< draft-mathis-ippm-model-based-metrics-00.txt   draft-mathis-ippm-model-based-metrics-01.txt >
IP Performance Working Group M. Mathis IP Performance Working Group M. Mathis
Internet-Draft Google, Inc Internet-Draft Google, Inc
Intended status: Experimental Oct 15, 2012 Intended status: Experimental A. Morton
Expires: April 18, 2013 Expires: August 29, 2013 AT&T Labs
Feb 25, 2013
Model Based Internet Performance Metrics Model Based Internet Performance Metrics
draft-mathis-ippm-model-based-metrics-00.txt draft-mathis-ippm-model-based-metrics-01.txt
Abstract Abstract
We introduce a new class of model based Internet metrics designed to We introduce a new class of model based metrics designed to determine
determine if a long path can be expected to meet a predefined end-to- if a long path can meet predefined end-to-end application performance
end application performance target by applying a suite of single targets. This is done by section-by-section testing -- by applying a
property tests to successive sections of the long path. In many suite of single property tests to successive sections of a long path.
cases these single property tests are based on existing IPPM metrics, In many cases these single property tests are based on existing IPPM
with the addition of success and validity criteria. The sub-path at metrics, with the addition of success and validity criteria. The
a time tests are designed to eliminate all known conditions that sub-path at a time tests are designed to facilitate IP providers
might prevent the full path from meeting the target performance. eliminating all known conditions that might prevent the full end-to-
end path from meeting the users target performance.
This approach makes it possible to to determine the IP performance
requirements needed to support the desired end-to-end TCP
performance. The IP metrics are based on traffic patterns that mimic
TCP but are precomputed independently of the actual behavior of TCP
over the sub-path under test. This makes the measurements open loop,
eliminating nearly all of the difficulties encountered by traditional
bulk transport metrics, which rely on congestion control equilibrium
behavior.
A natural consequence of this methodology is verifiable network
measurement: measurements from any given vantage point are repeatable
from other vantage points.
Status of this Memo Status of this Memo
This Internet-Draft is submitted in full conformance with the This Internet-Draft is submitted in full conformance with the
provisions of BCP 78 and BCP 79. provisions of BCP 78 and BCP 79.
Internet-Drafts are working documents of the Internet Engineering Internet-Drafts are working documents of the Internet Engineering
Task Force (IETF). Note that other groups may also distribute Task Force (IETF). Note that other groups may also distribute
working documents as Internet-Drafts. The list of current Internet- working documents as Internet-Drafts. The list of current Internet-
Drafts is at http://datatracker.ietf.org/drafts/current/. Drafts is at http://datatracker.ietf.org/drafts/current/.
Internet-Drafts are draft documents valid for a maximum of six months Internet-Drafts are draft documents valid for a maximum of six months
and may be updated, replaced, or obsoleted by other documents at any and may be updated, replaced, or obsoleted by other documents at any
time. It is inappropriate to use Internet-Drafts as reference time. It is inappropriate to use Internet-Drafts as reference
material or to cite them other than as "work in progress." material or to cite them other than as "work in progress."
This Internet-Draft will expire on April 18, 2013. This Internet-Draft will expire on August 29, 2013.
Copyright Notice Copyright Notice
Copyright (c) 2012 IETF Trust and the persons identified as the Copyright (c) 2013 IETF Trust and the persons identified as the
document authors. All rights reserved. document authors. All rights reserved.
This document is subject to BCP 78 and the IETF Trust's Legal This document is subject to BCP 78 and the IETF Trust's Legal
Provisions Relating to IETF Documents Provisions Relating to IETF Documents
(http://trustee.ietf.org/license-info) in effect on the date of (http://trustee.ietf.org/license-info) in effect on the date of
publication of this document. Please review these documents publication of this document. Please review these documents
carefully, as they describe your rights and restrictions with respect carefully, as they describe your rights and restrictions with respect
to this document. Code Components extracted from this document must to this document. Code Components extracted from this document must
include Simplified BSD License text as described in Section 4.e of include Simplified BSD License text as described in Section 4.e of
the Trust Legal Provisions and are provided without warranty as the Trust Legal Provisions and are provided without warranty as
described in the Simplified BSD License. described in the Simplified BSD License.
Table of Contents Table of Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 5
2. Background . . . . . . . . . . . . . . . . . . . . . . . . . . 4 2. New requirements relative to RFC 2330 . . . . . . . . . . . . 7
3. Common Models and Parameters . . . . . . . . . . . . . . . . . 5 3. Background . . . . . . . . . . . . . . . . . . . . . . . . . . 7
3.1. End-to-end parameters . . . . . . . . . . . . . . . . . . 5 4. Common Models and Parameters . . . . . . . . . . . . . . . . . 9
3.2. Per sub-path parameters . . . . . . . . . . . . . . . . . 7 4.1. Target End-to-end parameters . . . . . . . . . . . . . . . 9
3.3. Common Calculations for Single Property Tests . . . . . . 7 4.2. End-to-end parameters from sub-paths . . . . . . . . . . . 11
3.4. Parameter Derating . . . . . . . . . . . . . . . . . . . . 8 4.3. Per sub-path parameters . . . . . . . . . . . . . . . . . 11
3.5. Single Property Tests Results . . . . . . . . . . . . . . 9 4.4. Common Calculations for Single Property Tests . . . . . . 11
4. Single Property Tests . . . . . . . . . . . . . . . . . . . . 9 4.5. Parameter Derating . . . . . . . . . . . . . . . . . . . . 13
4.1. Verify the absence of cross traffic . . . . . . . . . . . 9 5. Common testing procedures . . . . . . . . . . . . . . . . . . 13
4.1.1. Parameter Calculation . . . . . . . . . . . . . . . . 10 5.1. Traffic generating techniques . . . . . . . . . . . . . . 13
4.1.2. Cross traffic Measurement . . . . . . . . . . . . . . 10 5.1.1. Paced transmission . . . . . . . . . . . . . . . . . . 14
4.2. Full Data Rate Loss Rate Tests . . . . . . . . . . . . . . 10 5.1.2. Constant window pseudo CBR . . . . . . . . . . . . . . 14
4.2.1. Loss Rate Measurement . . . . . . . . . . . . . . . . 11 5.1.3. Scanned window pseudo CBR . . . . . . . . . . . . . . 14
4.3. Background Loss Rate Tests . . . . . . . . . . . . . . . . 11 5.1.4. Intermittent Testing . . . . . . . . . . . . . . . . . 15
4.3.1. Background Loss Rate Measurement . . . . . . . . . . . 11 5.1.5. Intermittent Scatter Testing . . . . . . . . . . . . . 15
4.4. Queue Capacity Test . . . . . . . . . . . . . . . . . . . 12 5.2. Interpreting the Results . . . . . . . . . . . . . . . . . 15
4.4.1. Model Calculation . . . . . . . . . . . . . . . . . . 12 5.2.1. Inconclusive test outcomes . . . . . . . . . . . . . . 15
4.4.2. Queue Capacity Measurement . . . . . . . . . . . . . . 12 5.2.2. Statistical criteria for measuring run_length . . . . 16
4.5. AQM Test . . . . . . . . . . . . . . . . . . . . . . . . . 12 5.2.3. Classifications of tests . . . . . . . . . . . . . . . 17
4.5.1. Model Calculation . . . . . . . . . . . . . . . . . . 12 5.3. Reordering Tolerance . . . . . . . . . . . . . . . . . . . 18
4.5.2. AQM Measurement . . . . . . . . . . . . . . . . . . . 12 5.4. Verify the absence of cross traffic . . . . . . . . . . . 18
4.6. Reordering Test . . . . . . . . . . . . . . . . . . . . . 12 5.5. Additional test preconditions . . . . . . . . . . . . . . 19
