draft-ietf-ippm-delay-04.txt   draft-ietf-ippm-delay-05.txt 
Network Working Group G. Almes Network Working Group G. Almes
Internet Draft S. Kalidindi Internet Draft S. Kalidindi
Expiration Date: March 1999 M. Zekauskas Expiration Date: May 1999 M. Zekauskas
Advanced Network & Services Advanced Network & Services
August 1998 November 1998
A One-way Delay Metric for IPPM A One-way Delay Metric for IPPM
<draft-ietf-ippm-delay-04.txt> <draft-ietf-ippm-delay-05.txt>
1. Status of this Memo 1. Status of this Memo
This document is an Internet-Draft. Internet-Drafts are working This document is an Internet-Draft. Internet-Drafts are working
documents of the Internet Engineering Task Force (IETF), its areas, documents of the Internet Engineering Task Force (IETF), its areas,
and its working groups. Note that other groups may also distribute and its working groups. Note that other groups may also distribute
working documents as Internet Drafts. working documents as Internet-Drafts.
Internet-Drafts are draft documents valid for a maximum of six Internet-Drafts are draft documents valid for a maximum of six
months, and may be updated, replaced, or obsoleted by other documents months, and may be updated, replaced, or obsoleted by other documents
at any time. It is inappropriate to use Internet- Drafts as at any time. It is inappropriate to use Internet-Drafts as reference
reference material or to cite them other than as "work in progress." material or to cite them other than as "work in progress."
To view the entire list of current Internet-Drafts, please check the To view the entire list of current Internet-Drafts, please check the
"1id-abstracts.txt" listing contained in the Internet-Drafts shadow "1id-abstracts.txt" listing contained in the Internet-Drafts shadow
directories on ftp.is.co.za (Africa), nic.nordu.net (Northern directories on ftp.is.co.za (Africa), nic.nordu.net (Northern
Europe), ftp.nis.garr.it (Southern Europe), munnari.oz.au (Pacific Europe), ftp.nis.garr.it (Southern Europe), munnari.oz.au (Pacific
Rim), ftp.ietf.org (US East Coast), or ftp.isi.edu (US West Coast). Rim), ftp.ietf.org (US East Coast), or ftp.isi.edu (US West Coast).
This memo provides information for the Internet community. This memo This memo provides information for the Internet community. This memo
does not specify an Internet standard of any kind. Distribution of does not specify an Internet standard of any kind. Distribution of
this memo is unlimited. this memo is unlimited.
2. Introduction 2. Introduction
This memo defines a metric for one-way delay of packets across This memo defines a metric for one-way delay of packets across
Internet paths. It builds on notions introduced and discussed in the Internet paths. It builds on notions introduced and discussed in the
IPPM Framework document, RFC 2330 [1]; the reader is assumed to be IPPM Framework document, RFC 2330 [1]; the reader is assumed to be
familiar with that document. familiar with that document.
This memo is intended to be parallel in structure to a companion This memo is intended to be parallel in structure to a companion
document for Packet Loss ("A Packet Loss Metric for IPPM" document for Packet Loss ("A Packet Loss Metric for IPPM"
<draft-ietf-ippm-loss-04.txt>) [2]. <draft-ietf-ippm-loss-05.txt>) [2].
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 [6].
Although RFC 2119 was written with protocols in mind, the key words
are used in this document for similar reasons. They are used to
ensure the results of measurements from two different implementations
are comparable, and to note instances when an implementation could
perturb the network.
The structure of the memo is as follows: The structure of the memo is as follows:
+ A 'singleton' analytic metric, called Type-P-One-way-Delay, will + A 'singleton' analytic metric, called Type-P-One-way-Delay, will
be introduced to measure a single observation of one-way delay. be introduced to measure a single observation of one-way delay.
+ Using this singleton metric, a 'sample', called Type-P-One-way- + Using this singleton metric, a 'sample', called Type-P-One-way-
Delay-Poisson-Stream, will be introduced to measure a sequence of Delay-Poisson-Stream, will be introduced to measure a sequence of
singleton delays measured at times taken from a Poisson process. singleton delays measured at times taken from a Poisson process.
skipping to change at page 2, line 46 skipping to change at page 3, line 8
+ The minimum value of this metric provides an indication of the + The minimum value of this metric provides an indication of the
delay due only to propagation and transmission delay. delay due only to propagation and transmission delay.
+ The minimum value of this metric provides an indication of the + The minimum value of this metric provides an indication of the
delay that will likely be experienced when the path* traversed is delay that will likely be experienced when the path* traversed is
lightly loaded. lightly loaded.
+ Values of this metric above the minimum provide an indication of + Values of this metric above the minimum provide an indication of
the congestion present in the path. the congestion present in the path.
The measurement of one-way delay instead of round-trip delay is
motivated by the following factors:
+ In today's Internet, the path from a source to a destination may
be different than the path from the destination back to the source
("asymmetric paths"), such that different sequences of routers are
used for the forward and reverse paths. Therefore round-trip
measurements actually measure the performance of two distinct
paths together. Measuring each path independently highlights the
performance difference between the two paths which may traverse
different Internet service providers, and even radically different
types of networks (for example, research versus commodity
networks, or ATM versus packet-over-SONET).
+ Even when the two paths are symmetric, they may have radically
different performance characteristics due to asymmetric queueing.
+ Performance of an application may depend mostly on the performance
in one direction. For example, a file transfer using TCP may
depend more on the performance in the direction that data flows,
rather than the direction in which acknowledgements travel.
