way-Packet-Loss-Vector 4.3.2. Metric Parameters + Src*, the IP address of the sender. + Dst*, the IP address of the receiver. + i, An integer which ordered the hosts in the path. + Hi, exchange points of the path digest. + T*, a time, the sending (or initial observation) time for a measured packet. + dT1,..., dTn, dT, a list of delay. + P*, the specification of the packet type. + <Src, H1, H2,..., Hn, Dst>, a path digest. + B1, B2, ..., Bi, ..., Bn, a list of~~boolean~~Booleanvalues. 4.3.3. Metric Units A sequence of~~boolean~~Booleanvalues. 4.3.4. Definition Given a Type-P packet sent by the sender Src at time T to the receiver Dst in the path <H1, H2, ..., Hn>. Given the sequence of times <T+dT1,T+dT2,...,T+dTn,T+dT> the packet passes <H1, H2 ..., Hn, Dst>, Type-P-One-way-Packet-Lost-Vector metric is defined as the sequence of values <B1, B2, ..., Bn> such that for each Hi of the path, a value of Bi of 0 means that dTi is a finite value, and a value of 1 means that dTi is undefined. 4.3.5. Discussion Following are specific issues~~wich~~whichmay occur: o the result includes the sequence 1,0. This case means that the packet was seen by a host but not by it successor on the path; o The location of the meter in the device influences the result: o Even if the packet is received by a device, it may be not observed by a meter located after a buffer; 4.3.6. Reporting Section in progress. 4.4. A Definition for Spatial One-way Jitter Vector This section uses parameters from the definition of Type-P-One-way- ipdv. When a parameter from section 2 of [RFC3393] is first used in this section, it will be tagged with a trailing asterisk. Sections 3.5 to 3.7 of [RFC3393] give requirements and applicability statements for end-to-end one-way-ipdv measurements. They are applicable to each point of interest Hi involved in the measure. Spatial one-way-ipdv measurement SHOULD be respectful of them, especially those related to methodology, clock,~~uncertainities~~uncertaintiesand reporting. Following we adapt some of them and introduce points specific to spatial measurement. 4.4.1. Metric Name Type-P-Spatial-One-way-Jitter-Vector 4.4.2. Metric Parameters + Src*, the IP address of the sender. + Dst*, the IP address of the receiver. + i, An integer which ordered the hosts in the path. + Hi, exchange points of the path digest. + T1*, the time the first packet was sent. + T2*, the time the second packet was sent. + P, the specification of the packet type. + P1, the first packet sent at time T1. + P2, the second packet sent at time T2. + <Src, H1, H2,..., Hn, Dst>, a path digest. + <T1,dT1.1, dT1.2,..., dT1.n,dT1>, the Type-P-Spatial-One-way-Delay-Vector for packet sent at time T1; + <T2,dT2.1, dT2.2,..., dT2.n,dT2>, the Type-P-Spatial-One-way-Delay-Vector for packet sent at time T2; + L*, a packet length in bits. The packets of a Type P packet stream from which the Type-P-Spatial-One-way-Delay-Vector metric is taken MUST all be of the same length. 4.4.3. Metric Units A sequence of times. 4.4.4. Definition Given the Type-P packet having the size L and sent by the sender Src at wire-time (first bit) T1 to the receiver Dst in the path <H1, H2,..., Hn>. Given the Type-P packet having the size L and sent by the sender Src at wire-time (first bit) T2 to the receiver Dst in the same path. Given the Type-P-Spatial-One-way-Delay-Vector <T1,dT1.1, dT1.2,..., dT1,n,dT1> of the packet P1. Given the Type-P-Spatial-One-way-Delay-Vector <T2,dT2.1, dT2.2,..., dT2,n,dT2> of the packet P2. Type-P-Spatial-One-way-Jitter-Vector metric is defined as the sequence of values <T2-T1,dT2.1-dT1.1,dT2.2-dT1.2,...,dT2.n- dT1.n,dT2-dT1> Such that for each Hi of the path <H1, H2,..., Hn>, dT2.i-dT1.i is either a real number if the packets P1 and P2 passes Hi at wire-time (last bit) dT1.i, respectively dT2.i, or undefined if at least one of them never passes Hi. T2-T1 is the inter-packet emission interval and dT2-dT1 is ddT* the Type-P-One-way-ipdv at T1,T2*. 4.4.5. Sections in progress See sections 3.5 to 3.7 of [RFC3393]. 4.5. Pure Passive Metrics Spatial metrics may be measured without injecting test~~traffic as described in [I-D.boschi-ipfix-reducing-redundancy] .~~traffic.4.5.1. Discussion on Passive measurement One might says that most of the operational issues occur in the last mile and that consequently such measure are less useful than active~~measuremeent.~~measurement.Nevertheless they are usable for network TE and interdomain QoS monitoring, and composition of metric. Such a technique have some limitations that are discussed below. 4.5.1.1. Passive One way delay As the packet is not a test packet, it does not include the time it was sent. Consequently a point of interest Hi ignores the time the packet was send. So It is not possible to measure the delay between Src and Hi in the same manner it is not possible to measure the delay betwwen Hi and Dst. 4.5.1.2. Passive Packet loss The packet is not a test packet, so it does not include a sequence number. Packet lost measurement doe not require time synchronization and require only one point of observation. Nevertheless it requires the point of interest Hi to be expecting the packet. Practically Hi may not detect a lost of packet that occurs between Src and Hi. A point of interest Hi ignores the time the packet is send because the packet does not carry the time it was injected in the network. So a probe Hi can not compute dTi. An alternative to these issues consist in considering sample spatial One-way delay that T is the time when H1 (the first passive probe of the path) observed the packet. 4.5.2. Reporting and composition To avoid misunderstanding and to address specific reporting constraint a proposal consists in defining distinct metrics for pure passive measurement based on the definition above. It is crucial to know the~~methodologie~~methodologiesused because of the difference of method of detection (expecting Seq++); because of the difference of source of time (H1 vs Src) and because of the difference of~~behavior~~behaviourof the source (Poisson/unknown). 