4.6.1. Model Calculation . . . . . . . . . . . . . . . . . . 12 6. Single Property Tests . . . . . . . . . . . . . . . . . . . . 20
4.6.2. Reordering Measurement . . . . . . . . . . . . . . . . 12 6.1. CBR Tests . . . . . . . . . . . . . . . . . . . . . . . . 21
5. Calibration . . . . . . . . . . . . . . . . . . . . . . . . . 12 6.1.1. Loss Rate at Full Data Rate . . . . . . . . . . . . . 21
6. References . . . . . . . . . . . . . . . . . . . . . . . . . . 13 6.1.2. Background Loss Rate Tests . . . . . . . . . . . . . . 21
6.1. Normative References . . . . . . . . . . . . . . . . . . . 13 6.2. Standing Queue tests . . . . . . . . . . . . . . . . . . . 22
6.2. Informative References . . . . . . . . . . . . . . . . . . 13 6.2.1. Congestion Avoidance . . . . . . . . . . . . . . . . . 22
Appendix A. Model Derivations . . . . . . . . . . . . . . . . . . 13 6.2.2. Buffer Bloat . . . . . . . . . . . . . . . . . . . . . 23
Appendix B. Old text from an earlier document . . . . . . . . . . 13 6.2.3. Self Interference . . . . . . . . . . . . . . . . . . 23
Author's Address . . . . . . . . . . . . . . . . . . . . . . . . . 15 6.3. Slow Start tests . . . . . . . . . . . . . . . . . . . . . 23
6.3.1. Full Window slow start test . . . . . . . . . . . . . 23
6.3.2. Slowstart AQM test . . . . . . . . . . . . . . . . . . 24
6.4. Server Rate tests . . . . . . . . . . . . . . . . . . . . 24
6.4.1. Server TCP Send Offload (TSO) tests . . . . . . . . . 24
6.4.2. Server Full Window test . . . . . . . . . . . . . . . 24
7. Combined Tests . . . . . . . . . . . . . . . . . . . . . . . . 24
7.1. Sustained burst test . . . . . . . . . . . . . . . . . . . 25
8. Calibration . . . . . . . . . . . . . . . . . . . . . . . . . 26
9. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 26
10. References . . . . . . . . . . . . . . . . . . . . . . . . . . 26
10.1. Normative References . . . . . . . . . . . . . . . . . . . 26
10.2. Informative References . . . . . . . . . . . . . . . . . . 26
Appendix A. Model Derivations . . . . . . . . . . . . . . . . . . 27
Appendix B. Old text from an earlier document . . . . . . . . . . 27
Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 29
1. Introduction 1. Introduction
We introduce a new class of model based metrics designed to determine We introduce a new class of model based metrics designed to determine
if a long path can be expected to meet a predefined application end- if a long path can be expected to meet a predefined application end-
to-end performance target by applying a suite of single property to-end performance target by applying a suite of single property
tests to successive sections of the long path. In many cases these tests to successive sections of the long path. In many cases these
single property tests are based on existing IPPM metrics, with the single property tests are based on existing IPPM metrics, with the
addition of success and validity criteria. The sub-path at a time addition of specific success and validity criteria. The sub-path at
tests are designed to eliminate all known conditions that might a time tests are designed to eliminate all known conditions that will
prevent the full path from meeting the target performance. The end- potentially prevent the full path from meeting the target
to-end target performance must be specified in advance, in order to performance.
be able to open-loop the control systems (such as congestion control)
that are present in all Internet transport protocols and The end-to-end target performance must be specified in advance, and
applications. Since a singleton (see [RFC2330]) is only a pass/fail models are used to compute the IP layer properties necessary to
measurement of a sub-path, these metrics are most useful in support that performance. The IP metrics are based on traffic
composition over large pools of samples, such as across a collection patterns that mimic TCP but are precomputed independently of the
of paths or a time interval [RFC5835]. actual behavior of TCP over the sub-path under test.
This approach makes the measurements open loop, eliminating nearly
all of the difficulties encountered by traditional bulk transport
metrics, which depend on congestion control equilibrium behavior.
Otherwise these control systems inherently have a number of
properties that interfere with measurement: they have circular
dependencies such that every component affects every property.
Since a singleton (see [RFC2330]) is only a pass/fail measurement of
a sub-path, these metrics are most useful in composition over large
pools of samples, such as across a collection of paths or a time
interval [RFC5835] [RFC6049] .
For Bulk Transport Capacity (BTC) the target performance to be For Bulk Transport Capacity (BTC) the target performance to be
measured is a data rate. TCP's ability to compensate for less than confirmed is a data rate. TCP's ability to compensate for less than
ideal network conditions is fundamentally affected by the RTT and MTU ideal network conditions is fundamentally affected by the RTT and MTU
of the end-to-end Internet path that it traverses. Since the minimum of the end-to-end Internet path that it traverses. Since the minimum
RTT and maximum MTU are both fixed properties of the path, they are RTT and maximum MTU are both fixed properties of the path, they are
also taken as parameters. The target values for these three also taken as parameters to the modeling process. The target values
parameters, Data Rate, RTT and MTU, are determined by the for these three parameters, Data Rate, RTT and MTU, are determined by
application, its intended use and the physical infrastructure over the application, its intended use and the physical infrastructure
which it traverses. They are described in more detail in Section 3 over which it traverses.
together with the models used to infer the required performance of
the underlying Internet fabric. For BTC the following tests are sufficient:
o raw data rate,
o background loss rate,
o queue burst capacity,
o reordering extent [RFC4737],
o onset of congestion/AQM
o and corresponding metrics on return path quality.
If every sub-path passes all of these tests, then an end-to-end
application using any reasonably modern TCP or similar protocol
should be able to attain the specified target data rate, over the
full end-to-end path at the specified RTT and MTU.
Traditional end-to-end BTC metrics have proven to be difficult or Traditional end-to-end BTC metrics have proven to be difficult or
unsatisfactory for the reasons described in Section 2. Rather than unsatisfactory due to some overlooked requirements described in
testing the end-to-end path with TCP or other some other BTC, each Section 2 and some intrinsic difficulties with using protocols for
sub-path is evaluated using suite of far simpler and more predictable measurement described in Section 3. In Section 4 we describe the
single property tests described in Section 4. For BTC the following models and common parameters used to derive single property test
tests are sufficient: raw data rate, background loss rate, queue parameters. In Section 5 we describe common testing procedures used
burst capacity, reordering extent, onset of congestion/AQM and return by all of the tests. Rather than testing the end-to-end path with
path quality. If every sub-path passes all of these tests, then an TCP or other some other BTC, each sub-path is evaluated using suite
end-to-end application using any reasonably modern TCP or similar of far simpler and more predictable single property tests described
protocol should be able to attain the specified target data rate, in Section 6. Section 7 describes some combined tests that are more
over the full end-to-end path at the specified RTT and MTU. efficient to implement and deploy. However, if they fail they may
not clearly indicate the nature of the problem.