+ In quality-of-service (QoS) enabled networks, provisioning in one
direction may be radically different than provisioning in the
reverse direction, and thus the QoS guarantees differ. Measuring
the paths independently allows the verification of both
guarantees.
It is outside the scope of this document to say precisely how delay It is outside the scope of this document to say precisely how delay
metrics would be applied to specific problems. metrics would be applied to specific problems.
2.2. General Issues Regarding Time 2.2. General Issues Regarding Time
Whenever a time (i.e., a moment in history) is mentioned here, it is Whenever a time (i.e., a moment in history) is mentioned here, it is
understood to be measured in seconds (and fractions) relative to UTC. understood to be measured in seconds (and fractions) relative to UTC.
As described more fully in the Framework document, there are four As described more fully in the Framework document, there are four
distinct, but related notions of clock uncertainty: distinct, but related notions of clock uncertainty:
skipping to change at page 3, line 37 skipping to change at page 4, line 23
on an old Unix host might tick only once every 10 msec, and thus on an old Unix host might tick only once every 10 msec, and thus
have a resolution of only 10 msec. have a resolution of only 10 msec.
skew* skew*
measures the change of accuracy, or of synchronization, with measures the change of accuracy, or of synchronization, with
time. For example, the clock on a given host might gain 1.3 time. For example, the clock on a given host might gain 1.3
msec per hour and thus be 27.1 msec behind UTC at one time and msec per hour and thus be 27.1 msec behind UTC at one time and
only 25.8 msec an hour later. In this case, we say that the only 25.8 msec an hour later. In this case, we say that the
clock of the given host has a skew of 1.3 msec per hour relative clock of the given host has a skew of 1.3 msec per hour relative
to UTC, and this threatens accuracy. We might also speak of the to UTC, which threatens accuracy. We might also speak of the
skew of one clock relative to another clock, and this threatens skew of one clock relative to another clock, which threatens
synchronization. synchronization.
3. A Singleton Definition for One-way Delay 3. A Singleton Definition for One-way Delay
3.1. Metric Name: 3.1. Metric Name:
Type-P-One-way-Delay Type-P-One-way-Delay
3.2. Metric Parameters: 3.2. Metric Parameters:
skipping to change at page 6, line 39 skipping to change at page 7, line 21
to expect the test packet, and the means by which Src and Dst are to expect the test packet, and the means by which Src and Dst are
synchronized are outside the scope of this document. {Comment: We synchronized are outside the scope of this document. {Comment: We
plan to document elsewhere our own work in describing such more plan to document elsewhere our own work in describing such more
detailed implementation techniques and we encourage others to as detailed implementation techniques and we encourage others to as
well.} well.}
3.7. Errors and Uncertainties: 3.7. Errors and Uncertainties:
The description of any specific measurement method should include an The description of any specific measurement method should include an
accounting and analysis of various sources of error or uncertainty. accounting and analysis of various sources of error or uncertainty.
The Framework document provides general guidence on this point, but The Framework document provides general guidance on this point, but
we note here the following specifics related to delay metrics: we note here the following specifics related to delay metrics:
+ Errors or uncertainties due to uncertainties in the clocks of the + Errors or uncertainties due to uncertainties in the clocks of the
Src and Dst hosts. Src and Dst hosts.
+ Errors or uncertainties due to the difference between 'wire time' + Errors or uncertainties due to the difference between 'wire time'
and 'host time'. and 'host time'.
In addition, the loss threshold may affect the results. Each of In addition, the loss threshold may affect the results. Each of
these are discussed in more detail below, along with a section these are discussed in more detail below, along with a section
("Calibration") on accounting for these errors and uncertainties. ("Calibration") on accounting for these errors and uncertainties.
3.7.1. Errors or uncertainties related to Clocks 3.7.1. Errors or uncertainties related to Clocks
The uncertainty in a measurement of one-way delay is related, in The uncertainty in a measurement of one-way delay is related, in
part, to uncertainties in the clocks of the Src and Dst hosts. In part, to uncertainties in the clocks of the Src and Dst hosts. In
the following, we refer to the clock used to measure when the packet the following, we refer to the clock used to measure when the packet
was sent from Src as the source clock, we refer to the clock used to was sent from Src as the source clock, we refer to the clock used to
measure when the packet was received by Dst as the dest clock, we measure when the packet was received by Dst as the destination clock,
refer to the observed time when the packet was sent by the source we refer to the observed time when the packet was sent by the source
clock as Tsource, and the observed time when the packet was received clock as Tsource, and the observed time when the packet was received
by the dest clock as Tdest. Alluding to the notions of by the destination clock as Tdest. Alluding to the notions of
synchronization, accuracy, resolution, and skew mentioned in the synchronization, accuracy, resolution, and skew mentioned in the
Introduction, we note the following: Introduction, we note the following:
+ Any error in the synchronization between the source clock and the + Any error in the synchronization between the source clock and the
dest clock will contribute to error in the delay measurement. We destination clock will contribute to error in the delay
say that the source clock and the dest clock have a measurement. We say that the source clock and the destination
synchronization error of Tsynch if the source clock is Tsynch clock have a synchronization error of Tsynch if the source clock
ahead of the dest clock. Thus, if we know the value of Tsynch is Tsynch ahead of the destination clock. Thus, if we know the
exactly, we could correct for clock synchronization by adding value of Tsynch exactly, we could correct for clock
Tsynch to the uncorrected value of Tdest-Tsource. synchronization by adding Tsynch to the uncorrected value of
Tdest-Tsource.