4.5.3. naming and registry Having distinct metrics identifiers for spatial metrics and passive spatial metrics in the [RFC4148] will avoid~~interoperabily~~interoperabilityissues especially during composition of metrics. 4.5.4. Passive One way delay metrics 4.5.5. Passive One way PacketLoss metrics 4.5.6. Passive One way jitter metrics 4.6. Discussion on spatial statistics Do we define min, max, avg of spatial metrics ? having the maximum loss metric value could be interesting. Say, the segment between router A and B always contributes loss metric value of "1" means it could be the potential problem segment. Uploading dTi of each Hi consume a lot of bandwidth. Computing statistics (min, max and avg) of dTi locally in each Hi reduce the bandwidth consumption. 5. One-to-group metrics definitions 5.1. A Definition for one-to-group One-way Delay 5.1.1. Metric Name Type-P-one-to-group-One-way-Delay-Vector 5.1.2. Metric Parameters o Src, the IP address of a host acting as the source. o Recv1,..., RecvN, the IP addresses of the N hosts acting as receivers. o T, a time. o dT1,...,dTn a list of time. o P, the specification of the packet type. o Gr, the multicast group address (optional). The parameter Gr is the multicast group address if the measured packets are transmitted by multicast. This parameter is to identify the measured traffic from other unicast and multicast traffic. It is set to be optional in the metric to avoid losing any generality, i.e. to make the metric also applicable to unicast measurement where there is only one receivers. 5.1.3. Metric Units The value of a Type-P-one-to-group-One-way-Delay-Vector is a set of singletons metrics Type-P-One-way-Delay [RFC2679]. 5.1.4. Definition Given a Type P packet sent by the source Src at Time T, given the N hosts { Recv1,...,RecvN } which receive the packet at the time { T+dT1,...,T+dTn }, a Type-P-one-to-group-One-way-Delay-Vector is defined as the set of the Type-P-One-way-Delay singleton between Src and each receiver with value of { dT1, dT2,...,dTn }. 5.2. A Definition for one-to-group One-way Packet Loss 5.2.1. Metric Name Type-P-one-to-group-One-way-Packet-Loss-Vector 5.2.2. Metric Parameters o Src, the IP address of a host acting as the source. o Recv1,..., RecvN, the IP addresses of the N hosts acting as receivers. o T, a time. o T1,...,Tn a list of time. o P, the specification of the packet type. o Gr, the multicast group address (optional). 5.2.3. Metric Units The value of a Type-P-one-to-group-One-way-Packet-Loss-Vector is a set of singletons metrics Type-P-One-way-Packet-Loss [RFC2680]. 5.2.4. Definition Given a Type P packet sent by the source Src at T and the N hosts, Recv1,...,RecvN, which should receive the packet at T1,...,Tn, a Type-P-one-to-group-One-way-Packet-Loss-Vector is defined as a set of the Type-P-One-way-Packet-Loss singleton between Src and each of the receivers {<T1,0|1>,<T2,0|1>,..., <Tn,0|1>}. 5.3. A Definition for one-to-group One-way Jitter 5.3.1. Metric Name Type-P-one-to-group-One-way-Jitter-Vector 5.3.2. Metric Parameters + Src, the IP address of a host acting as the source. + Recv1,..., RecvN, the IP addresses of the N hosts acting as receivers. + T1, a time. + T2, a time. + ddT1,...,ddTn, a list of time. + P, the specification of the packet type. + F, a selection function defining unambiguously the two packets from the stream selected for the metric. + Gr, the multicast group address (optional) 5.3.3. Metric Units The value of a Type-P-one-to-group-One-way-Jitter-Vector is a set of singletons metrics Type-P-One-way-ipdv [RFC3393]. 5.3.4. Definition Given a Type P packet stream, Type-P-one-to-group-One-way-Jitter- Vector is defined for two packets from the source Src to the N hosts {Recv1,...,RecvN },which are selected by the selection function F, as the difference between the value of the Type-P-one-to-group-One-way- Delay-Vector from Src to { Recv1,..., RecvN } at time T1 and the value of the Type-P-one-to-group- One-way-Delay-Vector from Src to { Recv1,...,RecvN } at time T2. T1 is the wire-time at which~~Scr~~Srcsent the first bit of the first packet, and T2 is the wire-time at which Src sent the first bit of the second packet. This metric is derived from the Type-P-one-to- group-One-way-Delay-Vector metric. Therefore, for a set of real number {ddT1,...,ddTn},Type-P-one- to- group-One-way-Jitter-Vector from Src to { Recv1,...,RecvN } at T1, T2 is {ddT1,...,ddTn} means that Src sent two packets, the first at wire-time T1 (first bit), and the second at wire-time T2 (first bit) and the packets were received by { Recv1,...,RecvN } at wire-time {dT1+T1,...,dTn+T1}(last bit of the first packet), and at wire-time {dT'1+T2,...,dT'n+T2} (last bit of the second packet), and that {dT'1-dT1,...,dT'n-dTn} ={ddT1,...,ddTn}.~~5.4. Discussion on one-to-group statistics~~6. One-to-Group Sample StatisticsThe defined one-to-group metrics above can all be directly achieved from the relevant unicast one-way metrics. They managed to collect all unicast measurement results of one-way metrics together in one profile and sort them by receivers and packets in a multicast group. They can provide sufficient information regarding the network performance in terms of each receiver and guide engineers to identify potential problem happened on each branch of a multicast routing tree. However, these metrics can not be directly used to conveniently present the performance in terms of a group and neither to identify the relative performance situation.