There exists the potential that model based metric might fail, in the There exists the potential that model based metric itself might yield
sense that every sub-path of an end-to-end path passes every single a false pass result, in the sense that every sub-path of an end-to-
property test and yet a application might still fall to attain its end path passes every single property test and yet a real application
performance target. If so, then a traditional BTC needs to be used might still fall to attain its performance target over the path. If
to validate the tests for each sub-path, as described in Section 5. so, then a traditional BTC needs to be used to validate the tests for
each sub-path, as described in Section 8.
Future text (or a more likely a future document) will describe model Future text (or a more likely a future document) will describe model
based metrics for real time traffic. The salient point will be that based metrics for real time traffic. The salient point will be that
concurrently meeting the goals of both RT and throughput maximizing concurrently meeting the goals of both RT and throughput maximizing
traffic implicitly requires some form of traffic segregation, such traffic implicitly requires some form of traffic segregation, such
that the two traffic classes are not placed in the same queue. Some that the two traffic classes are not placed in the same queue. Some
technique as simple as SFQ[SFQ] might be a sufficient alternative to technique as simple as Fair Queueing [SFQ] might be a sufficient
full QoS. alternative to full QoS.
TODO: TODO:
Add discussion of protocol overhead: MSS vs IP MTU vs link MTU o Better terminology for: single property test, test targets (as
opposed to end-to-end targets), packet layer properties(?),
testing suites, combined tests, etc. All to strengthen of the
linguistic differences between transport and network layer.
o Make it clear that this document is about traffic patterns and
delivery statistics. Other aspects of the test procedures are out
of scope.
o Add description of assumed TCP behaviors used to derived the
models.
o Eliminate sequentiality both as a modeling process and for section
by section testing. Treatment of link under test being different
from the bottleneck link (e.g. testing to an aggregation point).
2. Background o Add "effective bottleneck rate" as an end-to-end parameter.
Discussion of ACK compression and its intrinsic consequences.
o Tests for a given subpath can be designed w/o knowing the rest of
the path. Tests suites can be designed for link types in the
abstract and standardized independently. Add example "complete
test suites" following the combined tests.
o Add add concept of untargeted tests and algebra on loss rate.
o Make the background traffic test have an explicit procedure, and
clearly delineate between users background traffic and other
traffic. Connect preloading to intermittent and not intermittent,
and as a way to control radio power states. Note that nearly all
devices have some preloading effects (e.g. ARP on LANs)
o Better uniformity about: applies to all transport protocols, but
defined in terms of TCP parameters.
o Clean and uniform descriptions of all tests.
o Write model appendix. Deprecate "old doc" appendix.
(Fragments of earlier text) 2. New requirements relative to RFC 2330
The Model Based Metrics are designed to fulfil some additional
requirement that were not recognized at the time RFC 2330 was
written. These missing requirements may have significantly
contributed to policy difficulties in the IP measurement space. The
additional requirements are:
o Metrics must be actionable by the ISP - they have to be
interpreted in terms of behaviors or properties at the IP or lower
layers, that an ISP can test, repair and verify.
o Metrics must be vantage point invariant over a significant range
of measurement point choices (e.g., measurement points as
described in [I-D.morton-ippm-lmap-path]), including off path
measurement points. The only requirements on MP selection should
be that the portion of the path that is not under test is
effectively ideal (or is non ideal in calibratable ways) and the
end-to-end RTT between MPs is below some reasonable bound.
o Metrics must be repeatable by multiple parties. It must be
possible for different parties to make the same measurement and
observe the same results. In particular it is specifically
important that both a consumer (or their delegate) and ISP be able
to perform the same measurement and get the same result.
NB: all of the requirements for metrics in RFC 2330 should be
reviewed and potentially revised.
3. Background
The holy grail of IPPM has been BTC measurement, but it has proven to The holy grail of IPPM has been BTC measurement, but it has proven to
be a very hard problem for a number of reasons: be a very hard problem for a number of reasons:
TCP is a control systems - everything affects performance, o TCP is a control system with circular dependencies - everything
including components that are explicitly not part of the the test. affects performance, including components that are explicitly not
Congestion control is an equilibrium process, transport protocols part of the test.
o Congestion control is an equilibrium process, transport protocols
change the network (raise loss probability and/or RTT) to conform change the network (raise loss probability and/or RTT) to conform
to their behavior. to their behavior.
TCP's ability to compensate for network flaws is directly o TCP's ability to compensate for network flaws is directly
proportional to the number of round trips per second (e.g. proportional to the number of round trips per second (e.g.
inversely proportional to the RTT). As a consequence a flawed inversely proportional to the RTT). As a consequence a flawed
link that passes a local test is likely to completely fail when link that passes a local test is likely to completely fail when
the path is extended by a perfect network to some larger RTT. the path is extended by a perfect network to some larger RTT.
TCP has a meta Heisenberg problem - Measurement and cross traffic o TCP has a meta Heisenberg problem - Measurement and cross traffic
interact in unknown and ill defined ways. The situation is interact in unknown and ill defined ways. The situation is
actually worse than the traditional physics problem where you can actually worse than the traditional physics problem where you can
at least estimate the relative masses of the measurement and at least estimate the relative masses of the measurement and
measured particles. For network measurement you can not in measured particles. For network measurement you can not in
general determine the relative "masses" of the measurement traffic general determine the relative "masses" of the measurement traffic
and cross traffic, so you can not even gage the relative magnitude and cross traffic, so you can not even gage the relative magnitude
of their effects on each other. of their effects on each other.
The new approach is to "open loop" congestion control. Defeat CC, The new approach is to "open loop" mandatory congestion control
typically by throttling TCP to a lower rate, such that it does not algorithms, typically by throttling TCP (or other protocol) to a
respond to network conditions. In this mode the measurement software lower rate, such that it does not react to changing network
explicitly controls TCP's state variables (e.g. cwnd) to create conditions. In this approack the measurement software explicitly
controlled traffic patterns, which are manipulated to measure the controls the data rate, transmission pattern or cwnd (TCP's primary
network. congestion control state variables) to create repeatable traffic
patterns that mimic TCP behavior but are almost entirely independent
of the actual network behavior. These patterns are manipulated to
probe the network to verify that it can deliver all of the traffic
patterns that a transport protocol is likely to generate under normal
operation at the target rate and RTT.
Models are used to determine the actual test parameters (loss rate, Models are used to determine the actual test parameters (burst size,
etc) from the target parameters. The basic method is to use models loss rate, etc) from the target parameters. The basic method is to
to estimate simple network properties required to sustain a given use models to estimate specific network properties required to
transport flow (or set of flows), and using a suite of simpler sustain a given transport flow (or set of flows), and using a suite
metrics to confirm that the network meets the required properties. of simpler metrics to confirm that the network meets the required
For example a network can sustain a Bulk TCP flow of a given data properties. For example a network can sustain a Bulk TCP flow of a
rate, MTU and RTT when 4 (and probably more) conditions are met: given data rate, MTU and RTT when 4 (and probably more) conditions
The raw link rate is higher than the target data rate. are met:
The raw packet loss rate is lower than required by a suitable TCP o The raw link rate is higher than the target data rate.
o The raw packet loss rate is lower than required by a suitable TCP
performance model performance model
There is sufficient buffering at any bottleneck smooth bursts. o There is sufficient buffering at any bottleneck smooth bursts.