+ The accuracy of a clock is important only in identifying the time + The accuracy of a clock is important only in identifying the time
at which a given delay was measured. Accuracy, per se, has no at which a given delay was measured. Accuracy, per se, has no
importance to the accuracy of the measurement of delay. When importance to the accuracy of the measurement of delay. When
computing delays, we are interested only in the differences computing delays, we are interested only in the differences
between clock values, not the values themselves. between clock values, not the values themselves.
+ The resolution of a clock adds to uncertainty about any time + The resolution of a clock adds to uncertainty about any time
measured with it. Thus, if the source clock has a resolution of measured with it. Thus, if the source clock has a resolution of
10 msec, then this adds 10 msec of uncertainty to any time value 10 msec, then this adds 10 msec of uncertainty to any time value
measured with it. We will denote the resolution of the source measured with it. We will denote the resolution of the source
clock and the dest clock as Rsource and Rdest, respectively. clock and the destination clock as Rsource and Rdest,
respectively.
+ The skew of a clock is not so much an additional issue as it is a + The skew of a clock is not so much an additional issue as it is a
realization of the fact that Tsynch is itself a function of time. realization of the fact that Tsynch is itself a function of time.
Thus, if we attempt to measure or to bound Tsynch, this needs to Thus, if we attempt to measure or to bound Tsynch, this needs to
be done periodically. Over some periods of time, this function be done periodically. Over some periods of time, this function
can be approximated as a linear function plus some higher order can be approximated as a linear function plus some higher order
terms; in these cases, one option is to use knowledge of the terms; in these cases, one option is to use knowledge of the
linear component to correct the clock. Using this correction, the linear component to correct the clock. Using this correction, the
residual Tsynch is made smaller, but remains a source of residual Tsynch is made smaller, but remains a source of
uncertainty that must be accounted for. We use the function uncertainty that must be accounted for. We use the function
Esynch(t) to denote an upper bound on the uncertainty in Esynch(t) to denote an upper bound on the uncertainty in
synchronization. Thus, |Tsynch(t)| <= Esynch(t). synchronization. Thus, |Tsynch(t)| <= Esynch(t).
Taking these items together, we note that naive computation Tdest- Taking these items together, we note that naive computation Tdest-
Tsource will be off by Tsynch(t) +/- (|Rsource|+|Rdest|). Using the Tsource will be off by Tsynch(t) +/- (Rsource + Rdest). Using the
notion of Esynch(t), we note that these clock-related problems notion of Esynch(t), we note that these clock-related problems
introduce a total uncertainty of Esynch(t)+|Rsource|+|Rdest|. This introduce a total uncertainty of Esynch(t)+ Rsource + Rdest. This
estimate of total clock-related uncertainty should be included in the estimate of total clock-related uncertainty should be included in the
error/uncertainty analysis of any measurement implementation. error/uncertainty analysis of any measurement implementation.
3.7.2. Errors or uncertainties related to Wire-time vs Host-time 3.7.2. Errors or uncertainties related to Wire-time vs Host-time
As we have defined one-way delay, we would like to measure the time As we have defined one-way delay, we would like to measure the time
between when the test packet leaves the network interface of Src and between when the test packet leaves the network interface of Src and
when it (completely) arrives at the network interface of Dst, and we when it (completely) arrives at the network interface of Dst, and we
refer to this as 'wire time'. If the timings are themselves refer to these as "wire times." If the timings are themselves
performed by software on Src and Dst, however, then this software can performed by software on Src and Dst, however, then this software can
only directly measure the time between when Src grabs a timestamp only directly measure the time between when Src grabs a timestamp
just prior to sending the test packet and when Dst grabs a timestamp just prior to sending the test packet and when Dst grabs a timestamp
just after having received the test packet, and we refer to this as just after having received the test packet, and we refer to these two
'host time'. points as "host times".
To the extent that the difference between wire time and host time is To the extent that the difference between wire time and host time is
accurately known, this knowledge can be used to correct for host time accurately known, this knowledge can be used to correct for host time
measurements and the corrected value more accurately estimates the measurements and the corrected value more accurately estimates the
desired (wire time) metric. desired (wire time) metric.
To the extent, however, that the difference between wire time and To the extent, however, that the difference between wire time and
host time is uncertain, this uncertainty must be accounted for in an host time is uncertain, this uncertainty must be accounted for in an
analysis of a given measurement method. We denote by Hsource an analysis of a given measurement method. We denote by Hsource an
upper bound on the uncertainty in the difference between wire time upper bound on the uncertainty in the difference between wire time
skipping to change at page 9, line 4 skipping to change at page 9, line 34
measured value = true value + systematic error + random error measured value = true value + systematic error + random error
If the systematic error (the constant bias in measured values) can be If the systematic error (the constant bias in measured values) can be
determined, it can be compensated for in the reported results. determined, it can be compensated for in the reported results.