~~One may say that no matter how many people join the communication, the connections can still be treated as a set of one-to-one connection. However, we might not describe a multiparty communication by a set of one-way measurement metrics because of the difficulty for understanding and the lack of convenience. For instance, an engineer might not describe the connections of a multiparty online conference in terms of one-to-group one-way delay for user A and B, B and C, and C and A because people might be confused. If there are more users in the same communication, the description might be very long. And he might use the one-way metrics with worst and the best value to give users an idea of the performance range of the service they are providing. But it is not clear enough and might not be accurate in a large multiparty communication scenario.~~From the performance point of view, the multiparty communication services not only require the absolute performance support but also the relative performance. The relative performance means the difference between absolute performance of all users. Directly using the one-way metrics cannot present the relative performance situation. However, if we use the variations of all users one-way parameters, we can have new metrics to measure the difference of the absolute performance and hence provide the threshold value of relative performance that a multiparty service might demand. A very good example of the high relative performance requirement is the online gaming. A very light~~worse~~difference indelay~~will~~mightresult in failure in the game. We have to use~~the new~~multicast specificstatistic metrics to define exactly how small the relative delay the online gaming requires. There are many other services, e.g. online biding, online stock market, etc.,~~need a rule~~that require multicast metrics in orderto~~judge~~evaluatethe~~relative performance requirement.~~network against their requirements.Therefore, we can see the importance of~~new~~new, multicast specific,statistic metrics to feed this need. We mightalsouse some one-to-group statistic conceptions to present and report the group performance and relative performance to save the report transmission bandwidth. Statistics have been defined for One- way metrics in corresponding FRCs. They provide the foundation of definition for performance statistics. For instance, there are definitions for minimum and maximum One-way delay in~~[RFC2679] and One-way delay mean in [I-D.ietf-ippm-spatial-composition].~~[RFC2679].However, there is a dramatic difference between the statistics for one-to-one communications and for one-to-many communications. The former one only has statistics over the time dimension while the later one can have statistics over both time~~dimension~~and space~~dimention.~~dimensions.This space dimension is introduced by the Matrix~~concept. For a~~concept as illustrated in Figure 7. For aMatrix M~~shown in the Fig. 2,~~each row is a set of One-way singletons spreading over the~~space~~timedimension and each~~colume~~columnis another set of One-way singletons spreading over the~~time~~spacedimension.~~(preamble)~~Receivers Space ^ 1 |/R1dT1 R1dT2 R1dT3 ... R3dTk\ |~~dT11, dT12,..., dT1N~~| |~~dT21, dT22,..., dT2N~~2| |~~:~~R2dT1 R2dT2 R2dT3 ... R3dTk| |~~:~~| |~~dTm1, dTm2,..., dTmN~~3 | | R3dT1 R3dT2 R3dT3 ... R3dTk | . | | | . | | | . | | | n| \RndT1 RndT2 RndT3 ... RndTk/~~Fig. 2~~+--------------------------------------------> time T0 Figure 7:Matrix M~~(m*N)~~(n*m)In Matrix M, each element is a One-way delay singleton. Each~~row~~columnis a delay vector contains the One-way delays of the same packet observed at~~N~~Mpoints of interest. It implies the geographical factor of the performance within a group. Each~~colume~~rowis a set of One-way delays observed during a sampling interval at one of the points of interest. It presents the delay performance at a receiver over the time dimension. Therefore, one can either calculate statistics by rows over the space dimension or by~~columes~~columnsover the time dimension. It's up to the operators or service provides which dimension they are interested in. For example, a TV broadcast service provider might want to know the statistical performance of each user in a long term run to make sure their services are acceptable and stable. While for an online gaming service provider, he might be more interested to know if all users are served~~farely~~fairlyby calculating the statistics over the space dimension. This memo does not~~intent~~intendto recommend which of the statistics are better than the other. To save the report transmission bandwidth, each point of interest can send statistics in a pre-defined time interval to the reference point rather than sending every One-way singleton it observed. As long as an appropriate time interval is decided, appropriate~~stantistics~~statisticscan represent the performance in a certain accurate scale. How to decide the time interval and how to bootstrap all points of interest and the reference point depend on applications. For instance, applications with lower transmission rate can have the time interval longer and ones with higher transmission rate can have the time interval shorter. However, this is out of the scope of this memo. Moreover, after knowing the statistics over the time dimension, one might want to know how this statistics distributed over the space dimension. For instance, a TV broadcast service provider had the performance Matrix M and calculated the One-way delay mean over the time dimension to obtain a delay Vector as {V1,V2,..., VN}. He then calculated the mean of all the elements in the Vector to see what level of delay he has served to all N users. This new delay mean gives information on how good the service has been delivered to a group of users during a sampling interval in terms of delay. It needs twice calculation to have this statistic over both time and space dimensions. We name this kind of statistics 2-level statistics to distinct with those 1-level statistics calculated over either space or time dimension. It can be easily prove that no matter over which dimension a 2-level statistic is calculated first, the results are the same. I.e. one can calculate the 2-level delay mean using the Matrix M by having the 1-level delay mean over the time dimension first and then calculate the mean of the obtained vector to find out the 2-level delay mean. Or, he can do the 1-level statistic calculation over the space~~dimention~~dimensionfirst and then have the 2-level delay mean. Both two results will be exactly the same. Therefore, when define a 2-level statistic,~~it~~thereis no need to specify in which procedure the calculation should follow.