When the link is overfilled (congested), the onset of packet loss o When the link is overfilled (congested), the onset of packet loss
is progressive. is progressive.
These condition can all be verified with simple tests, using model These condition can all be verified with simple tests, using model
parameters and acceptance thresholds derived from the target data parameters and acceptance thresholds derived from the target data
rate, MTU and RTT. Note that this procedure is not invertible: a rate, MTU and RTT. Note that this procedure is not invertible: a
singleton measurement is a pass/fail evaluation of a given path or singleton measurement is a pass/fail evaluation of a given path or
subpath at a given performance. Measurements to confirm that a link subpath at a given performance. Measurements to confirm that a link
passes at one particular performance may not be generally be useful passes at one particular performance may not be generally be useful
to predict if the link will pass at a different performance. to predict if the link will pass at a different performance.
Although they are not invertible, they do have several other valuable Although they are not invertible, they do have several other valuable
properties, such as natural ways to define several different properties, such as natural ways to define several different
composition metrics. composition metrics.
3. Common Models and Parameters [Add text on algebra on metrics (A-Frame) and tomography.] The
Spatial Composition of fundamental IPPM metrics has been studied and
standardized. For example, the algebra to combine empirical
assessments of loss ratio to estimate complete path performance is
described in section 5.1.5. of [RFC6049]. We intend to use this and
other composition metrics as necessary.
4. Common Models and Parameters
Transport performance models are used to derive the test parameters Transport performance models are used to derive the test parameters
for each single property test from the end-to-end target parameters for each single property test from the end-to-end target parameters
and additional ancillary parameters. and additional ancillary parameters.
It is envisioned that the modeling phase (to compute the test It is envisioned that the modeling phase (to compute the test
parameters) and testing phases will be decoupled. This section parameters) and testing phases will be decoupled. This section
covers common derived parameters, used by multiple single property covers common derived parameters, used by multiple single property
tests. For some tests, additional modeling is described with the tests. For some tests, additional modeling is described with the
tests. tests. MAKE THIS NON SEQUENTIAL
Since some aspects of the models are very conservative, the modeling Since some aspects of the models may be excessively conservative, the
framework permits some latitude in derating test parameters, as modeling framework permits some latitude in derating some test
described in Section 3.4. parameters, as described in Section 4.5.
For certain sub-paths (e.g. common types of access links) it would be For certain sub-paths (e.g. common types of access links) it would be
appropriate for the single property test parameters to be documented appropriate for the single property test parameters to be documented
as a "measurement profile" together with the modeling assumptions and as a "measurement profile" together with the modeling assumptions and
derating factors described in Section 3.3 and Section 3.4. derating factors described in Section 4.4 and Section 4.5.
3.1. End-to-end parameters 4.1. Target End-to-end parameters
These parameters are determined by the needs of the application or These parameters are determined by the needs of the application or
the ultimate end user and the end-to-end Internet path. the ultimate end user and the end-to-end Internet path. They are in
units that make sense to the upper layer: payload bytes delivered,
excluding header overheads for IP, TCP and other protocol.
Target Data Rate: The application or ultimate user's performance Target Data Rate: The application or ultimate user's performance
goal (in aggregate across all connections). goal.
Permitted Number of Connections: The target rate can be more easily
obtained by dividing the traffic across more than one connection.
In general the number of concurrent connections is determined by
the application, however see the comments below on multiple
connections.
Target RTT (Round Trip Time): For fundamental reasons a long path Target RTT (Round Trip Time): For fundamental reasons a long path
makes it more difficult for TCP or other transport protocol to makes it more difficult for TCP or other transport protocol to
meet the target rate. The target RTT must be representative for meet the target rate. The target RTT must be representative of
the actual applications expected to use the network. This the actual expected application use of the network. It may be
parameter may be subject to a future convention (e.g. continental subject to conventions about assumed application usage (e.g.
scale paths should be assumed to be some fixed RTT, such as 100 continental scale paths should be assumed to be some fixed RTT,
ms) or alternatively be an property of an ISP's topology (e.g. a such as 100 ms) or alternatively be an property of an ISP's
ISP with richer and better placed peering may actually have lower topology (e.g. a ISP with richer or better placed peering may be
RTTs for typical users.) able to justify assuming lower RTTs than other ISPs.)
Target MTU (Maximum Transmission Unit): Assume 1500 Bytes unless Target MTU (Maximum Transmission Unit): Assume 1500 Bytes per packet
otherwise specified. If some sub-path forces a smaller MTU, then unless otherwise specified. If some sub-path forces a smaller
all sub-paths must be tested with the same smaller MTU. MTU, then all sub-paths must be tested with the same smaller MTU.
Header overhead The IP and TCP header sizes, which are the portion
of each MTU not available for carrying application payload. This
is also assumed to be the size for returning acknowledgements
(ACKs). The payload Maximum Segment Size (MSS) is the Target MTU
minus header overhead.
Permitted Number of Connections: The target rate can be more easily
obtained by dividing the traffic across more than one connection.
(Ideally this would be 1). In general the number of concurrent
connections is determined by the application, however see the
comments below on multiple connections.
Effective Bottleneck Data Rate This is the bottleneck data rate that
might be inferred from the ACK stream, by looking at how much data
the ACK stream reports was delivered per unit time. For
traditional networks, the effective bottleneck rate would be the
same as the actual bottleneck rate. If the forward path is
subject AFD [AFD] style policing using a virtual queue, the
effective bottleneck rate would be the same as the actual physical
link rate, even though this rate is not available to the user.
For systems that batch ACKs, for example due to half duplex
channel allocation, the effective bottleneck data rate might be
much higher than any link in the system.
The use of multiple connections has been very controversial since the The use of multiple connections has been very controversial since the
beginning of the World-Wide-Web[first complaint]. Modern browsers beginning of the World-Wide-Web[first complaint]. Modern browsers
open many connections [BScope]. Experts in the IETF transport area open many connections [BScope]. Experts associated with IETF
have frequently spoken against this practice [long list]. It is not transport area have frequently spoken against this practice [long
inappropriate to assume some small number of concurrent connections list]. It is not inappropriate to assume some small number of
(e.g. 4 or 6), to compensate for limitation in TCP. However, concurrent connections (e.g. 4 or 6), to compensate for limitation in
choosing too large a number is at risk of being taken as a signal by TCP. However, choosing too large a number is at risk of being
the web browser community that this practice has been embraced by the interpreted as a signal by the web browser community that this
Internet community. It may not be desirable to send such a signal. practice has been embraced by the Internet service provider
community. It may not be desirable to send such a signal.
4.2. End-to-end parameters from sub-paths
[This entire section needs to be overhauled and should be skipped on
a first reading. The concepts defined here are not used elsewhere.]