reported value = measured value - systematic error reported value = measured value - systematic error
therefore therefore
reported value = true value + random error reported value = true value + random error
The goal of calibration is to determine the systematic and random The goal of calibration is to determine the systematic and random
error in as much detail as possible. At a minimum, a bound ("e") error generated by the instruments themselves in as much detail as
should be found such that the reported value is in the range (true possible. At a minimum, a bound ("e") should be found such that the
value - e) to (true value + e) at least 95 percent of the time. We reported value is in the range (true value - e) to (true value + e)
call "e" the error bar for the measurements. {Comment: 95 percent at least 95 percent of the time. We call "e" the calibration error
was chosen because (1) some confidence level is desirable to be able for the measurements. It represents the degree to which the values
to remove outliers which will be found in measuring any physical produced by the measurement instrument are repeatable; that is, how
property; (2) a particular confidence level should be specified so closely an actual delay of 30 ms is reported as 30 ms. {Comment: 95
that the results of independent implementations can be compared; and percent was chosen because (1) some confidence level is desirable to
(3) even with a prototype user-level implementation, 95% was loose be able to remove outliers which will be found in measuring any
enough to exclude outliers.} physical property; (2) a particular confidence level should be
specified so that the results of independent implementations can be
compared; and (3) even with a prototype user-level implementation,
95% was loose enough to exclude outliers.}
From the discussion in the previous two sections, the error in From the discussion in the previous two sections, the error in
measurements could be bounded by determining all the individual measurements could be bounded by determining all the individual
uncertainties, and adding them together to form uncertainties, and adding them together to form
Esynch(t) + |Rsource| + |Rdest| + Hsource + Hdest. Esynch(t) + Rsource + Rdest + Hsource + Hdest.
However, reasonable bounds on both the clock-related uncertainty However, reasonable bounds on both the clock-related uncertainty
captured by the first three terms and the host-related uncertainty captured by the first three terms and the host-related uncertainty
captured by the last two terms should be possible by careful design captured by the last two terms should be possible by careful design
techniques and calibrating the instruments using a known, isolated, techniques and calibrating the instruments using a known, isolated,
network in a lab. network in a lab.
For example, the clock-related uncertainties are greatly reduced For example, the clock-related uncertainties are greatly reduced
through the use of a GPS time source. The sum of Esynch(t) + through the use of a GPS time source. The sum of Esynch(t) + Rsource
|Rsource| + |Rdest| is small, and is also bounded for the duration of + Rdest is small, and is also bounded for the duration of the
the measurement because of the global time source. measurement because of the global time source.
The host-related uncertainties, Hsource + Hdest, could be bounded by The host-related uncertainties, Hsource + Hdest, could be bounded by
connecting two instruments back-to-back with a high-speed serial link connecting two instruments back-to-back with a high-speed serial link
or isolated LAN (depending on the intended network connection for or isolated LAN segment. In this case, repeated measurements are
actual measurement), and performing repeated measurements. In this measuring the same one-way delay.
case, unlike measuring live networks, repeated measurements are
measuring the same wire time. (When measuring live networks, the
wire time is what you are measuring, and varies with the load
encountered on the path traversed by the test packets.)
If the test packets are small, such a network connection has a If the test packets are small, such a network connection has a
minimal wire time that may be approximated by zero. The measured minimal delay that may be approximated by zero. The measured delay
delay therefore contains only systematic and random error in the therefore contains only systematic and random error in the
instrumentation. The "average value" of repeated measurements is the instrumentation. The "average value" of repeated measurements is the
systematic error, and the variation is the random error. systematic error, and the variation is the random error.
One way to compute the systematic error, and the random error to a One way to compute the systematic error, and the random error to a
95% confidence is to repeat the experiment many times - at least 95% confidence is to repeat the experiment many times - at least
hundreds of tests. The systematic error would then be the median, hundreds of tests. The systematic error would then be the median.
and likely the mode (the most frequently occuring value). {Comment: The random error could then be found by removing the systematic error
It's likely the systematic error is represented by the minimum value from the measured values. The 95% confidence interval would be the
(which is also the median and the mode); with unloaded instruments on range from the 2.5th percentile to the 97.5th percentile of these
a single test path all the random error will tend to be increased deviations from the true value. The calibration error "e" could then
time due to host processing. The only error resulting an a delay be taken to be the largest absolute value of these two numbers, plus
less than the systematic error would be due to clock-related the clock-related uncertainty. {Comment: as described, this bound is
uncertainties (resolution and relative skew).} The random error could relatively loose since the uncertainties are added, and the absolute
then be found by removing the systematic error from the measured value of the largest deviation is used. As long as the resulting
values. The 95% confidence interval would be the range from the 2nd value is not a significant fraction of the measured values, it is a
percentile to the 97th percentile of these deviations from the true reasonable bound. If the resulting value is a significant fraction
value. The error bar "e" could then be taken to be the largest of the measured values, then more exact methods will be needed to
absolute value of these two numbers, plus the clock-related compute the calibration error.}
uncertainty. If all of the deviations are positive, then the 95%
confidence interval is simply the 95th percentile, and that value
should be used instead of the larger of the 2nd and 97th percentiles.
{Comment: as described, this bound is relatively loose since the
uncertainties are added, and the absolute value of the largest
deviation is used. As long as the resulting value is not a
significant fraction of the measured values, it is a reasonable
bound. If the resulting value is a significant fraction of the
measured values, then more exact methods will be needed to compute an
error bar.}
Note that random error is a function of measurement load. For Note that random error is a function of measurement load. For
example, if many paths will be measured by one instrument, this might example, if many paths will be measured by one instrument, this might
increase interrupts, process scheduling, and disk I/O (for example, increase interrupts, process scheduling, and disk I/O (for example,
recording the measurements), all of which may increase the random recording the measurements), all of which may increase the random
error in measured singletons. Therefore, in addition to minimal load error in measured singletons. Therefore, in addition to minimal load
measurements to find the systematic error, calibration measurements measurements to find the systematic error, calibration measurements
should be performed with the same measurement load that the should be performed with the same measurement load that the
instruments will see in the field. instruments will see in the field.
