RFC 6049
Spatial Composition of Metrics
Internet Engineering Task Force (IETF) A. Morton Request for Comments: 6049 AT&T Labs Category: Standards Track E. Stephan ISSN: 2070-1721 France Telecom Orange January 2011 Spatial Composition of MetricsAbstract
This memo utilizes IP performance metrics that are applicable to both complete paths and sub-paths, and it defines relationships to compose a complete path metric from the sub-path metrics with some accuracy with regard to the actual metrics. This is called "spatial composition" in RFC 2330. The memo refers to the framework for metric composition, and provides background and motivation for combining metrics to derive others. The descriptions of several composed metrics and statistics follow. Status of This Memo This is an Internet Standards Track document. This document is a product of the Internet Engineering Task Force (IETF). It represents the consensus of the IETF community. It has received public review and has been approved for publication by the Internet Engineering Steering Group (IESG). Further information on Internet Standards is available in Section 2 of RFC 5741. Information about the current status of this document, any errata, and how to provide feedback on it may be obtained at http://www.rfc-editor.org/info/rfc6049.
Copyright Notice Copyright (c) 2011 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. This document may contain material from IETF Documents or IETF Contributions published or made publicly available before November 10, 2008. The person(s) controlling the copyright in some of this material may not have granted the IETF Trust the right to allow modifications of such material outside the IETF Standards Process. Without obtaining an adequate license from the person(s) controlling the copyright in such materials, this document may not be modified outside the IETF Standards Process, and derivative works of it may not be created outside the IETF Standards Process, except to format it for publication as an RFC or to translate it into languages other than English.
Table of Contents
1. Introduction ....................................................4 1.1. Motivation .................................................6 1.2. Requirements Language ......................................6 2. Scope and Application ...........................................6 2.1. Scope of Work ..............................................6 2.2. Application ................................................7 2.3. Incomplete Information .....................................7 3. Common Specifications for Composed Metrics ......................8 3.1. Name: Type-P ...............................................8 3.1.1. Metric Parameters ...................................8 3.1.2. Definition and Metric Units .........................9 3.1.3. Discussion and Other Details ........................9 3.1.4. Statistic ...........................................9 3.1.5. Composition Function ................................9 3.1.6. Statement of Conjecture and Assumptions ............10 3.1.7. Justification of the Composition Function ..........10 3.1.8. Sources of Deviation from the Ground Truth .........10 3.1.9. Specific Cases where the Conjecture Might Fail .....11 3.1.10. Application of Measurement Methodology ............12 4. One-Way Delay Composed Metrics and Statistics ..................12 4.1. Name: Type-P-Finite-One-way-Delay-<Sample>-Stream .........12 4.1.1. Metric Parameters ..................................12 4.1.2. Definition and Metric Units ........................12 4.1.3. Discussion and Other Details .......................13 4.1.4. Statistic ..........................................13 4.2. Name: Type-P-Finite-Composite-One-way-Delay-Mean ..........13 4.2.1. Metric Parameters ..................................13 4.2.2. Definition and Metric Units of the Mean Statistic ..14 4.2.3. Discussion and Other Details .......................14 4.2.4. Statistic ..........................................14 4.2.5. Composition Function: Sum of Means .................14 4.2.6. Statement of Conjecture and Assumptions ............15 4.2.7. Justification of the Composition Function ..........15 4.2.8. Sources of Deviation from the Ground Truth .........15 4.2.9. Specific Cases where the Conjecture Might Fail .....15 4.2.10. Application of Measurement Methodology ............16 4.3. Name: Type-P-Finite-Composite-One-way-Delay-Minimum .......16 4.3.1. Metric Parameters ..................................16 4.3.2. Definition and Metric Units of the Minimum Statistic ..........................................16 4.3.3. Discussion and Other Details .......................16 4.3.4. Statistic ..........................................16 4.3.5. Composition Function: Sum of Minima ................16 4.3.6. Statement of Conjecture and Assumptions ............17 4.3.7. Justification of the Composition Function ..........17 4.3.8. Sources of Deviation from the Ground Truth .........17
4.3.9. Specific Cases where the Conjecture Might Fail .....17
4.3.10. Application of Measurement Methodology ............17
5. Loss Metrics and Statistics ....................................18
5.1. Type-P-Composite-One-way-Packet-Loss-Empirical-Probability 18
5.1.1. Metric Parameters ..................................18
5.1.2. Definition and Metric Units ........................18
5.1.3. Discussion and Other Details .......................18
5.1.4. Statistic:
Type-P-One-way-Packet-Loss-Empirical-Probability ...18
5.1.5. Composition Function: Composition of
Empirical Probabilities ............................18
5.1.6. Statement of Conjecture and Assumptions ............19
5.1.7. Justification of the Composition Function ..........19
5.1.8. Sources of Deviation from the Ground Truth .........19
5.1.9. Specific Cases where the Conjecture Might Fail .....19
5.1.10. Application of Measurement Methodology ............19
6. Delay Variation Metrics and Statistics .........................20
6.1. Name: Type-P-One-way-pdv-refmin-<Sample>-Stream ...........20
6.1.1. Metric Parameters ..................................20
6.1.2. Definition and Metric Units ........................20
6.1.3. Discussion and Other Details .......................21
6.1.4. Statistics: Mean, Variance, Skewness, Quantile .....21
6.1.5. Composition Functions ..............................22
6.1.6. Statement of Conjecture and Assumptions ............23
6.1.7. Justification of the Composition Function ..........23
6.1.8. Sources of Deviation from the Ground Truth .........23
6.1.9. Specific Cases where the Conjecture Might Fail .....24
6.1.10. Application of Measurement Methodology ............24
7. Security Considerations ........................................24
7.1. Denial-of-Service Attacks .................................24
7.2. User Data Confidentiality .................................24
7.3. Interference with the Metrics .............................24
8. IANA Considerations ............................................25
9. Contributors and Acknowledgements ..............................27
10. References ....................................................28
10.1. Normative References .....................................28
10.2. Informative References ...................................28
1. Introduction
The IP Performance Metrics (IPPM) framework [RFC2330] describes two
forms of metric composition: spatial and temporal. The composition
framework [RFC5835] expands and further qualifies these original
forms into three categories. This memo describes spatial
composition, one of the categories of metrics under the umbrella of
the composition framework.