The following optional parameters apply for testing generalized end- The following optional parameters apply for testing generalized end-
to-end paths that include subpaths with known specific types of to-end paths that include subpaths with known specific types of
behaviors that are not well represented by simple queueing models: behaviors that are not well represented by simple queueing models:
Bottleneck link clock rate: This applies to links that are using Bottleneck link clock rate: This applies to links that are using
virtual queues or other techniques to police or shape users virtual queues or other techniques to police or shape users
traffic at lower rates full link rate. The bottleneck link clock traffic at lower rates full link rate. The bottleneck link clock
rate should be representative of queue drain times for short rate should be representative of queue drain times for short
bursts of packets on an otherwise unloaded link. bursts of packets on an otherwise unloaded link.
Channel hold time: For channels that have relatively expensive Channel hold time: For channels that have relatively expensive
channel arbitration algorithms, this is the typical (maximum?) channel arbitration algorithms, this is the typical (maximum?)
time that data and or ACKs are held pending acquiring the channel. time that data and or ACKs are held pending acquiring the channel.
WHile under heavy load, the RTT may be inflated by this parameter, While under heavy load, the RTT may be inflated by this parameter,
unless it is built into the target RTT unless it is built into the target RTT
Preload traffic volume: If the user's traffic is shaped on the basis Preload traffic volume: If the user's traffic is shaped on the basis
of average traffic volume, this is volume necessary to invoke of average traffic volume, this is volume necessary to invoke
"heavy hitter" policies. "heavy hitter" policies.
Unloaded traffic volume: If the user's traffic is shaped on the Unloaded traffic volume: If the user's traffic is shaped on the
basis of average traffic volume, this is the maximum traffic basis of average traffic volume, this is the maximum traffic
volume that a test can use and stay within a "light user" volume that a test can use and stay within a "light user"
policies. policies.
Note on a ConEx enabled network [ConEx], the word "traffic" in the Note on a ConEx enabled network [ConEx], the word "traffic" in the
last two items should be replaced by "congestion" i.e. "preload last two items should be replaced by "congestion" i.e. "preload
congestion volume" and "unloaded congestion volume". congestion volume" and "unloaded congestion volume".
3.2. Per sub-path parameters 4.3. Per sub-path parameters
[This entire section needs to be overhauled and should be skipped on
a first reading. The concepts defined here are not used elsewhere.]
Some single parameter tests also need parameter of the sub-path. Some single parameter tests also need parameter of the sub-path.
sub-path RTT: RTT of the sub-path under test. sub-path RTT: RTT of the sub-path under test.
sub-path link clock rate: If different than the Bottleneck link sub-path link clock rate: If different than the Bottleneck link
clock rate clock rate
3.3. Common Calculations for Single Property Tests 4.4. Common Calculations for Single Property Tests
The most important derived parameter is target_pipe_size (in The most important derived parameter is target_pipe_size (in
packets), which is the number of packets needed exactly meet the packets), which is the number of packets needed exactly meet the
target rate, with no cross traffic for the specified RTT and MTU. It target rate, with no cross traffic for the specified target RTT and
is given by: MTU. It is given by:
target_pipe_size = target_rate * target_RTT / target_MTU [Need to add multiple connections]
target_pipe_size = target_rate * target_RTT / ( target_MTU -
header_overhead )
If the transport protocol (e.g. TCP) average window size is smaller If the transport protocol (e.g. TCP) average window size is smaller
than this, the link will be under filled. than this, it will not meet the target rate.
If target_data_rate is equal to bottleneck link_data_rate, then The reference_target_run_length, which is the most conservative model
target_pipe_size also predicts the onset of queueing. If the for the minimum spacing between losses, can derived as follows:
transport protocol (e.g. TCP) average window size is larger than the assume the target_data_rate is equal to bottleneck link_data_rate.
target_pipe_size, the excess packets will be in a standing queue at Then target_pipe_size also predicts the onset of queueing. If the
the bottleneck. transport protocol (e.g. TCP) has an average window size that is
larger than the target_pipe_size, the excess packets will form a
standing queue at the bottleneck.
If the transport protocol is using Reno congestion control [RFC5681], If the transport protocol is using traditional Reno style Additive
then there must be target_pipe_size roundtrips between losses. Increase, Multiplicative Decrease congestion control [RFC5681], then
Otherwise the multiplicative window reduction triggered by a loss there must be target_pipe_size roundtrips between losses. Otherwise
would cause the network to be underfilled. Following [MSMO97], we the multiplicative window reduction triggered by a loss would cause
derive the losses must be no more frequent than every 1 in the network to be underfilled. Following [MSMO97], we derive the
losses must be no more frequent than every 1 in
(3/2)(target_pipe_size^2) packets. This provides the reference value (3/2)(target_pipe_size^2) packets. This provides the reference value
for target_run_length which is typically the number of packets that for target_run_length which is typically the number of packets that
must be delivered between loss episodes in hte tests below: must be delivered between loss episodes in the tests below:
reference_target_run_length = (3/2)(target_pipe_size^2) reference_target_run_length = (3/2)(target_pipe_size^2)
Note that this calculation is based on a number of assumptions that Note that this calculation is based on a number of assumptions that
may not apply. Appendix A discusses these assumptions and provides may not apply. Appendix A discusses these assumptions and provides
some alternative models. The actual method for computing some alternative models. The actual method for computing
target_run_length MUST be published along with the rationale for the target_run_length MUST be documented along with the rationale for the
underlying assumptions and the ratio of chosen target_run_length to underlying assumptions and the ratio of chosen target_run_length to
reference_target_run_length. reference_target_run_length.
Although this document gives a lot of latitude for calculating Although this document gives a lot of latitude for calculating
target_run_length people specifying profiles for suites of single target_run_length, people specifying profiles for suites of single
property tests need to consider the effect of their choices on the property tests need to consider the effect of their choices on the
ongoing conversation and tussle about the relevance of "TCP ongoing conversation and tussle about the relevance of "TCP
friendliness" as an appropriate model for capacity allocation. friendliness" as an appropriate model for capacity allocation.
Choosing a target_run_length that is substantially smaller than Choosing a target_run_length that is substantially smaller than
reference_target_run_length is equivalent to saying that it is reference_target_run_length is equivalent to saying that it is
appropriate for the research community to abandon "TCP friendliness" appropriate for the transport research community to abandon "TCP
as a fairness model and to develop more aggressive Internet transport friendliness" as a fairness model and to develop more aggressive
protocols, and for applications to continue (or even increase) the Internet transport protocols, and for applications to continue (or
number of connections that they open. even increase) the number of connections that they open concurrently.
The calculations for individual parameters are presented with the The calculations for individual parameters are presented with the
each single property test. In general these calculations are each single property test. In general these calculations are
permitted some as described in Section 3.4 permitted some derating as described in Section 4.5
3.4. Parameter Derating 4.5. Parameter Derating
Since some aspects of the models are very conservative, the modeling Since some aspects of the models are very conservative, the modeling
framework permits some latitude in derating some specific test framework permits some latitude in derating some specific test
parameters, as indicated in Section 4. For example classical parameters, as indicated in Section 6. For example classical
performance models suggest that in order to be sure that a single TCP performance models suggest that in order to be sure that a single TCP
stream can fill a link, it needs to have a full bandwidth-delay- stream can fill a link, it needs to have a full bandwidth-delay-
product worth of buffering at the bottleneck[QueueSize]. In real product worth of buffering at the bottleneck[QueueSize]. In real
networks with real applications this is sometimes overly networks with real applications this is often overly conservative.
conservative. Rather than trying to formalize more complicated Rather than trying to formalize more complicated models we permit
models we permit some test parameters to be relaxed as long as they some test parameters to be relaxed as long as they meet some
meet some additional procedural constraints: additional procedural constraints:
The method used compute and justify the derated metrics is o The method used compute and justify the derated metrics is
published in such a way that it becomes a matter of public record. published in such a way that it becomes a matter of public record.