We wish to reiterate that this statistical treatment refers to the
calibration of the instrument; it is used to "calibrate the meter
stick" and say how well the meter stick reflects reality.
In addition to calibrating the instruments for finite one-way delay, In addition to calibrating the instruments for finite one-way delay,
two checks should be made to ensure that packets reported as losses two checks should be made to ensure that packets reported as losses
were really lost. First, the threshold for loss should be verified. were really lost. First, the threshold for loss should be verified.
In particular, ensure the "reasonable" threshold is reasonable: that In particular, ensure the "reasonable" threshold is reasonable: that
it is very unlikely a packet will arrive after the threshold value, it is very unlikely a packet will arrive after the threshold value,
and therefore the number of packets lost over an interval is not and therefore the number of packets lost over an interval is not
sensitive to the error bound on measurements. Second, consider the sensitive to the error bound on measurements. Second, consider the
probability that a packet arrives at the network interface, but is possibility that a packet arrives at the network interface, but is
lost due to congestion on that interface or to other resource lost due to congestion on that interface or to other resource
exhaustion (e.g. buffers) in the instrument. exhaustion (e.g. buffers) in the instrument.
3.8. Reporting the metric: 3.8. Reporting the metric:
The calibration and context in which the metric is measured must be The calibration and context in which the metric is measured MUST be
carefully considered, and should always be reported along with metric carefully considered, and SHOULD always be reported along with metric
results. We now present four items to consider: the Type-P of test results. We now present four items to consider: the Type-P of test
packets, the threshold of infinite delay (if any), error calibration, packets, the threshold of infinite delay (if any), error calibration,
and the path traversed by the test packets. This list is not and the path traversed by the test packets. This list is not
exhaustive; any additional information that could be useful in exhaustive; any additional information that could be useful in
interpreting applications of the metrics should also be reported. interpreting applications of the metrics should also be reported.
3.8.1. Type-P 3.8.1. Type-P
As noted in the Framework document [1], the value of the metric may As noted in the Framework document [1], the value of the metric may
depend on the type of IP packets used to make the measurement, or depend on the type of IP packets used to make the measurement, or
"type-P". The value of Type-P-One-way-Delay could change if the "type-P". The value of Type-P-One-way-Delay could change if the
protocol (UDP or TCP), port number, size, or arrangement for special protocol (UDP or TCP), port number, size, or arrangement for special
treatment (e.g., IP precedence or RSVP) changes. The exact Type-P treatment (e.g., IP precedence or RSVP) changes. The exact Type-P
used to make the measurements must be accurately reported. used to make the measurements MUST be accurately reported.
3.8.2. Loss threshold 3.8.2. Loss threshold
In addition, the threshold (or methodology to distinguish) between a In addition, the threshold (or methodology to distinguish) between a
large finite delay and loss should be reported. large finite delay and loss MUST be reported.
3.8.3. Calibration results 3.8.3. Calibration results
+ If the systematic error can be determined, it should be removed + If the systematic error can be determined, it SHOULD be removed
from the measured values. from the measured values.
+ Report an error bar, e, such that the true value is the reported + You SHOULD also report the calibration error, e, such that the
value plus or minus e, with 95% confidence. true value is the reported value plus or minus e, with 95%
confidence (see the last section.)
+ If possible, report the probability that a test packet with finite + If possible, the conditions under which a test packet with finite
delay is reported as lost due to resource exhaustion on the delay is reported as lost due to resource exhaustion on the
measurement instrument. measurement instrument SHOULD be reported.
3.8.4. Path 3.8.4. Path
Finally, the path traversed by the packet should be reported, if Finally, the path traversed by the packet SHOULD be reported, if
possible. In general it is impractical to know the precise path a possible. In general it is impractical to know the precise path a
given packet takes through the network. The precise path may be given packet takes through the network. The precise path may be
known for certain Type-P on short or stable paths. If Type-P known for certain Type-P on short or stable paths. If Type-P
includes the record route (or loose-source route) option in the IP includes the record route (or loose-source route) option in the IP
header, and the path is short enough, and all routers* on the path header, and the path is short enough, and all routers* on the path
support record (or loose-source) route, then the path will be support record (or loose-source) route, then the path will be
precisely recorded. This is impractical because the route must be precisely recorded. This is impractical because the route must be
short enough, many routers do not support (or are not configured for) short enough, many routers do not support (or are not configured for)
record route, and use of this feature would often artificially worsen record route, and use of this feature would often artificially worsen
the performance observed by removing the packet from common-case the performance observed by removing the packet from common-case
processing. However, partial information is still valuable context. processing. However, partial information is still valuable context.
For example, if a host can choose between two links* (and hence two For example, if a host can choose between two links* (and hence two
separate routes from src to dst), then the initial link used is separate routes from Src to Dst), then the initial link used is
valuable context. {Comment: For example, with Merit's NetNow setup, valuable context. {Comment: For example, with Merit's NetNow setup,
a Src on one NAP can reach a Dst on another NAP by either of several a Src on one NAP can reach a Dst on another NAP by either of several
different backbone networks.} different backbone networks.}
4. A Definition for Samples of One-way Delay 4. A Definition for Samples of One-way Delay
Given the singleton metric Type-P-One-way-Delay, we now define one Given the singleton metric Type-P-One-way-Delay, we now define one
particular sample of such singletons. The idea of the sample is to particular sample of such singletons. The idea of the sample is to
select a particular binding of the parameters Src, Dst, and Type-P, select a particular binding of the parameters Src, Dst, and Type-P,
then define a sample of values of parameter T. The means for then define a sample of values of parameter T. The means for
defining the values of T is to select a beginning time T0, a final defining the values of T is to select a beginning time T0, a final
time Tf, and an average rate lambda, then define a pseudo-random time Tf, and an average rate lambda, then define a pseudo-random
Poisson arrival process of rate lambda, whose values fall between T0 Poisson process of rate lambda, whose values fall between T0 and Tf.
and Tf. The time interval between successive values of T will then The time interval between successive values of T will then average
average 1/lambda. 1/lambda.