~~There are many~~Comment: The above statement depends on whether the order of operations has any affect on the outcome. Manystatistics can be defined for the proposed~~one-to- group~~one-to-groupmetrics over either the space dimension or the time dimension or both.~~In this memo, we define one-to-group mean and one-to-group variation over the space dimension. These statistics are offered mostly to be illustrative of what could be done. One-to-group mean are trying to measure~~This memo treatsthe~~overall performance for a multicast group associated to one source. It is~~case wherea~~reflection~~streamofpackets fromthe~~absolute performance of~~Source results ina~~multiparty communication service when we treat all receivers as one customer. It can also present the trend of the absolute performance of all receivers, i.e., it shows that most~~sample at eachof the~~receivers~~Receiversin the~~multiparty communication service trend to receive an absolute performance close to~~Group, and these samples are each summarized withthe~~mean. One-to-group variation streams~~usual statistics employed in one-to-one communication. New statistic definitionsare~~trying to measure how~~presented, which summarizethe~~performance varies among~~one-to-one statistics overall~~of~~the~~users~~Receiversin~~a multicast group associated to one source.~~the Group. 6.1. Discussion on the Impact of packet loss on statisticsThe~~word "variation" in this memo is~~packet loss does have effects on one-way metrics and their statistics. For example,the~~population standard deviation.~~lost packet can result an infinite one- way delay.It~~reflects~~is easy to handlethe~~relative performancesituation~~problem by simply ignoring the infinite valuein~~a multiparty communication service, i.e.,~~the~~level of~~metrics and inthe~~difference between~~calculation ofthe~~absolute performanceof each receivers. Using~~corresponding statistics. However,the~~one-to-group~~packet loss has so strong impact on the statistics calculation for the one-to-group metrics that it can not be solved by the same method used for one-way metrics. This is due to the complex of building a Matrix, which is needed for calculation of the statistics proposed in this memo. The situation is that measurement results obtained by different end users might have different packet loss pattern. For example, for User1, packet A was observed lost. And for User2, packet A was successfully received but packet B was lost. If the method to overcome the packet loss for one-way metrics is applied, the two singleton sets reported by User1 and User2 will be different in terms of the transmitted packets. Moreover, if User1 and User2 have different number of lost packets, the size of the results will be different. Therefore, for the centralized calculation, the reference point will not be able to use these two results to build up the group Matrix and can not calculate the statistics. In an extreme situation, no single packet arrives all users in the measurement and the Matrix will be empty. One of the possible solutions is to replace the infinite/undefined delay value by the average of the two adjacent values. For example, if the result reported by user1 is { R1dT1 R1dT2 R1dT3 ... R1dTK-1 UNDEF R1dTK+1... R1DM } where "UNDEF" is an undefined value, the reference point can replace it by R1dTK = {(R1dTK-1)+( R1dTK+1)}/2. Therefore, this result can be used to build up the group Matrix with an estimated value R1dTK. There are other possible solutions such as using the overallmeanof the whole result to replace the infinite/undefined value, and so on. It is out of the scope of this memo. For the distributed calculation, the reported statistics might have different "weight" to present the group performance, which is especially true for delay and jitter relevant metrics. For example, User1 calculates the Type-P-Finite-One-way-Delay-Mean R1DM as shown in Figure. 8 without any packet loss and User2 calculates the R2DM with N-2 packet loss. The R1DM and R2DM should not be treated with equal weight because R2DM was calculated only based on 2 delay values in the whole sample interval. One possible solution is to use a weight factor to mark every statistic value sent by users and use this factor for further statistic calculation. 6.2. General Metric Parameters o Src, the IP address of a host o G, the Group IP address o N, the number of Receivers (Recv1, Recv2, ... RecvN) o T, a time (start of test interval) o Tf, a time (end of test interval) o K, the number of packets sent from the source during the test interval o J[n], the number of packets received at a particular Receiver, n, where 1<=n<=N o lambda, a rate in reciprocal seconds (for Poisson Streams) o incT, the nominal duration of inter-packet interval, first bit to first bit (for Periodic Streams) o T0, a time that MUST be selected at random from the interval [T, T+I] to start generating packets and taking measurements (for Periodic Streams) o TstampSrc, the wire time of the packet as measured at MP(Src) (the Source Measurement Point) o TstampRecv, the wire time of the packet as measured at MP(Recv), assigned to packets that arrive within a "reasonable" time o Tmax, a maximum waiting time for packets at the destination, set sufficiently long to disambiguate packets with long delays from packets that are discarded (lost), thus the distribution of delay is not truncated o dT, shorthand notation for a one-way delay singleton value o L, shorthand notation for a one-way loss singleton value, either zero or