Spatial composition encompasses the definition of performance metrics that are applicable to a complete path, based on metrics collected on various sub-paths. The main purpose of this memo is to define the deterministic functions that yield the complete path metrics using metrics of the sub-paths. The effectiveness of such metrics is dependent on their usefulness in analysis and applicability with practical measurement methods. The relationships may involve conjecture, and [RFC2330] lists four points that the metric definitions should include: o the specific conjecture applied to the metric and assumptions of the statistical model of the process being measured (if any; see [RFC2330], Section 12), o a justification of the practical utility of the composition in terms of making accurate measurements of the metric on the path, o a justification of the usefulness of the composition in terms of making analysis of the path using A-frame concepts more effective, and o an analysis of how the conjecture could be incorrect. Also, [RFC2330] gives an example using the conjecture that the delay of a path is very nearly the sum of the delays of the exchanges and clouds of the corresponding path digest. This example is particularly relevant to those who wish to assess the performance of an inter-domain path without direct measurement, and the performance estimate of the complete path is related to the measured results for various sub-paths instead. Approximate functions between the sub-path and complete path metrics are useful, with knowledge of the circumstances where the relationships are/are not applicable. For example, we would not expect that delay singletons from each sub-path would sum to produce an accurate estimate of a delay singleton for the complete path (unless all the delays were essentially constant -- very unlikely). However, other delay statistics (based on a reasonable sample size) may have a sufficiently large set of circumstances where they are applicable.
1.1. Motivation
One-way metrics defined in other RFCs (such as [RFC2679] and [RFC2680]) all assume that the measurement can be practically carried out between the source and the destination of interest. Sometimes there are reasons that the measurement cannot be executed from the source to the destination. For instance, the measurement path may cross several independent domains that have conflicting policies, measurement tools and methods, and measurement time assignment. The solution then may be the composition of several sub-path measurements. This means each domain performs the one-way measurement on a sub-path between two nodes that are involved in the complete path, following its own policy, using its own measurement tools and methods, and using its own measurement timing. Under the appropriate conditions, one can combine the sub-path one-way metric results to estimate the complete path one-way measurement metric with some degree of accuracy.1.2. 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 RFC 2119 [RFC2119]. In this memo, the characters "<=" should be read as "less than or equal to" and ">=" as "greater than or equal to".2. Scope and Application
2.1. Scope of Work
For the primary IP Performance Metrics RFCs for loss [RFC2680], delay [RFC2679], and delay variation [RFC3393], this memo gives a set of metrics that can be composed from the same or similar sub-path metrics. This means that the composition function may utilize: o the same metric for each sub-path; o multiple metrics for each sub-path (possibly one that is the same as the complete path metric); o a single sub-path metric that is different from the complete path metric; o different measurement techniques like active [RFC2330], [RFC3432] and passive [RFC5474].
We note a possibility: using a complete path metric and all but one sub-path metric to infer the performance of the missing sub-path, especially when the "last" sub-path metric is missing. However, such de-composition calculations, and the corresponding set of issues they raise, are beyond the scope of this memo.2.2. Application
The composition framework [RFC5835] requires the specification of the applicable circumstances for each metric. In particular, each section addresses whether the metric: o Requires the same test packets to traverse all sub-paths or may use similar packets sent and collected separately in each sub-path. o Requires homogeneity of measurement methodologies or can allow a degree of flexibility (e.g., active, active spatial division [RFC5644], or passive methods produce the "same" metric). Also, the applicable sending streams will be specified, such as Poisson, Periodic, or both. o Needs information or access that will only be available within an operator's domain, or is applicable to inter-domain composition. o Requires synchronized measurement start and stop times in all sub-paths or largely overlapping measurement intervals, or no timing requirements. o Requires the assumption of sub-path independence with regard to the metric being defined/composed or other assumptions. o Has known sources of inaccuracy/error and identifies the sources.2.3. Incomplete Information
In practice, when measurements cannot be initiated on a sub-path (and perhaps the measurement system gives up during the test interval), then there will not be a value for the sub-path reported, and the entire test result SHOULD be recorded as "undefined". This case should be distinguished from the case where the measurement system continued to send packets throughout the test interval, but all were declared lost. When a composed metric requires measurements from sub-paths A, B, and C, and one or more of the sub-path results are undefined, then the composed metric SHOULD also be recorded as undefined.