{Comment: Note that Poisson sampling is only one way of defining a
sample. Poisson has the advantage of limiting bias, but other
methods of sampling might be appropriate for different situations.
We encourage others who find such appropriate cases to use this
general framework and submit their sampling method for
standardization.}
4.1. Metric Name: 4.1. Metric Name:
Type-P-One-way-Delay-Poisson-Stream Type-P-One-way-Delay-Poisson-Stream
4.2. Metric Parameters: 4.2. Metric Parameters:
+ Src, the IP address of a host + Src, the IP address of a host
+ Dst, the IP address of a host + Dst, the IP address of a host
skipping to change at page 13, line 31 skipping to change at page 14, line 11
beginning at or before T0, with average arrival rate lambda, and beginning at or before T0, with average arrival rate lambda, and
ending at or after Tf. Those time values greater than or equal to T0 ending at or after Tf. Those time values greater than or equal to T0
and less than or equal to Tf are then selected. At each of the times and less than or equal to Tf are then selected. At each of the times
in this process, we obtain the value of Type-P-One-way-Delay at this in this process, we obtain the value of Type-P-One-way-Delay at this
time. The value of the sample is the sequence made up of the time. The value of the sample is the sequence made up of the
resulting <time, delay> pairs. If there are no such pairs, the resulting <time, delay> pairs. If there are no such pairs, the
sequence is of length zero and the sample is said to be empty. sequence is of length zero and the sample is said to be empty.
4.5. Discussion: 4.5. Discussion:
Note first that, since a pseudo-random number sequence is employed, The reader should be familiar with the in-depth discussion of Poisson
the sequence of times, and hence the value of the sample, is not sampling in the Framework document [1], which includes methods to
fully specified. Pseudo-random number generators of good quality compute and verify the pseudo-random Poisson process.
will be needed to achieve the desired qualities.
We specifically do not constrain the value of lambda, except to note
the extremes. If the rate is too large, then the measurement traffic
will perturb the network, and itself cause congestion. If the rate
is too small, then you might not capture interesting network
behavior. {Comment: We expect to document our experiences with, and
suggestions for, lambda elsewhere, culminating in a "best current
practices" document.}
Since a pseudo-random number sequence is employed, the sequence of
times, and hence the value of the sample, is not fully specified.
Pseudo-random number generators of good quality will be needed to
achieve the desired qualities.
The sample is defined in terms of a Poisson process both to avoid the The sample is defined in terms of a Poisson process both to avoid the
effects of self-synchronization and also capture a sample that is effects of self-synchronization and also capture a sample that is
statistically as unbiased as possible. {Comment: there is, of statistically as unbiased as possible. {Comment: there is, of
course, no claim that real Internet traffic arrives according to a course, no claim that real Internet traffic arrives according to a
Poisson arrival process.} Poisson arrival process.} The Poisson process is used to schedule
the delay measurements. The test packets will generally not arrive
at Dst according to a Poisson distribution, since they are influenced
by the network.
All the singleton Type-P-One-way-Delay metrics in the sequence will All the singleton Type-P-One-way-Delay metrics in the sequence will
have the same values of Src, Dst, and Type-P. have the same values of Src, Dst, and Type-P.
Note also that, given one sample that runs from T0 to Tf, and given Note also that, given one sample that runs from T0 to Tf, and given
new time values T0' and Tf' such that T0 <= T0' <= Tf' <= Tf, the new time values T0' and Tf' such that T0 <= T0' <= Tf' <= Tf, the
subsequence of the given sample whose time values fall between T0' subsequence of the given sample whose time values fall between T0'
and Tf' are also a valid Type-P-One-way-Delay-Poisson-Stream sample. and Tf' are also a valid Type-P-One-way-Delay-Poisson-Stream sample.
4.6. Methodologies: 4.6. Methodologies:
skipping to change at page 14, line 26 skipping to change at page 15, line 26
arrival of test packets; it is possible that the Src could send one arrival of test packets; it is possible that the Src could send one
test packet at TS[i], then send a second one (later) at TS[i+1], test packet at TS[i], then send a second one (later) at TS[i+1],
while the Dst could receive the second test packet at TR[i+1], and while the Dst could receive the second test packet at TR[i+1], and
then receive the first one (later) at TR[i]. then receive the first one (later) at TR[i].