The calibration procedures described in Section 5 are used to o The calibration procedures described in Section 8 are used to
demonstrate the feasibility of meeting the performance targets demonstrate the feasibility of meeting the performance targets
with the derated test parameters. with the derated test parameters.
The calibration process itself is documented is such a way that o The calibration process itself is documented is such a way that
other researchers can duplicate the experiments and validate the other researchers can duplicate the experiments and validate the
results. results.
In the test specifications in Section 6 assume 0 < derate <= 1, is a
derating parameter. These will be individually named in the final
document. In all cases making derate smaller makes the test more
tolerant. Derate = 1 is "full strenght".
Note that some single property test parameters are not permitted to Note that some single property test parameters are not permitted to
be derated. be derated.
3.5. Single Property Tests Results 5. Common testing procedures
TBD Define: Pass, Fail and inconclusive test results. 5.1. Traffic generating techniques
The inconclusive outcome is needed to address the case where a test A key property of Model Based Metrics is that the traffic patterns
failed to attain the specified test conditions. This is important to are determined by the end-to-end target parameters and not the
network. The only exception are the constant window tests, which
rely on a TCP style self clock. This makes the tests "open loop",
which is key to preventing circular dependencies between test
prameters. The transmission pattern does not depend on the details
of the network's reaction to traffic.
5.1.1. Paced transmission
Paced (burst) transmissions: send bursts of data on a timer to meet a
particular target rate and pattern.
Single: Send individual packets at the specified rate or headway.
Burst: Send server interface rate bursts on a timer. Specify any 3
of average rate, packet size, burst size (number of packets) and
burst headway (start to start). These bursts are typically sent
as back-to-back packets on high speed media.
Slow start: Send 4 packets bursts at an average rate equal to twice
the effective bottleneck link rate (but not faster than the server
interface rate). This corresponds to the average rate during a
TCP slowstart when Appropriate Byte Counting [ABC] is present or
delayed ack is disabled. Note that slow start pacing itself is
typically part of larger scale burst patterns, such as sending
target_pipe_size packets as slow start bursts every on a
target_RTT headway (burst start to burst start). Such a stream
has three different average rates, depending on the averaging time
scale. At the finest time scale the average rate is the same as
the server rate, at a medium scale the average rate is twice the
bottleneck link rate and at the longest time scales the average
rate is the target data rate, adjusted to include header overhead.
Note that if the effective bottleneck link rate is more than half of
the server interface rate, slowstart bursts become server interface
rate bursts.
5.1.2. Constant window pseudo CBR
Implement pseudo CBR by running a standard protocol such as TCP at a
fixed window size. This has the advantage that it can be implemented
under real content delivery. The rate is only maintained in average
over each RTT, and is subject to limitations of the transport
protocol.
For tests that have strongly prescribed data rates, if the transport
protocol fails to maintain the test rate for any reason (especially
due to network congestion) the test should be considered
inconclusive, otherwise there are some cases where tester failures
might cause incorrect link tests results.
5.1.3. Scanned window pseudo CBR
Same as the above, except the window is incremented once per
2*target_RTT, starting from below target_pipe and sweeping up to
first loss or some other event. This is analogous to the tests
implemented in Windowed Ping [WPING] and pathdiag [PATHDIAG].
5.1.4. Intermittent Testing
Any test which does not depend on queueing (e.g. the CBR tests) or
normally experiences periodic zero outstanding data (e.g. between
bursts for burst tests), can be formulated as an intermittent test.
The Intermittent testing can be used for ongoing monitoring for
changes in sub-path quality with minimal disruption users. It should
be used in conjunction with the full rate test because this method
assesses an average_run_length over a long time interval w.r.t. user
sessions. It may false fail due to other legitimate congestion
causing traffic or may false pass changes in underlying link
properties (e.g. a modem retraining to an out of contract lower
rate).
5.1.5. Intermittent Scatter Testing
Intermittent scatter testing: when testing the network path to or
from an ISP subscriber aggregation point (Cable headend [better
word?] or DSLAM, etc), intermittent tests can be spread across a pool
of (idle) users such that no one users experiences the full impact of
the testing, even though the traffic to or from the ISP subscriber
aggregation point is sustained at full rate. EXPAND
5.2. Interpreting the Results
(This section needs major reorganization.)
MOVE ELSEWHERE: General comment about types of loss: masking BER type
loss with ARQ/FEC is not an issue. What we are most concerned with
is congestion or AQM related losses. These are explicitly considered
to be a signal back to the sender to slow down. Note that even with
ARQ or FEC at some point the link will accumulate enough backlog
where it will need to cause AQM (or drop tail overflow) losses.
5.2.1. Inconclusive test outcomes
A singleton is a pass fail measurement.
In addition we use "inconclusive" outcome to indicate that a test
failed to attain the required test conditions. This is important to
the extent that the tests themselves have built in control systems the extent that the tests themselves have built in control systems
which might interfere with some aspepect of the test. It is required which might interfere with some aspect of the test.
for example to use TCP for testing.
4. Single Property Tests For example if a test is implemented with an controled and
instrumented TCP, failing to attain the specified data rate may
indicate a problem with either the TCP implementation or the test
vantage point.
The single property tests confirm that each sub-path can sustain the One of the goal of the testing process should be to drive the number
normal traffic patterns caused by TCP running at the specified target of inconclusive tests to zero.
performance. Specifically they confirm that each sub-path has:
sufficient raw capacity (e.g. sufficient data rate); low enough
background loss rate where mandatory congestion control stays out of
the way; large enough queue space to absorb TCP's normal bursts; does
not cause unreasonable packet reordering; progressive AQM to
appropriately invoke congestion control. Appropriately invoking
congestion control requires that packet losses or ECN marks start
progressively before TCP creates an excessive sustained
queues[BufferBloat] or excessively bursty losses. The return path
must also subject to a similar suite of tests, although potentially
with different test parameters.
Note that many of the sub-path tests resemble metrics that have 5.2.2. Statistical criteria for measuring run_length
already been defined in the IPPM context, with the addition of
criteria for passing or failing the test. The models used to derive
the test parameters make specific assumptions about network
conditions, a test is deemed "inconclusive" (as opposed to failing)
if tester does not meet the underlying assumption. For example a
loss rate test at a specified data rate is inconclusive if the tester
fails to send data at the specified rate for some reason. This
concept of an inconclusive test is necessary to build tests out of
protocols or technologies that they themselves have built in or
implicit control systems.
Some single property test can be combined, since their parameters are When evaluating the traget_run_length, we need to determine
not mutually exclusive. appropriate packet stream sizes and acceptable error levels to test
efficiently. In practice, can we compare the empirically estimated
loss probabilities with the targets as the sample size grows? How
large a sample is needed to say that the measurements of packet
transfer indicate a particular run-length is present?
4.1. Verify the absence of cross traffic The generalized measurement can be described as recursive testing:
send a flight of packets and observe the packet transfer performance
(loss ratio or other metric, any defect we define).