3. Common Specifications for Composed Metrics
To reduce the redundant information presented in the detailed metrics sections that follow, this section presents the specifications that are common to two or more metrics. The section is organized using the same subsections as the individual metrics, to simplify comparisons. Also, the index variables are represented as follows: o m = index for packets sent. o n = index for packets received. o s = index for involved sub-paths.3.1. Name: Type-P
All metrics use the "Type-P" convention as described in [RFC2330]. The rest of the name is unique to each metric.3.1.1. Metric Parameters
o Src, the IP address of a host. o Dst, the IP address of a host. o T, a time (start of test interval). o Tf, a time (end of test interval). 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 dT, the duration of the allowed interval for Periodic Stream sample start times. o T0, a time that MUST be selected at random from the interval [T, T + dT] to start generating packets and taking measurements (for Periodic Streams). o TstampSrc, the wire time of the packet as measured at MP(Src) (measurement point at the source). o TstampDst, the wire time of the packet as measured at MP(Dst), 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 M, the total number of packets sent between T0 and Tf.
o N, the total number of packets received at Dst (sent between T0
and Tf).
o S, the number of sub-paths involved in the complete Src-Dst path.
o Type-P, as defined in [RFC2330], which includes any field that may
affect a packet's treatment as it traverses the network.
In metric names, the term "<Sample>" is intended to be replaced by
the name of the method used to define a sample of values of parameter
TstampSrc. This can be done in several ways, including:
1. Poisson: a pseudo-random Poisson process of rate lambda, whose
values fall between T and Tf. The time interval between
successive values of TstampSrc will then average 1/lambda, as per
[RFC2330].
2. Periodic: a Periodic stream process with pseudo-random start time
T0 between T and dT, and nominal inter-packet interval incT, as
per [RFC3432].
3.1.2. Definition and Metric Units
This section is unique for every metric.
3.1.3. Discussion and Other Details
This section is unique for every metric.
3.1.4. Statistic
This section is unique for every metric.
3.1.5. Composition Function
This section is unique for every metric.
3.1.6. Statement of Conjecture and Assumptions
This section is unique for each metric. The term "ground truth" is frequently used in these sections and is defined in Section 4.7 of [RFC5835].3.1.7. Justification of the Composition Function
It is sometimes impractical to conduct active measurements between every Src-Dst pair. Since the full mesh of N measurement points grows as N x N, the scope of measurement may be limited by testing resources. There may be varying limitations on active testing in different parts of the network. For example, it may not be possible to collect the desired sample size in each test interval when access link speed is limited, because of the potential for measurement traffic to degrade the user traffic performance. The conditions on a low-speed access link may be understood well enough to permit use of a small sample size/rate, while a larger sample size/rate may be used on other sub-paths. Also, since measurement operations have a real monetary cost, there is value in re-using measurements where they are applicable, rather than launching new measurements for every possible source-destination pair.3.1.8. Sources of Deviation from the Ground Truth
3.1.8.1. Sub-Path List Differs from Complete Path
The measurement packets, each having source and destination addresses intended for collection at edges of the sub-path, may take a different specific path through the network equipment and links when compared to packets with the source and destination addresses of the complete path. Example sources of parallel paths include Equal Cost Multi-Path and parallel (or bundled) links. Therefore, the performance estimated from the composition of sub-path measurements may differ from the performance experienced by packets on the complete path. Multiple measurements employing sufficient sub-path address pairs might produce bounds on the extent of this error. We also note the possibility of re-routing during a measurement interval, as it may affect the correspondence between packets traversing the complete path and the sub-paths that were "involved" prior to the re-route.