4.7. Errors and Uncertainties: 4.7. Errors and Uncertainties:
In addition to sources of errors and uncertainties associated with In addition to sources of errors and uncertainties associated with
methods employed to measure the singleton values that make up the methods employed to measure the singleton values that make up the
sample, care must be given to analyze the accuracy of the Poisson sample, care must be given to analyze the accuracy of the Poisson
arrival process of the wire-time of the sending of the test packets. process with respect to the wire-times of the sending of the test
Problems with this process could be caused by several things, packets. Problems with this process could be caused by several
including problems with the pseudo-random number techniques used to things, including problems with the pseudo-random number techniques
generate the Poisson arrival process, or with jitter in the value of used to generate the Poisson arrival process, or with jitter in the
Hsource (mentioned above as uncertainty in the singleton delay value of Hsource (mentioned above as uncertainty in the singleton
metric). The Framework document shows how to use the Anderson- delay metric). The Framework document shows how to use the Anderson-
Darling test to verify the accuracy of the Poisson process. Darling test to verify the accuracy of a Poisson process over small
time frames. {Comment: The goal is to ensure that test packets are
sent "close enough" to a Poisson schedule, and avoid periodic
behavior.}
4.8. Reporting the metric: 4.8. Reporting the metric:
You should report the calibration and context for the underlying You MUST report the calibration and context for the underlying
singletons along with the stream. (See "Reporting the metric" for singletons along with the stream. (See "Reporting the metric" for
Type-P-One-way-Delay.) Type-P-One-way-Delay.)
5. Some Statistics Definitions for One-way Delay 5. Some Statistics Definitions for One-way Delay
Given the sample metric Type-P-One-way-Delay-Poisson-Stream, we now Given the sample metric Type-P-One-way-Delay-Poisson-Stream, we now
offer several statistics of that sample. These statistics are offer several statistics of that sample. These statistics are
offered mostly to be illustrative of what could be done. offered mostly to be illustrative of what could be done.
5.1. Type-P-One-way-Delay-Percentile 5.1. Type-P-One-way-Delay-Percentile
skipping to change at page 15, line 25 skipping to change at page 16, line 31
Stream1 = < Stream1 = <
<T1, 100 msec> <T1, 100 msec>
<T2, 110 msec> <T2, 110 msec>
<T3, undefined> <T3, undefined>
<T4, 90 msec> <T4, 90 msec>
<T5, 500 msec> <T5, 500 msec>
> >
Then the 50th percentile would be 110 msec, since 90 msec and 100 Then the 50th percentile would be 110 msec, since 90 msec and 100
msec are smaller and 110 msec and 'undefined' are larger. msec are smaller and 110 msec and 'undefined' are larger.
Note that if the probability that a finite packet is reported as lost Note that if the possibility that a packet with finite delay is
is significant, then a high percentile (90th or 95th) might be reported as lost is significant, then a high percentile (90th or
reported as infinite instead of finite. 95th) might be reported as infinite instead of finite.
5.2. Type-P-One-way-Delay-Median 5.2. Type-P-One-way-Delay-Median
Given a Type-P-One-way-Delay-Poisson-Stream, the median of all the dT Given a Type-P-One-way-Delay-Poisson-Stream, the median of all the dT
values in the Stream. In computing the median, undefined values are values in the Stream. In computing the median, undefined values are
treated as infinitely large. treated as infinitely large. As with Type-P-One-way-Delay-
Percentile, Type-P-One-way-Delay-Median is undefined if the sample is
empty.
As noted in the Framework document, the median differs from the 50th As noted in the Framework document, the median differs from the 50th
percentile only when the sample contains an even number of values, in percentile only when the sample contains an even number of values, in
which case the mean of the two central values is used. which case the mean of the two central values is used.
Example: suppose we take a sample and the results are: Example: suppose we take a sample and the results are:
Stream2 = < Stream2 = <
<T1, 100 msec> <T1, 100 msec>
<T2, 110 msec> <T2, 110 msec>
<T3, undefined> <T3, undefined>
<T4, 90 msec> <T4, 90 msec>
> >
Then the median would be 105 msec, the mean of 100 msec and 110 msec, Then the median would be 105 msec, the mean of 100 msec and 110 msec,
the two central values. the two central values.
5.3. Type-P-One-way-Delay-Minumum 5.3. Type-P-One-way-Delay-Minimum
Given a Type-P-One-way-Delay-Poisson-Stream, the minimum of all the Given a Type-P-One-way-Delay-Poisson-Stream, the minimum of all the
dT values in the Stream. In computing this, undefined values are dT values in the Stream. In computing this, undefined values are
treated as infinitely large. Note that this means that the minimum treated as infinitely large. Note that this means that the minimum
could thus be undefined (informally, infinite) if all the dT values could thus be undefined (informally, infinite) if all the dT values
are undefined. In addition, the Type-P-One-way-Delay-Minimum is are undefined. In addition, the Type-P-One-way-Delay-Minimum is
undefined if the sample is empty. undefined if the sample is empty.
In the above example, the minimum would be 90 msec. In the above example, the minimum would be 90 msec.
5.4. Type-P-One-way-Delay-Inverse-Percentile 5.4. Type-P-One-way-Delay-Inverse-Percentile
Given a Type-P-One-way-Delay-Poisson-Stream and a non-negative time Given a Type-P-One-way-Delay-Poisson-Stream and a non-negative time
duration threshold, the fraction of all the dT values in the Stream duration threshold, the fraction of all the dT values in the Stream
less than or equal to the threshold. The result could be as low as less than or equal to the threshold. The result could be as low as
0% (if all the dT values exceed threshold) or as high as 100%. 0% (if all the dT values exceed threshold) or as high as 100%. Type-
P-One-way-Delay-Inverse-Percentile is undefined if the sample is
empty.
In the above example, the Inverse-Percentile of 103 msec would be In the above example, the Inverse-Percentile of 103 msec would be
50%. 50%.
6. Security Considerations 6. Security Considerations
Conducting Internet measurements raises both security and privacy Conducting Internet measurements raises both security and privacy
concerns. This memo does not specify an implementation of the concerns. This memo does not specify an implementation of the
metrics, so it does not directly affect the security of the Internet metrics, so it does not directly affect the security of the Internet
nor of applications which run on the Internet. However, nor of applications which run on the Internet. However,
implementations of these metrics must be mindful of security and implementations of these metrics must be mindful of security and
privacy concerns. privacy concerns.