one, where L=1 indicates loss and L=0 indicates arrival at the destination within TstampSrc + Tmax, may be indexed over n Receivers o DV, shorthand notation for a one-way delay variation singleton value 6.3. One-to-Group one-way Delay Statistics This section defines the overall one-way delay statistics for an entire Group or receivers. For example, we can define the group mean delay, as illustrated below. This is a metric designed to summarize the entire Matrix. Recv /----------- Sample -------------\ Stats Group Stat 1 R1dT1 R1dT2 R1dT3 ... R1dTk R1DM \ | 2 R2dT1 R2dT2 R2dT3 ... R2dTk R2DM | | 3 R3dT1 R3dT2 R3dT3 ... R3dTk R2DM > GMD . | . | . | n RndT1 RndT2 RndT3 ... RndTk RnDM / Figure 8: One-to-GroupGroup Mean Delay where: R1dT1 is the Type-P-Finite-One-way-Delay singleton evaluated at Receiver 1 for packet 1. R1DM is the Type-P-Finite-One-way-Delay-Mean evaluated at Receiver 1 for the sample of packets (1,...K). GMD is the mean of the sample means over all Receivers (1, ...N). 6.3.1. Definition and Metric Units Using the parameters above, we obtain the value of Type-P-One-way- Delay singleton for all packets sent during the test interval at each Receiver (Destination), as per [RFC2679]. For each packet that arrives within Tmax of its sending time, TstampSrc, the one-way delay singleton (dT) will be a finite value in units of seconds. Otherwise, the value of the singleton is Undefined. For each packet [i] that has a finite One-way Delay at Receiver n (in other words, excluding packets which have undefined one-way delay): Type-P-Finite-One-way-Delay-Receiver-n-[i] = = TstampRecv[i] - TstampSrc[i] The units of Finite one-way delay are seconds, with sufficient resolution to convey 3 significant digits. 6.3.2. Sample Mean Statistic This section defines the Sample Mean at each of N Receivers. Type-P-Finite-One-way-Delay-Mean-Receiver-n = RnDM = J[n] --- 1 \ --- * > Type-P-Finite-One-way-Delay-Receiver-n-[i] J[n] / --- i = 1 Figure 9: Type-P-Finite-One-way-Delay-Mean-Receiver-n where all packets i= 1 through J[n] have finite singleton delays. 6.3.3. One-to-Group Mean Delay Statistic This section defines the Mean One-way Delay calculated over the entire Group (or Matrix). Type-P-One-to-Group-Mean-Delay = GMD = N --- 1 \ - * > RnDM N / --- n = 1 Figure 10: Type-P-One-to-Group-Mean-Delay Note that the Group Mean Delay can also be calculated by summing the Finite one-way Delay singletons in the Matrix, and dividing by the number of Finite One-way Delay singletons. 6.3.4. One-to-Group Range of Mean Delays This section defines a metric for the range of mean delays over all N receivers in the Group, (R1DM, R2DM,...RnDM). Type-P-One-to-Group-Range-Mean-Delay = GRMD = max(RnDM) - min(RnDM) 6.3.5. One-to-Group Maximum of Mean Delays This section defines a metrics for the maximum of mean delays over all N receivers in the Group, (R1DM, R2DM,...RnDM). Type-P-One-to-Group-Max-Mean-Delay = GMMD = max(RnDM) 6.4. One-to-Group one-way Loss Statistics This section defines the overall 1-way loss statistics for an entire Group. For example, we can define the group loss ratio, as illustrated below. This is a metric designed to summarize the entire Matrix. Recv /----------- Sample ----------\ Stats Group Stat 1 R1L1 R1L2 R1L3 ... R1Lk R1LR \ | 2 R2L1 R2L2 R2L3 ... R2Lk R2LR | | 3 R3L1 R3L2 R3L3 ... R3Lk R3LR > GLR . | . | . | n RnL1 RnL2 RnL3 ... RnLk RnLR / Figure 11: One-to-Group Loss Ratio where: R1L1 is the Type-P-One-way-Loss singleton (L) evaluated at Receiver 1 for packet 1. R1LR is the Type-P-One-way-Loss-Ratio evaluated at Receiver 1 for the sample of packets (1,...K). GLR is the loss ratio over all Receivers (1, ..., N). 6.4.1. One-to-Group Loss Ratio The overall Group loss ratio id defined as Type-P-One-to-Group-Loss-Ratio = K,N --- 1 \ = --- * > L(k,n) K*N / --- k,n = 1 Figure 12 ALL Loss ratios are expressed in units of packets lost to total packets sent. 6.4.2. One-to-Group Loss Ratio Range Given a Matrix of loss singletons as illustrated above, determine the Type-P-One-way-Packet-Loss-Average for the sample at each receiver, according to the definitions and method of [RFC2680]. The Type-P- One-way-Packet-Loss-Average, RnLR for receiver n, and the Type-P-One- way-Loss-Ratio illustrated above are equivalent metrics. In terms of the parameters used here, these metrics definitions can be expressed as Type-P-One-way-Loss-Ratio-Receiver-n = RnLR = K --- 1 \ - * > RnLk K / --- k = 1 Figure 13: Type-P-One-way-Loss-Ratio-Receiver-n The One-to-Group Loss Ratio Range is defined as Type-P-One-to-Group-Loss-Ratio-Range = max(RnLR) - min(RnLR) It is most effective to indicate the range by giving both the max and minimum loss ratios for the Group, rather than only reporting the difference between them. 6.4.3. Comparative Loss Ratio Usually, the number of packets sent is used in the denominator of packet loss ratio metrics. For the comparative metrics defined here, the denominator is the maximum number of packets received at any receiver for the sample and test interval of interest. The Comparative Loss Ratio is defined as Type-P-Comp-Loss-Ratio-Receiver-n = RnCLR = K --- \ > Ln(k) / --- k=1 = ----------------------------- / K \ | --- | | \ | K - Min | > Ln(k) | | / | | --- | \ k=1 / N Figure 14: Type-P-Comp-Loss-Ratio-Receiver-n 6.5. One-to-Group one-way Delay Variation Statistics There is are two delay variation (DV) statistics to summarize the performance over the Group: the maximum DV over all receivers and the range of DV over all receivers. The detailed definitions are T0 BE PROVIDED. 