3.1.8.2. Sub-Path Contains Extra Network Elements
Related to the case of an alternate path described above is the case where elements in the measured path are unique to measurement system connectivity. For example, a measurement system may use a dedicated link to a LAN switch, and packets on the complete path do not traverse that link. The performance of such a dedicated link would be measured continuously, and its contribution to the sub-path metrics SHOULD be minimized as a source of error.3.1.8.3. Sub-Paths Have Incomplete Coverage
Measurements of sub-path performance may not cover all the network elements on the complete path. For example, the network exchange points might be excluded unless a cooperative measurement is conducted. In this example, test packets on the previous sub-path are received just before the exchange point, and test packets on the next sub-path are injected just after the same exchange point. Clearly, the set of sub-path measurements SHOULD cover all critical network elements in the complete path.3.1.8.4. Absence of Route
At a specific point in time, no viable route exists between the complete path source and destination. The routes selected for one or more sub-paths therefore differ from the complete path. Consequently, spatial composition may produce finite estimation of a ground truth metric (see Section 4.7 of [RFC5835]) between a source and a destination, even when the route between them is undefined.3.1.9. Specific Cases where the Conjecture Might Fail
This section is unique for most metrics (see the metric-specific sections). For delay-related metrics, one-way delay always depends on packet size and link capacity, since it is measured in [RFC2679] from first bit to last bit. If the size of an IP packet changes on its route (due to encapsulation), this can influence delay performance. However, the main error source may be the additional processing associated with encapsulation and encryption/decryption if not experienced or accounted for in sub-path measurements. Fragmentation is a major issue for composition accuracy, since all metrics require all fragments to arrive before proceeding, and fragmented complete path performance is likely to be different from performance with non-fragmented packets and composed metrics based on non-fragmented sub-path measurements.
Highly manipulated routing can cause measurement error if not expected and compensated for. For example, policy-based MPLS routing could modify the class of service for the sub-paths and complete path.3.1.10. Application of Measurement Methodology
o The methodology SHOULD use similar packets sent and collected separately in each sub-path, where "similar" in this case means that Type-P contains as many equal attributes as possible, while recognizing that there will be differences. Note that Type-P includes stream characteristics (e.g., Poisson, Periodic). o The methodology allows a degree of flexibility regarding test stream generation (e.g., active or passive methods can produce an equivalent result, but the lack of control over the source, timing, and correlation of passive measurements is much more challenging). o Poisson and/or Periodic streams are RECOMMENDED. o The methodology applies to both inter-domain and intra-domain composition. o The methodology SHOULD have synchronized measurement time intervals in all sub-paths, but largely overlapping intervals MAY suffice. o Assumption of sub-path independence with regard to the metric being defined/composed is REQUIRED.4. One-Way Delay Composed Metrics and Statistics
4.1. Name: Type-P-Finite-One-way-Delay-<Sample>-Stream
This metric is a necessary element of delay composition metrics, and its definition does not formally exist elsewhere in IPPM literature.4.1.1. Metric Parameters
See the common parameters section (Section 3.1.1).4.1.2. Definition and Metric Units
Using the parameters above, we obtain the value of the Type-P-One- way-Delay singleton as per [RFC2679].
For each packet "[i]" that has a finite one-way delay (in other
words, excluding packets that have undefined one-way delay):
Type-P-Finite-One-way-Delay-<Sample>-Stream[i] =
FiniteDelay[i] = TstampDst - TstampSrc
This metric is measured in units of time in seconds, expressed in
sufficiently low resolution to convey meaningful quantitative
information. For example, resolution of microseconds is usually
sufficient.
4.1.3. Discussion and Other Details
The "Type-P-Finite-One-way-Delay" metric permits calculation of the
sample mean statistic. This resolves the problem of including lost
packets in the sample (whose delay is undefined) and the issue with
the informal assignment of infinite delay to lost packets (practical
systems can only assign some very large value).
The Finite-One-way-Delay approach handles the problem of lost packets
by reducing the event space. We consider conditional statistics, and
estimate the mean one-way delay conditioned on the event that all
packets in the sample arrive at the destination (within the specified
waiting time, Tmax). This offers a way to make some valid statements
about one-way delay, at the same time avoiding events with undefined
outcomes. This approach is derived from the treatment of lost
packets in [RFC3393], and is similar to [Y.1540].
4.1.4. Statistic
All statistics defined in [RFC2679] are applicable to the finite one-
way delay, and additional metrics are possible, such as the mean (see
below).
4.2. Name: Type-P-Finite-Composite-One-way-Delay-Mean
This section describes a statistic based on the Type-P-Finite-One-
way-Delay-<Sample>-Stream metric.
4.2.1. Metric Parameters
See the common parameters section (Section 3.1.1).
4.2.2. Definition and Metric Units of the Mean Statistic
We define Type-P-Finite-One-way-Delay-Mean = N --- 1 \ MeanDelay = - * > (FiniteDelay [n]) N / --- n = 1 where all packets n = 1 through N have finite singleton delays. This metric is measured in units of time in seconds, expressed in sufficiently fine resolution to convey meaningful quantitative information. For example, resolution of microseconds is usually sufficient.4.2.3. Discussion and Other Details
The Type-P-Finite-One-way-Delay-Mean metric requires the conditional delay distribution described in Section 4.1.3.4.2.4. Statistic
This metric, a mean, does not require additional statistics.4.2.5. Composition Function: Sum of Means
The Type-P-Finite-Composite-One-way-Delay-Mean, or CompMeanDelay, for the complete source to destination path can be calculated from the sum of the mean delays of all of its S constituent sub-paths.
Then the
Type-P-Finite-Composite-One-way-Delay-Mean =
S
---
\
CompMeanDelay = > (MeanDelay [s])
/
---
s = 1
where sub-paths s = 1 to S are involved in the complete path.