Use a passive packet or SNMP monitoring to verify that the traffic As each flight is sent and measured, we have an ongoing estimate of
volume on the sub-path agrees with the traffic generated by each the performance in terms of defect to total packet ratio (or an
test. Ideally this should be performed before during and after each empirical probability). Continue to send until conditions support a
conclusion or a maximum sending limit has been reached.
We have a target_defect_probability, 1 defect per target_run_length,
where a "defect" is defined as a lost packet, a packet with ECN mark,
or other impairment. This constitutes the null Hypothesis:
H0: no more than one defects in target_run_length = (3/2)*(flight)^2
packets
and we can stop sending flights of packets if measurements support
accepting H0 with the specified Type I error = alpha (= 0.05 for
example).
We also have an alternative Hypothesis to evaluate: if performance is
significantly lower than the target_defect_probability, say half the
target:
H1: one or more defects in target_run_length/2 packets
and we can stop sending flights of packets if measurements support
rejecting H0 with the specified Type II error = beta, thus preferring
the alternate H1.
H0 and H1 constitute the Success and Failure outcomes described
elsewhere in the memo, and while the ongoing measurements do not
support either hypothesis the current status of measurements is
indeterminate.
The problem above is formulated to match the Sequential Probability
Ratio Test (SPRT) [StatQC] [temp ref:
http://en.wikipedia.org/wiki/Sequential_probability_ratio_test ],
which also starts with a pair of hypothesis specified as above:
H0: p = p0 = one defect in target_run_length
H1: p = p1 = one defect in target_run_length/2
As flights are sent and measurements collected, the tester evaluates
the cumulative log-likelihood ratio:
S_i = S_i-1 + log(Lambda_i)
where Lambda_i is the ratio of the two likelihood functions
(calculated on the measurement at packet i, and index i increases
linearly over all flights of packets ) for p0 and p1 [temp ref:
http://en.wikipedia.org/wiki/Likelihood_function ].
The SPRT specifies simple stopping rules:
o a < S_i < b: continue testing
o S_i <= a: Accept H0
o S_i >= b: Accept H1
where a and b are based on the Type I and II errors, alpha and beta:
a ~= Log((beta/(1-alpha)) and b ~= Log((1-beta)/alpha)
with the error probabilities decided beforehand, as above.
The calculations above are implemented in the R-tool for Statistical
Analysis, in the add-on package for Cross-Validation via Sequential
Testing (CVST) [http://www.r-project.org/] [Rtool] [CVST] .
5.2.3. Classifications of tests
These tests are annotated with "(load)", "(engineering)" or
"(monitoring)". WHY DO WE CARE?
A network would be expected to fail load tests in the presence excess
or uncontrolled cross traffic. As such, load tests identify
parameters that can transition from passing to failing as a
consequence of insufficient network capacity and the actions of other
network users.
Monitoring tests are design to capture the most important aspects of
a load test, but without causing unreasonable ongoing load
themselves. As such they may miss some details of the network
performance, but can serve as a useful reduced cost proxy for a load
test. test.
Engineering tests evaluate how network algorithms (such as AQM and
channel allocation) interact with transport protocols. Although
these tests may be to be sensitive to load, the interaction may be
quite complicated and might even have an inverted sensitivity. For
example a test to verify that an AQM algorithm causes ECN marks or
packet drops early enough to limit queue occupancy may experience a
false pass results in the presence of bursty cross traffic. It is
important that engineering tests be performed under a wide range of
conditions, including both in situ and bench testing, and under a
full range of load conditions. Ongoing monitoring is less likely to
be useful for these tests, although sparse testing might be
appropriate.
5.3. Reordering Tolerance
All tests must be instrumented for excessive reordering [RFC4737].
NB: there is no global consensus for how much reordering tolerance is
appropriate or reasonable. ("None" is absolutely unreasonable.)
Section 5 of [RFC4737] proposed a metric that may be sufficient to
designate isolated reordered packets as effectively lost, because
TCP's retransmission response would be the same.
[As a strawman, we propose the following:] TCP should be able to
adapt to reordering as long as the reordering extent is no more than
the maximum of one half window or 1 mS, whichever is larger. Note
that there is a fundamental tradeoff between tolerance to reordering
and how quickly algorithms such as fast retransmit can repair losses.
Within this limit on extent, there should be no bound on the
frequency.
Parameters:
dispalcement the maximum of one half of target_pipe_size or 1 mS.
5.4. Verify the absence of cross traffic
Cross traffic should be monitored prior to and during testing. In
sub-paths where traffic of many users is aggregated, an excessive
level of cross traffic should be noted and prevent testing (the test
should be recorded as "inconclusive"). In sub-paths that include
individual subscriber service, the current approach is to suspend
testing when subscriber traffic is detected, because neither a flawed
test nor degraded user performance conditions are desired.
Note that canceling tests due to load on subscriber lines may
introduce sampling errors for other parts of the infrastructure. For
this reason tests that are scheduled but not run due to load should
be treated as a special case of "inconclusive".
The test is deemed to have passed only if the observed data rate
matches the target_data_rate and it is statistically significant that
the average_run_lenght is larger than target_run_lenght. It is
deemed inconclusive if: the statistical test is inconclusive; there
is too much background load; or the target_data_rate could not be
attained.
Use a passive packet or SNMP monitoring to verify that the traffic
volume on the sub-path agrees with the traffic generated by a test.
Ideally this should be performed before during and after each test.
The goal is provide quality assurance on the overall measurement The goal is provide quality assurance on the overall measurement
process, and specifically to detect the following measurement process, and specifically to detect the following measurement
failure: a user observes unexpectedly poor application performance, failure: a user observes unexpectedly poor application performance,
the ISP observes that the access link is running at the rated the ISP observes that the access link is running at the rated
capacity. Both fail to observe that the user's computer has been capacity. Both fail to observe that the user's computer has been
infected by a virus which is spewing traffic as fast as it can. infected by a virus which is spewing traffic as fast as it can.
Parameters: Parameters:
Maximum Cross Traffic Data Rate The amount of excess traffic Maximum Cross Traffic Data Rate The amount of excess traffic
permitted. Note that this might be different for different tests. permitted. Note that this might be different for different tests.
Maximum Data Rate underage The permitted amount that the traffic can Maximum Data Rate underage The permitted amount that the traffic can
be less than predicted for the current test. Normally this would be less than predicted for the current test. Normally this would
just be a statement of the maximum permitted measurement error, just be a statement of the maximum permitted measurement error,
however it might also detect cases where the passive and active however it might also detect cases where the passive and active
tests are misaligned: testing different subscriber lines. This is tests are misaligned: testing different subscriber lines. This is
important because the vantage points are so different: in-band important because the vantage points are so different: in-band
active measurement vs out-of-band passive measurement. active measurement vs out-of-band passive measurement.
4.1.1. Parameter Calculation
TBA
4.1.2. Cross traffic Measurement
One possible method is an adaptation of: www-didc.lbl.gov/papers/ One possible method is an adaptation of: www-didc.lbl.gov/papers/
SCNM-PAM03.pdf D Agarwal etal. "An Infrastructure for Passive SCNM-PAM03.pdf D Agarwal etal. "An Infrastructure for Passive
Network Monitoring of Application Data Streams". Use the same Network Monitoring of Application Data Streams". Use the same
technique as that paper to trigger the capture of SNMP statistics for technique as that paper to trigger the capture of SNMP statistics for
the link. the link.