7. Measurement Methods: Scaleability and Reporting Virtually all the guidance on measurement processes supplied by the earlier IPPM RFCs (such as [RFC2679] and [RFC2680]) for one-to-one scenarios is applicable here in the spatial and multiparty measurement scenario. The main difference is that the spatial and multiparty configurations require multiple measurement points where a stream of singletons will be collected. The amount of information requiring storage grows with both the number of metrics and the number of measurement points, so the scale of the measurement architecture multiplies the number of singleton results that must be collected and processed. It is possible that the architecture for results collection involves a single aggregation point with connectivity to all the measurement points. In this case, the number of measurement points determines both storage capacity and packet transfer capacity of the host acting as the aggregation point. However, both the storage and transfer capacity can be reduced if the measurement points are capable of computing the summary statistics that describe each measurement interval. This is consistent with many operational monitoring architectures today, where even the individual singletons may not be stored at each measurement point. In recognition of the likely need to minimize form of the results for storage and communication, the Group metrics above have been constructed to allow some computations on a per-Receiver basis. This means that each Receiver's statistics would normally have an equal weight with all other Receivers in the Group (regardless of the number of packets received). 7.1. Computation methods The scalability issue can be raised when there are thousands of points of interest in a group who are trying to send back the measurement results to the reference point for further processing and analysis. The points of interest can send either the whole measured sample or only the calculated statistics. The former one is a centralized statistic calculation method and the latter one is a distributed statistic calculation method. The sample should include all metrics parameters, the values and the corresponding sequence numbers. The transmission of the whole sample can cost much more bandwidth than the transmission of the statistics that should include all statistic parameters specified by policies and the additional information about the whole sample, such as the size of the sample, the group address, the address of the point of interest, the ID of the sample session,and~~one-to-group variation concepts, we~~so on. Apparently, the centralized calculation methodcan~~have~~require much more bandwidth than the distributed calculation method when the sample size is big. This is especially true when the measurement has huge number of the points of interest. It can lead toascalability issue at the reference point by over load the network resources. The distributed calculation method can savemuch~~clear understand~~more bandwidth and release the pressure of the scalability issue at the reference point side. However, it can result in the lack of information because not all measured singletons are obtained for building up the group matrix. The performance over time can be hidden from the analysis. For example, the loss pattern can be missed by simply accepting the loss ratio as well as the delay pattern. This tradeoff between the bandwidth consuming and the information acquiring has to be taken into account when design the measurement campaign to optimize the measurement results delivery. The possible solution could be to transit the statistic parameters to the reference point first to obtain the general information of the group performance. If the detail results are required, the reference point should send the requests to the points of interest, which could be particular ones or the whole group. This procedure can happen in the off peak time and can be well scheduled to avoid delivery of too many points of interest at the same time. Compression techniques can also be used to minimize the bandwidth required by the transmission. This could be a measurement protocol to report the measurement results. It is out of the scope of this memo. 7.2. Measurement To prevent any biais in the result, the configuration of a one-to- many measure must take in consideration that implicitly more packets will to be routed than send and selects a test packets rate that will not impact the network performance. 7.3. effect of Time and Space Aggregation OrderonGroup Stats This section presentsthe~~performanceof a multiparty communication service in terms~~impactof~~its trend and range. There can be mean~~the aggregation order on the scalability of the reportingand~~variation stream definitions for each~~of the~~three one-to- group metrics defined above. We only present~~the~~definition of Type- P-one-to-group-One-way-Delay-Space-Mean~~computation. It makes the hypothesis that receivers are managed remotlyand~~Type-P-one-to-group- One- way-Delay-Space-Variation as examples in this memo. 5.4.1. Type-P-one-to-group-One-way-Delay-Space-Mean Given a Type-P-one-to-group-One-way-Delay-Vector,~~not co-located. 2 methods are available to compute group statistics: Figure 8and (Figure 11) illustratethe~~mean { dT1, dT2,...,dTN } for~~method method choosen:the~~packet from Src at~~one-to-one statistic is computed per interval oftime~~T to { Recv1,...,RecvN }. For example, suppose we take a delay vector and~~beforethe~~results is: Delay_Vector = {dT1,...,dTN} Then~~computation ofthe mean overthe group of receivers [method1]; Figure 15 presents the second one, metric is computed overspace~~dimension would be: Delay_Space_Mean = DsM = sum{dT1,...,dTN}/N 5.4.2. Type-P-one-to-group-One-way-Delay-Variation-Stream Given a Type-P-one-to-group-One-way-Delay-Vector,~~and then over time [method2]. They differ only bythe~~variation { dT1, dT2,...,dTN } for~~order ofthe~~packet from Src at~~time~~T to { Recv1,...,RecvN }. We still take~~and ofthe~~above Delay_Vector~~space aggregation. Viewas~~an~~a matrix this order is neutral as it does not impact the result, but the impact on a measurement deployement is critical. Recv 1 R1S1 R1S1 R1S1 ... R1Sk \ | 2 R2S1 R2S2 R2S3 ... R2Sk | | 3 R3S1 R3S2 R3S3 ... R3Sk >sampleover space . | . | . | n RnS1 RnS2 RnS3 ... RnSk / S1M S2M