4.2.6. Statement of Conjecture and Assumptions
The mean of a sufficiently large stream of packets measured on each
sub-path during the interval [T, Tf] will be representative of the
ground truth mean of the delay distribution (and the distributions
themselves are sufficiently independent), such that the means may be
added to produce an estimate of the complete path mean delay.
It is assumed that the one-way delay distributions of the sub-paths
and the complete path are continuous. The mean of multi-modal
distributions has the unfortunate property that such a value may
never occur.
4.2.7. Justification of the Composition Function
See the common section (Section 3).
4.2.8. Sources of Deviation from the Ground Truth
See the common section (Section 3).
4.2.9. Specific Cases where the Conjecture Might Fail
If any of the sub-path distributions are multi-modal, then the
measured means may not be stable, and in this case the mean will not
be a particularly useful statistic when describing the delay
distribution of the complete path.
The mean may not be a sufficiently robust statistic to produce a
reliable estimate, or to be useful even if it can be measured.
If a link contributing non-negligible delay is erroneously included
or excluded, the composition will be in error.
4.2.10. Application of Measurement Methodology
The requirements of the common section (Section 3) apply here as well.4.3. Name: Type-P-Finite-Composite-One-way-Delay-Minimum
This section describes a statistic based on the Type-P-Finite-One- way-Delay-<Sample>-Stream metric, and the composed metric based on that statistic.4.3.1. Metric Parameters
See the common parameters section (Section 3.1.1).4.3.2. Definition and Metric Units of the Minimum Statistic
We define Type-P-Finite-One-way-Delay-Minimum = MinDelay = (FiniteDelay [j]) such that for some index, j, where 1 <= j <= N FiniteDelay[j] <= FiniteDelay[n] for all n where all packets n = 1 through N have finite singleton delays. This metric is measured in units of time in seconds, expressed in sufficiently fine resolution to convey meaningful quantitative information. For example, resolution of microseconds is usually sufficient.4.3.3. Discussion and Other Details
The Type-P-Finite-One-way-Delay-Minimum metric requires the conditional delay distribution described in Section 4.1.3.4.3.4. Statistic
This metric, a minimum, does not require additional statistics.4.3.5. Composition Function: Sum of Minima
The Type-P-Finite-Composite-One-way-Delay-Minimum, or CompMinDelay, for the complete source to destination path can be calculated from the sum of the minimum delays of all of its S constituent sub-paths.
Then the
Type-P-Finite-Composite-One-way-Delay-Minimum =
S
---
\
CompMinDelay = > (MinDelay [s])
/
---
s = 1
4.3.6. Statement of Conjecture and Assumptions
The minimum of a sufficiently large stream of packets measured on
each sub-path during the interval [T, Tf] will be representative of
the ground truth minimum of the delay distribution (and the
distributions themselves are sufficiently independent), such that the
minima may be added to produce an estimate of the complete path
minimum delay.
It is assumed that the one-way delay distributions of the sub-paths
and the complete path are continuous.
4.3.7. Justification of the Composition Function
See the common section (Section 3).
4.3.8. Sources of Deviation from the Ground Truth
See the common section (Section 3).
4.3.9. Specific Cases where the Conjecture Might Fail
If the routing on any of the sub-paths is not stable, then the
measured minimum may not be stable. In this case the composite
minimum would tend to produce an estimate for the complete path that
may be too low for the current path.
4.3.10. Application of Measurement Methodology
The requirements of the common section (Section 3) apply here as
well.
5. Loss Metrics and Statistics
5.1. Type-P-Composite-One-way-Packet-Loss-Empirical-Probability
5.1.1. Metric Parameters
See the common parameters section (Section 3.1.1).5.1.2. Definition and Metric Units
Using the parameters above, we obtain the value of the Type-P-One- way-Packet-Loss singleton and stream as per [RFC2680]. We obtain a sequence of pairs with elements as follows: o TstampSrc, as above. o L, either zero or one, where L = 1 indicates loss and L = 0 indicates arrival at the destination within TstampSrc + Tmax.5.1.3. Discussion and Other Details
None.5.1.4. Statistic: Type-P-One-way-Packet-Loss-Empirical-Probability
Given the stream parameter M, the number of packets sent, we can define the Empirical Probability of Loss Statistic (Ep), consistent with average loss in [RFC2680], as follows: Type-P-One-way-Packet-Loss-Empirical-Probability = M --- 1 \ Ep = - * > (L[m]) M / --- m = 1 where all packets m = 1 through M have a value for L.5.1.5. Composition Function: Composition of Empirical Probabilities
The Type-P-One-way-Composite-Packet-Loss-Empirical-Probability, or CompEp, for the complete source to destination path can be calculated by combining the Ep of all of its constituent sub-paths (Ep1, Ep2, Ep3, ... Epn) as
Type-P-Composite-One-way-Packet-Loss-Empirical-Probability =
CompEp = 1 - {(1 - Ep1) x (1 - Ep2) x (1 - Ep3) x ... x (1 - EpS)}
If any Eps is undefined in a particular measurement interval,
possibly because a measurement system failed to report a value, then
any CompEp that uses sub-path s for that measurement interval is
undefined.