There are two types of security concerns: potential harm caused by There are two types of security concerns: potential harm caused by
the measurements, and potential harm to the measurements. The the measurements, and potential harm to the measurements. The
measurements could cause harm because they are active, and inject measurements could cause harm because they are active, and inject
packets into the network. The measurement parameters must be packets into the network. The measurement parameters MUST be
carefully selected so that the measurements inject trivial amounts of carefully selected so that the measurements inject trivial amounts of
additional traffic into the networks they measure. If they inject additional traffic into the networks they measure. If they inject
"too much" traffic, they can skew the results of the measurement, and "too much" traffic, they can skew the results of the measurement, and
in extreme cases cause congestion and denial of service. in extreme cases cause congestion and denial of service.
The measurements themselves could be harmed by routers giving The measurements themselves could be harmed by routers giving
measurement traffic a different priority than "normal" traffic, or by measurement traffic a different priority than "normal" traffic, or by
an attacker injecting artificial measurement traffic. If routers can an attacker injecting artificial measurement traffic. If routers can
recognize measurement traffic and treat it separately, the recognize measurement traffic and treat it separately, the
measurements will not reflect actual user traffic. If an attacker measurements will not reflect actual user traffic. If an attacker
injects artificial traffic that is accepted as legitimate, the loss injects artificial traffic that is accepted as legitimate, the loss
rate will be artificially lowered. Therefore, the measurement rate will be artificially lowered. Therefore, the measurement
methodologies should include appropriate techniques to reduce the methodologies SHOULD include appropriate techniques to reduce the
probability measurement traffic can be distinguished from "normal" probability measurement traffic can be distinguished from "normal"
traffic. Authentication techniques, such as digital signatures, may traffic. Authentication techniques, such as digital signatures, may
be used where appropriate to guard against injected traffic attacks. be used where appropriate to guard against injected traffic attacks.
The privacy concerns of network measurement are limited by the active The privacy concerns of network measurement are limited by the active
measurements described in this memo. Unlike passive measurements, measurements described in this memo. Unlike passive measurements,
there can be no release of existing user data. there can be no release of existing user data.
7. Acknowledgements 7. Acknowledgements
Special thanks are due to Vern Paxson of Lawrence Berkeley Labs for Special thanks are due to Vern Paxson of Lawrence Berkeley Labs for
his helpful comments on issues of clock uncertainty and statistics. his helpful comments on issues of clock uncertainty and statistics.
Thanks also to Will Leland, Sean Shapira, and Roland Wittig for Thanks also to Will Leland, Andy Scherrer, Sean Shapira, and
several useful suggestions. Roland Wittig for several useful suggestions.
8. References 8. References
[1] V. Paxson, G. Almes, J. Mahdavi, and M. Mathis, "Framework for [1] V. Paxson, G. Almes, J. Mahdavi, and M. Mathis, "Framework for
IP Performance Metrics", RFC 2330, May 1998. IP Performance Metrics", RFC 2330, May 1998.
[2] G. Almes, S. Kalidindi, and M. Zekauskas, "A Packet Loss Metric [2] G. Almes, S. Kalidindi, and M. Zekauskas, "A Packet Loss Metric
for IPPM", Internet-Draft <draft-ietf-ippm-loss-04.txt>, August for IPPM", Internet-Draft <draft-ietf-ippm-loss-05.txt>, August
1998. 1998.
[3] D. Mills, "Network Time Protocol (v3)", RFC 1305, April 1992. [3] D. Mills, "Network Time Protocol (v3)", RFC 1305, April 1992.
[4] J. Mahdavi and V. Paxson, "IPPM Metrics for Measuring [4] J. Mahdavi and V. Paxson, "IPPM Metrics for Measuring
Connectivity", Internet-Draft <draft-ietf-ippm- Connectivity", Internet-Draft <draft-ietf-ippm-
connectivity-02.txt>, August 1998. connectivity-03.txt>, October 1998.
[5] J. Postel, "Internet Protocol", RFC 791, September 1981. [5] J. Postel, "Internet Protocol", RFC 791, September 1981.
[6] S. Bradner, "Key words for use in RFCs to Indicate Requirement
Levels", RFC 2119, March 1997.
9. Authors' Addresses 9. Authors' Addresses
Guy Almes Guy Almes
Advanced Network & Services, Inc. Advanced Network & Services, Inc.
200 Business Park Drive 200 Business Park Drive
Armonk, NY 10504 Armonk, NY 10504
USA USA
Phone: +1 914 765 1120 Phone: +1 914 765 1120
EMail: almes@advanced.org EMail: almes@advanced.org
Sunil Kalidindi Sunil Kalidindi
skipping to change at page 18, line 31 skipping to change at page 19, line 34
Matthew J. Zekauskas Matthew J. Zekauskas
Advanced Network & Services, Inc. Advanced Network & Services, Inc.
200 Buisiness Park Drive 200 Buisiness Park Drive
Armonk, NY 10504 Armonk, NY 10504
USA USA
Phone: +1 914 765 1112 Phone: +1 914 765 1112
EMail: matt@advanced.org EMail: matt@advanced.org
Expiration date: March, 1999 Expiration date: May, 1999
 End of changes. 

This html diff was produced by rfcdiff 1.23, available from http://www.levkowetz.com/ietf/tools/rfcdiff/