S3M ... SnM Stats over space \------------- ------------/ \/ Group Stat over spaceandtime Figure 15: Impact of space aggregation on Group Stat In both casesthe~~variation would be: Delay_Variation_Stream = {SUM[(dT1-DsM)^2,...,(dTN- DsM)^2)}/N)^(1/2) 6. Extension from one-to-one~~volume of datato~~one-to-many measurement The above one-to-group metrics were defined~~report is proportionalto~~compose measurement results~~the numberofprobes. But there isa~~group of users who receive~~major difference between these 2 methods: method2: In space and time aggregation modethe~~same~~volume ofdata~~from one source. Moreover, this~~to collectis~~one~~proportionnal to the numberof~~efforts~~test packets received; Each received packet RiSi triggers out a block of data that must be reportedto~~introducing~~a common place for computingthe~~one-to- many concern to~~stat over space; method1: In time and space aggregation modethe~~IPPM working group with respect~~volume of data to collect is proportionnalto the~~fact that all existing documents in~~period of aggregation, so it does not depend onthe~~group are unicast oriented, which talk about only one-to-one single "path"~~number of packet received; Method 2 property has severe drawbacksin~~measurements. This concept can be extended from the "path" to "path tree" to cover both one-to- one~~terms of securityand~~one-to-many communications. Actually,~~dimensionning: The increasing ofthe~~one-to-one communications can be viewed as a special case~~rateof~~one-to-many from~~the~~routing point~~test packets may result in a sortof~~view.~~DoS toward the computation points;The~~one-to-many communications build up~~dimensioning ofa~~routing tree in~~measurement system is quite impossible to validate. The time agregation interval providesthe~~networks and one-to-one can be viewed as~~reporting side witha~~special simplified tree without branches but only the "trunk". Therefore, the one-to-group metrics described in this memo can even be viewed~~control of various collecting aspects suchas~~general metrics to measure the delay, jitter~~bandwidthand~~packet loss in IP networks. When~~computation and storage capacities. So this draft defines metrics based on method 1. Note: In some specific cases one may need sample of singletons over space. To adress this needit~~applies~~is suggested firstlyto~~one-to-one communications,~~limitthe~~metrics will have N receivers while N equal to 1. And~~number of test andthe~~statistic metrics for one-to-one communications are exactly~~number of test packets per seconds. Then reducingthe~~one- to-group metrics themselves when calculated using~~size ofthe~~methods given. 7.~~sample over time to one packet give sample of singleton over space.. 7.4. effect of Time and Space Aggregation Order on Spatial Stats TBD 8.Open issues~~8.~~9.Security Considerations Active~~measumrement: see~~measurement: (TODO:security~~section in~~considerations ofowd pl, jitter rfcsapplies(editor notes: add references).9.1.passive~~measurement:~~measurementThe generation of packets which match systematically the hash function may lead to a DoS attack toward the collector. The generation of packets with spoofing~~adresses~~addressesmay corrupt the results without any possibility to~~detect the spoofing. one-to-group metrics require collection~~detect the spoofing. 9.2. one-to-group metric The configuration of a measure must take in consideration that implicitly more packets will to be routed than send and selects a test packets rate accordingly. Collecting statistics from a huge numberof~~singletons which~~probesmay overloadany combination ofthe network the measurement controller is attach~~to. 9.~~to, measurement controller network interfaces and measurement controller computation capacities. one-to-group metrics: 10.Acknowledgments Lei would like to acknowledge Zhili Sun from CCSR, University of Surrey, for his instruction and helpful comments on this work.~~10.~~11.IANA Considerations Metrics defined in this memo~~will~~Metrics defined in this memo are designed tobe registered in the IANA IPPM METRICS REGISTRY as described in initial version of the registry~~[RFC4148]. 11.~~[RFC4148] : IANA is asked to register the following metrics in the IANA-IPPM- METRICS-REGISTRY-MIB : Spatial-One-way-Delay-Vector OBJECT-IDENTITY STATUS current DESCRIPTION "Type-P-Spatial-One-way-Delay-Vector" REFERENCE "Reference "RFCyyyy, section 4.1." -- RFC Ed.: replace yyyy with actual RFC number & remove this note := { ianaIppmMetrics nn } -- IANA assigns nn subpath-One-way-Delay-Stream OBJECT-IDENTITY STATUS current DESCRIPTION "Type-P-subpath-One-way-Delay-Stream" REFERENCE "Reference "RFCyyyy, section 4.2." -- RFC Ed.: replace yyyy with actual RFC number & remove this note := { ianaIppmMetrics nn } -- IANA assigns nn Spatial-One-way-Packet-Loss-Vector OBJECT-IDENTITY STATUS current DESCRIPTION "Type-P-Spatial-One-way-Packet-Loss-Vector" REFERENCE "Reference "RFCyyyy, section 4.3." -- RFC Ed.: replace yyyy with actual RFC number & remove this note := { ianaIppmMetrics nn } -- IANA assigns nn Spatial-One-way-Jitter-Vector OBJECT-IDENTITY STATUS current DESCRIPTION "Type-P-Spatial-One-way-Jitter-Vector" REFERENCE "Reference "RFCyyyy, section 4.4." -- RFC Ed.: replace yyyy with actual RFC number & remove this note := { ianaIppmMetrics nn } -- IANA assigns nn one-to-group-One-way-Delay-Vector OBJECT-IDENTITY STATUS current DESCRIPTION "Type-P-one-to-group-One-way-Delay-Vector" REFERENCE "Reference "RFCyyyy, section 5.1." -- RFC Ed.: replace yyyy with actual RFC number & remove this note := { ianaIppmMetrics nn } -- IANA assigns nn one-to-group-One-way-Packet-Loss-Vector OBJECT-IDENTITY STATUS current DESCRIPTION "Type-P-one-to-group-One-way-Packet-Loss-Vector" REFERENCE "Reference "RFCyyyy, section 5.2." -- RFC Ed.: replace yyyy with actual RFC number & remove this note := { ianaIppmMetrics nn } -- IANA assigns nn one-to-group-One-way-Jitter-Vector OBJECT-IDENTITY