5.1.6. Statement of Conjecture and Assumptions
The empirical probability of loss calculated on a sufficiently large
stream of packets measured on each sub-path during the interval
[T, Tf] will be representative of the ground truth empirical loss
probability (and the probabilities themselves are sufficiently
independent), such that the sub-path probabilities may be combined to
produce an estimate of the complete path empirical loss probability.
5.1.7. Justification of the Composition Function
See the common section (Section 3).
5.1.8. Sources of Deviation from the Ground Truth
See the common section (Section 3).
5.1.9. Specific Cases where the Conjecture Might Fail
A concern for loss measurements combined in this way is that root
causes may be correlated to some degree.
For example, if the links of different networks follow the same
physical route, then a single catastrophic event like a fire in a
tunnel could cause an outage or congestion on remaining paths in
multiple networks. Here it is important to ensure that measurements
before the event and after the event are not combined to estimate the
composite performance.
Or, when traffic volumes rise due to the rapid spread of an email-
borne worm, loss due to queue overflow in one network may help
another network to carry its traffic without loss.
5.1.10. Application of Measurement Methodology
See the common section (Section 3).
6. Delay Variation Metrics and Statistics
6.1. Name: Type-P-One-way-pdv-refmin-<Sample>-Stream
This packet delay variation (PDV) metric is a necessary element of Composed Delay Variation metrics, and its definition does not formally exist elsewhere in IPPM literature (with the exception of [RFC5481]).6.1.1. Metric Parameters
In addition to the parameters of Section 3.1.1: o TstampSrc[i], the wire time of packet[i] as measured at MP(Src) (measurement point at the source). o TstampDst[i], the wire time of packet[i] as measured at MP(Dst), assigned to packets that arrive within a "reasonable" time. o B, a packet length in bits. o F, a selection function unambiguously defining the packets from the stream that are selected for the packet-pair computation of this metric. F(current packet), the first packet of the pair, MUST have a valid Type-P-Finite-One-way-Delay less than Tmax (in other words, excluding packets that have undefined one-way delay) and MUST have been transmitted during the interval [T, Tf]. The second packet in the pair, F(min_delay packet) MUST be the packet with the minimum valid value of Type-P-Finite-One-way-Delay for the stream, in addition to the criteria for F(current packet). If multiple packets have equal minimum Type-P-Finite-One-way-Delay values, then the value for the earliest arriving packet SHOULD be used. o MinDelay, the Type-P-Finite-One-way-Delay value for F(min_delay packet) given above. o N, the number of packets received at the destination that meet the F(current packet) criteria.6.1.2. Definition and Metric Units
Using the definition above in Section 5.1.2, we obtain the value of Type-P-Finite-One-way-Delay-<Sample>-Stream[n], the singleton for each packet[i] in the stream (a.k.a. FiniteDelay[i]).
For each packet[n] that meets the F(first packet) criteria given
above: Type-P-One-way-pdv-refmin-<Sample>-Stream[n] =
PDV[n] = FiniteDelay[n] - MinDelay
where PDV[i] is in units of time in seconds, expressed in
sufficiently fine resolution to convey meaningful quantitative
information. For example, resolution of microseconds is usually
sufficient.
6.1.3. Discussion and Other Details
This metric produces a sample of delay variation normalized to the
minimum delay of the sample. The resulting delay variation
distribution is independent of the sending sequence (although
specific FiniteDelay values within the distribution may be
correlated, depending on various stream parameters such as packet
spacing). This metric is equivalent to the IP Packet Delay Variation
parameter defined in [Y.1540].
6.1.4. Statistics: Mean, Variance, Skewness, Quantile
We define the mean PDV as follows (where all packets n = 1 through N
have a value for PDV[n]):
Type-P-One-way-pdv-refmin-Mean = MeanPDV =
N
---
1 \
- * > (PDV[n])
N /
---
n = 1
We define the variance of PDV as follows:
Type-P-One-way-pdv-refmin-Variance = VarPDV =
N
---
1 \ 2
------- > (PDV[n] - MeanPDV)
(N - 1) /
---
n = 1
We define the skewness of PDV as follows:
Type-P-One-way-pdv-refmin-Skewness = SkewPDV =
N
--- 3
\ / \
> | PDV[n] - MeanPDV |
/ \ /
---
n = 1
-----------------------------------
/ \
| ( 3/2 ) |
\ (N - 1) * VarPDV /
(See Appendix X of [Y.1541] for additional background information.)
We define the quantile of the PDV sample as the value where the
specified fraction of singletons is less than the given value.
6.1.5. Composition Functions
This section gives two alternative composition functions. The
objective is to estimate a quantile of the complete path delay
variation distribution. The composed quantile will be estimated
using information from the sub-path delay variation distributions.
6.1.5.1. Approximate Convolution
The Type-P-Finite-One-way-Delay-<Sample>-Stream samples from each
sub-path are summarized as a histogram with 1-ms bins representing
the one-way delay distribution.
From [STATS], the distribution of the sum of independent random
variables can be derived using the relation:
Type-P-Composite-One-way-pdv-refmin-quantile-a =
. .
/ /
P(X + Y + Z <= a) = | | P(X <= a - y - z) * P(Y = y) * P(Z = z) dy dz
/ /
` `
z y
Note that dy and dz indicate partial integration above, and that y and z are the integration variables. Also, the probability of an outcome is indicated by the symbol P(outcome), where X, Y, and Z are random variables representing the delay variation distributions of the sub-paths of the complete path (in this case, there are three sub-paths), and "a" is the quantile of interest. This relation can be used to compose a quantile of interest for the complete path from the sub-path delay distributions. The histograms with 1-ms bins are discrete approximations of the delay distributions.6.1.5.2. Normal Power Approximation (NPA)
Type-P-One-way-Composite-pdv-refmin-NPA for the complete source to destination path can be calculated by combining the statistics of all the constituent sub-paths in the process described in [Y.1541], Clause 8 and Appendix X.6.1.6. Statement of Conjecture and Assumptions
The delay distribution of a sufficiently large stream of packets measured on each sub-path during the interval [T, Tf] will be sufficiently stationary, and the sub-path distributions themselves are sufficiently independent, so that summary information describing the sub-path distributions can be combined to estimate the delay distribution of the complete path. It is assumed that the one-way delay distributions of the sub-paths and the complete path are continuous.6.1.7. Justification of the Composition Function
See the common section (Section 3).6.1.8. Sources of Deviation from the Ground Truth
In addition to the common deviations, a few additional sources exist here. For one, very tight distributions with ranges on the order of a few milliseconds are not accurately represented by a histogram with 1-ms bins. This size was chosen assuming an implicit requirement on accuracy: errors of a few milliseconds are acceptable when assessing a composed distribution quantile. Also, summary statistics cannot describe the subtleties of an empirical distribution exactly, especially when the distribution is very different from a classical form. Any procedure that uses these statistics alone may incur error.
6.1.9. Specific Cases where the Conjecture Might Fail
If the delay distributions of the sub-paths are somehow correlated, then neither of these composition functions will be reliable estimators of the complete path distribution. In practice, sub-path delay distributions with extreme outliers have increased the error of the composed metric estimate.6.1.10. Application of Measurement Methodology
See the common section (Section 3).7. Security Considerations
7.1. Denial-of-Service Attacks
This metric requires a stream of packets sent from one host (source) to another host (destination) through intervening networks. This method could be abused for denial-of-service attacks directed at the destination and/or the intervening network(s). Administrators of source, destination, and intervening networks should establish bilateral or multilateral agreements regarding the timing, size, and frequency of collection of sample metrics. Use of this method in excess of the terms agreed upon between the participants may be cause for immediate rejection or discarding of packets, or other escalation procedures defined between the affected parties.7.2. User Data Confidentiality
Active use of this method generates packets for a sample, rather than taking samples based on user data, and does not threaten user data confidentiality. Passive measurement MUST restrict attention to the headers of interest. Since user payloads may be temporarily stored for length analysis, suitable precautions MUST be taken to keep this information safe and confidential. In most cases, a hashing function will produce a value suitable for payload comparisons.7.3. Interference with the Metrics
It may be possible to identify that a certain packet or stream of packets is part of a sample. With that knowledge at the destination and/or the intervening networks, it is possible to change the
processing of the packets (e.g., increasing or decreasing delay), which may distort the measured performance. It may also be possible to generate additional packets that appear to be part of the sample metric. These additional packets are likely to perturb the results of the sample measurement. To discourage the kind of interference mentioned above, packet interference checks, such as cryptographic hash, may be used.8. IANA Considerations
Metrics defined in the IETF are typically registered in the IANA IPPM Metrics Registry as described in the initial version of the registry [RFC4148]. IANA has registered the following metrics in the IANA-IPPM-METRICS-REGISTRY-MIB: ietfFiniteOneWayDelayStream OBJECT-IDENTITY STATUS current DESCRIPTION "Type-P-Finite-One-way-Delay-Stream" REFERENCE "RFC 6049, Section 4.1." ::= { ianaIppmMetrics 71 } ietfFiniteOneWayDelayMean OBJECT-IDENTITY STATUS current DESCRIPTION "Type-P-Finite-One-way-Delay-Mean" REFERENCE "RFC 6049, Section 4.2." ::= { ianaIppmMetrics 72 } ietfCompositeOneWayDelayMean OBJECT-IDENTITY STATUS current DESCRIPTION "Type-P-Finite-Composite-One-way-Delay-Mean" REFERENCE "RFC 6049, Section 4.2.5." ::= { ianaIppmMetrics 73 } ietfFiniteOneWayDelayMinimum OBJECT-IDENTITY STATUS current DESCRIPTION "Type-P-Finite-One-way-Delay-Minimum" REFERENCE "RFC 6049, Section 4.3.2." ::= { ianaIppmMetrics 74 }
ietfCompositeOneWayDelayMinimum OBJECT-IDENTITY
STATUS current
DESCRIPTION
"Type-P-Finite-Composite-One-way-Delay-Minimum"
REFERENCE "RFC 6049, Section 4.3."