STATUS current DESCRIPTION "Type-P-one-to-group-One-way-Jitter-Vector" REFERENCE "Reference "RFCyyyy, section 5.3." -- RFC Ed.: replace yyyy with actual RFC number & remove this note := { ianaIppmMetrics nn } -- IANA assigns nn One-to-Group-Mean-Delay OBJECT-IDENTITY STATUS current DESCRIPTION "Type-P-One-to-Group-Mean-Delay" REFERENCE "Reference "RFCyyyy, section 6.3.3." -- RFC Ed.: replace yyyy with actual RFC number & remove this note := { ianaIppmMetrics nn } -- IANA assigns nn One-to-Group-Range-Mean-Delay OBJECT-IDENTITY STATUS current DESCRIPTION "Type-P-One-to-Group-Range-Mean-Delay" REFERENCE "Reference "RFCyyyy, section 6.3.4." -- RFC Ed.: replace yyyy with actual RFC number & remove this note := { ianaIppmMetrics nn } -- IANA assigns nn One-to-Group-Max-Mean-Delay OBJECT-IDENTITY STATUS current DESCRIPTION "Type-P-One-to-Group-Max-Mean-Delay" REFERENCE "Reference "RFCyyyy, section 6.3.5." -- RFC Ed.: replace yyyy with actual RFC number & remove this note := { ianaIppmMetrics nn } -- IANA assigns nn One-to-Group-Loss-Ratio OBJECT-IDENTITY STATUS current DESCRIPTION "Type-P-One-to-Group-Loss-Ratio" REFERENCE "Reference "RFCyyyy, section 6.4.1." -- RFC Ed.: replace yyyy with actual RFC number & remove this note := { ianaIppmMetrics nn } -- IANA assigns nn -- One-to-Group-Loss-Ratio-Range OBJECT-IDENTITY STATUS current DESCRIPTION "Type-P-One-to-Group-Loss-Ratio-Range" REFERENCE "Reference "RFCyyyy, section 6.4.2." -- RFC Ed.: replace yyyy with actual RFC number & remove this note := { ianaIppmMetrics nn } -- IANA assigns nn -- 12.References~~11.1.~~12.1.Normative References [RFC2330] Paxson, V., Almes, G., Mahdavi, J., and M. Mathis, "Framework for IP Performance Metrics", RFC 2330, May 1998. [RFC2679] Almes, G., Kalidindi, S., and M. Zekauskas, "A One-way Delay Metric for IPPM", RFC 2679, September 1999. [RFC2680] Almes, G., Kalidindi, S., and M. Zekauskas, "A One-way Packet Loss Metric for IPPM", RFC 2680, September 1999. [RFC3393] Demichelis, C. and P. Chimento, "IP Packet Delay Variation Metric for IP Performance Metrics (IPPM)", RFC 3393, November 2002. [RFC4148] Stephan, E., "IP Performance Metrics (IPPM) Metrics Registry", BCP 108, RFC 4148, August 2005.~~11.2.~~12.2.Informative References~~[I-D.boschi-ipfix-reducing-redundancy] Boschi, E., "Reducing redundancy in IPFIX and PSAMP reports", draft-boschi-ipfix-reducing-redundancy-02 (work in progress), June 2006. [I-D.ietf-ippm-spatial-composition] Morton, A. and E. Stephan, "Spatial Composition of Metrics", draft-ietf-ippm-spatial-composition-01 (work in progress), June 2006. [I-D.quittek-ipfix-middlebox] Quittek, J., "Guidelines for IPFIX Implementations on Middleboxes", draft-quittek-ipfix-middlebox-00 (work in progress), February 2004.~~[RFC2678] Mahdavi, J. and V. Paxson, "IPPM Metrics for Measuring Connectivity", RFC 2678, September 1999. [RFC2681] Almes, G., Kalidindi, S., and M. Zekauskas, "A Round-trip Delay Metric for IPPM", RFC 2681, September 1999. [RFC3148] Mathis, M. and M. Allman, "A Framework for Defining Empirical Bulk Transfer Capacity Metrics", RFC 3148, July 2001. [RFC3357] Koodli, R. and R. Ravikanth, "One-way Loss Pattern Sample Metrics", RFC 3357, August 2002. [RFC3432] Raisanen, V., Grotefeld, G., and A. Morton, "Network performance measurement with periodic streams", RFC 3432, November 2002. [RFC3763] Shalunov, S. and B. Teitelbaum, "One-way Active Measurement Protocol (OWAMP) Requirements", RFC 3763, April 2004.[RFC4656] Shalunov, S., Teitelbaum, B., Karp, A., Boote, J., and M. Zekauskas, "A One-way Active Measurement Protocol (OWAMP)", RFC 4656, September 2006. [RFC4737] Morton, A., Ciavattone, L., Ramachandran, G., Shalunov, S., and J. Perser, "Packet Reordering Metrics", RFC 4737, November 2006.Authors' Addresses Stephan Emile France Telecom Division R&D 2 avenue Pierre Marzin Lannion, F-22307 Fax: +33 2 96 05 18 52 Email:~~emile.stephan@orange-ft.com~~emile.stephan@orange-ftgroup.comLei Liang CCSR, University of Surrey Guildford Surrey, GU2 7XH Fax: +44 1483 683641 Email: L.Liang@surrey.ac.uk Al Morton 200 Laurel Ave. South Middletown, NJ 07748 USA Phone: +1 732 420 1571 Email: acmorton@att.com Full Copyright Statement Copyright (C) The~~Internet Society (2006).~~IETF Trust (2007).This document is subject to the rights, licenses and restrictions contained in BCP 78, and except as set forth therein, the authors retain all their rights. This document and the information contained herein are provided on an "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS OR IS SPONSORED BY (IF ANY), THE INTERNET~~SOCIETY~~SOCIETY, THE IETF TRUSTAND THE INTERNET ENGINEERING TASK FORCE DISCLAIM ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Intellectual Property The IETF takes no position regarding the validity or scope of any Intellectual Property Rights or other rights that might be claimed to pertain to the implementation or use of the technology described in this document or the extent to which any license under such rights might or might not be available; nor does it represent that it has made any independent effort to identify any such rights. Information on the procedures with respect to rights in RFC documents can be found in BCP 78 and BCP 79. Copies of IPR disclosures made to the IETF Secretariat and any assurances of licenses to be made available, or the result of an attempt made to obtain a general license or permission for the use of such proprietary rights by implementers or users of this specification can be obtained from the IETF on-line IPR repository at http://www.ietf.org/ipr. The IETF invites any interested party to bring to its attention any copyrights, patents or patent applications, or other proprietary rights that may cover technology that may be required to implement this standard. Please address the information to the IETF at ietf-ipr@ietf.org. Acknowledgment Funding for the RFC Editor function is provided by the IETF Administrative Support Activity (IASA).