::= { ianaIppmMetrics 75 }
ietfOneWayPktLossEmpiricProb OBJECT-IDENTITY
STATUS current
DESCRIPTION
"Type-P-One-way-Packet-Loss-Empirical-Probability"
REFERENCE "RFC 6049, Section 5.1.4"
::= { ianaIppmMetrics 76 }
ietfCompositeOneWayPktLossEmpiricProb OBJECT-IDENTITY
STATUS current
DESCRIPTION
"Type-P-Composite-One-way-Packet-Loss-Empirical-Probability"
REFERENCE "RFC 6049, Section 5.1."
::= { ianaIppmMetrics 77 }
ietfOneWayPdvRefminStream OBJECT-IDENTITY
STATUS current
DESCRIPTION
"Type-P-One-way-pdv-refmin-Stream"
REFERENCE "RFC 6049, Section 6.1."
::= { ianaIppmMetrics 78 }
ietfOneWayPdvRefminMean OBJECT-IDENTITY
STATUS current
DESCRIPTION
"Type-P-One-way-pdv-refmin-Mean"
REFERENCE "RFC 6049, Section 6.1.4."
::= { ianaIppmMetrics 79 }
ietfOneWayPdvRefminVariance OBJECT-IDENTITY
STATUS current
DESCRIPTION
"Type-P-One-way-pdv-refmin-Variance"
REFERENCE "RFC 6049, Section 6.1.4."
::= { ianaIppmMetrics 80 }
ietfOneWayPdvRefminSkewness OBJECT-IDENTITY
STATUS current
DESCRIPTION
"Type-P-One-way-pdv-refmin-Skewness"
REFERENCE "RFC 6049, Section 6.1.4."
::= { ianaIppmMetrics 81 }
ietfCompositeOneWayPdvRefminQtil OBJECT-IDENTITY
STATUS current
DESCRIPTION
"Type-P-Composite-One-way-pdv-refmin-quantile-a"
REFERENCE "RFC 6049, Section 6.1.5.1."
::= { ianaIppmMetrics 82 }
ietfCompositeOneWayPdvRefminNPA OBJECT-IDENTITY
STATUS current
DESCRIPTION
"Type-P-One-way-Composite-pdv-refmin-NPA"
REFERENCE "RFC 6049, Section 6.1.5.2."
::= { ianaIppmMetrics 83 }
9. Contributors and Acknowledgements
The following people have contributed useful ideas, suggestions, or
the text of sections that have been incorporated into this memo:
- Phil Chimento <vze275m9@verizon.net>
- Reza Fardid <RFardid@cariden.com>
- Roman Krzanowski <roman.krzanowski@verizon.com>
- Maurizio Molina <maurizio.molina@dante.org.uk>
- Lei Liang <L.Liang@surrey.ac.uk>
- Dave Hoeflin <dhoeflin@att.com>
A long time ago, in a galaxy far, far away (Minneapolis), Will Leland
suggested the simple and elegant Type-P-Finite-One-way-Delay concept.
Thanks Will.
Yaakov Stein and Donald McLachlan also provided useful comments along
the way.
10. References
10.1. Normative References
[RFC2119] Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, March 1997. [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. [RFC3432] Raisanen, V., Grotefeld, G., and A. Morton, "Network performance measurement with periodic streams", RFC 3432, November 2002. [RFC4148] Stephan, E., "IP Performance Metrics (IPPM) Metrics Registry", BCP 108, RFC 4148, August 2005. [RFC5835] Morton, A. and S. Van den Berghe, "Framework for Metric Composition", RFC 5835, April 2010.10.2. Informative References
[RFC5474] Duffield, N., Chiou, D., Claise, B., Greenberg, A., Grossglauser, M., and J. Rexford, "A Framework for Packet Selection and Reporting", RFC 5474, March 2009. [RFC5481] Morton, A. and B. Claise, "Packet Delay Variation Applicability Statement", RFC 5481, March 2009. [RFC5644] Stephan, E., Liang, L., and A. Morton, "IP Performance Metrics (IPPM): Spatial and Multicast", RFC 5644, October 2009. [STATS] Mood, A., Graybill, F., and D. Boes, "Introduction to the Theory of Statistics, 3rd Edition", McGraw-Hill, New York, NY, 1974.
[Y.1540] ITU-T Recommendation Y.1540, "Internet protocol data communication service - IP packet transfer and availability performance parameters", November 2007. [Y.1541] ITU-T Recommendation Y.1541, "Network Performance Objectives for IP-based Services", February 2006.Authors' Addresses
Al Morton AT&T Labs 200 Laurel Avenue South Middletown, NJ 07748 USA Phone: +1 732 420 1571 Fax: +1 732 368 1192 EMail: acmorton@att.com URI: http://home.comcast.net/~acmacm/ Stephan Emile France Telecom Orange 2 avenue Pierre Marzin Lannion, F-22307 France EMail: emile.stephan@orange-ftgroup.com