4.2. Full Data Rate Loss Rate Tests 5.5. Additional test preconditions
We propose two versions of the loss rate test. One, performed at Send pre-load traffic as needed to activate radios with a sleep mode,
data full rate, is intrusive and recommend for infrequent testing, or other "reactive network" elements (term defined in
such as when a service is first turned up or as part of an auditing [draft-morton-ippm-2330-update-01]).
process. Note that this test also implicitly confirms that sub_path
has sufficient capacity to carry the target_data_rate.
The second background loss rate, described below, is designed for Use the procedure above to confirm that the pre-test background
ongoing monitoring for change is sub-path quality. traffic is low enough.
6. Single Property Tests
The following tests are fully decomposed to verify individual network
properties required for TCP meet the target parameters. It is
believed that these properties apply to all self clocked throughput
maximizing protocols. Failing to meet any one of these tests will
cause poor TCP performance in some specific context.
These tests are pedantically separated: It would be more practical to
combine them. Failing such a combined test might imply ambiguous
consequences for TCP: it would be expected to fail under some
conditions, but a single test might not be able to indicate exactly
which conditions. The following section describes some combined
tests.
The single property tests confirm that each sub-path can sustain the
normal traffic patterns caused by TCP running at the specified target
performance. Specifically they confirm that each sub-path has:
sufficient raw capacity (e.g. sufficient data rate); low enough
background loss rate where mandatory congestion control stays out of
the way; large enough queue space to absorb TCP's normal bursts; does
not cause unreasonable packet reordering; progressive AQM to
appropriately invoke congestion control. Appropriately invoking
congestion control requires that packet losses or ECN marks start
progressively before TCP creates an excessive sustained
queues[BufferBloat] or and without causing excessively bursty losses.
The return path must also subject to a similar suite of tests,
although potentially with different test parameters (due to the
asymmetrical capabilities of many access link technologies).
Furthermore it is important that the forward and return path interact
appropriately, for example if they share they share a channel that
has to be allocated.
Note that many of the sub-path tests resemble metrics that have
already been defined in the IPPM context, with the specification of
additional of test parameters and success critera. The models used
to derive the test parameters make specific assumptions about network
conditions, a test is deemed "inconclusive" (as opposed to failing)
if tester does not meet the underlying assumption. For example a
loss rate test at a specified data rate is inconclusive if the tester
fails to send data at the specified rate for some reason. This
concept of an inconclusive test is necessary to build tests out of
protocols or technologies that they themselves have built in or
implicit control systems.
6.1. CBR Tests
We propose several versions of the CBR loss rate test. One,
performed at data full rate, is intrusive and recommend for
infrequent testing, such as when a service is first turned up or as
part of an auditing process. The second, background loss rate, is
designed for ongoing monitoring for change is sub-path quality.
6.1.1. Loss Rate at Full Data Rate
Send at target_rate, confirm that the observed run length is at least
the target_run_lenght. No additional derating is permitted (except
through the choice of model to calculate target_run_lenght in the
first place).
Note that this test also implicitly confirms that sub_path has
sufficient capacity to carry the target_data_rate.
Parameters: Parameters:
Run Length Same as target_run_lenght Run Length Same as target_run_lenght
Data Rate Same as target_data_rate Data Rate Same as target_data_rate
Note that these parameters MUST NOT be derated. If the default Note that these parameters MUST NOT be derated. If the default
parameters are too stringent use an alternate model for parameters are too stringent use an alternate model for
arget_data_rate as described in Appendix A. target_run_lenght as described in Appendix A.
4.2.1. Loss Rate Measurement
Data is sent at the specified data_rate. The receiver accumulates Data is sent at the specified data_rate. The receiver accumulates
the total data delivered and packets lost [and ECN marks, which are the total data delivered and packets lost [and ECN marks, which are
nominally treated as losses by conforming transport protocols]. The nominally treated as losses by conforming transport protocols]. The
observed average_run_lenght is computed from total_data_delivered observed average_run_lenght is computed from total_data_delivered
divided by the total_loss_rate. A [TBD] statistical test is applied divided by the total_loss_rate. A [TBD] statistical test is applied
to determine when or if the average_run_lenght is larger than to determine when or if the average_run_length is larger than
target_run_lenght. target_run_length.
TODO: add language about monitoring cross traffic.
The test is deemed to have passed only if the observed data rate TODO: add language about monitoring cross traffic. acm attempted
matches the target_data_rate and it is statistically significant that below.
the average_run_lenght is larger than target_run_lenght. It is
deemed inconclusive if: the statistical test is inconclusive; there
is too much background load; or the target_data_rate could not be
attained.
4.3. Background Loss Rate Tests 6.1.2. Background Loss Rate Tests
The background loss rate is designed for ongoing monitoring for The background loss rate is a low rate version of the target rate
change is sub-path quality. It should be used in conjunction with test above, designed for ongoing monitoring for changes in sub-path
the above full rate test. quality without disrupting users. It should be used in conjunction
with the above full rate test because it may be subject to false
results under some conditions, in particular it may false pass
changes in underlying link properties (e.g. a modem retraining to an
out of contract lower rate).
Parameters: Parameters:
Run Length Same as target_run_lenght Run Length Same as target_run_lenght
Data Rate Some small fraction of target_data_rate, such as 1%. Data Rate Some small fraction of target_data_rate, such as 1%.
4.3.1. Background Loss Rate Measurement Once the preconditions described in Section 5.5 are met, the test
data is sent at the prescribed rate with a burst size of 1. The
receiver accumulates packet delivery statistics and the procedures
described in Section 5.2.1 and Section 5.4 are used to score the
outcome:
The receiver accumulates the total data delivered and packets losses Pass: it is statistically significantly that the observed run length
[and ECN marks, which are nominally treated as losses by conforming is larger than the target_run_length.
transport protocols]. The observed average_run_lenght is computed
from total_data_delivered divided by the total_loss_rate. A [TBD]
statistical test is applied to determine when or if the
average_run_lenght is larger than target_run_lenght.
TODO: add language about monitoring cross traffic. Fail: it is statistically significantly that the observed run length
is smaller than the target_run_length.
The test is deemed to have passed if it is statistically significant Inconclusive: Neither test was statistically significant or there was
that the average_run_lenght is larger than target_run_lenght. It is excess cross traffic during the test.
deemed inconclusive if there is too much background traffic or the
statistical test is inconclusive.
4.4. Queue Capacity Test 6.2. Standing Queue tests
Parameters: These test confirm that the bottleneck is well behaved across the
TBA TBA onset of queueing. For conventional bottlenecks this will be from
the onset of queuing to the point where there is a full target_pipe
of standing data. Well behaved generally means lossless for
target_run_length, followed by a small number of losses to signal to
the transport protocol that it should slow down. Losses that are too
early can prevent the transport from averaging above the target_rate.
Losses that are too late indicate that the queue might be subject to
bufferbloat and subject other flows to excess queuing delay. Excess
losses (more than half of of target_pipe) make loss recovery
problematic for the transport protcol.
4.4.1. Model Calculation These tests can also observe some problems with channel acquisition
systems, especially at the onset of persistent queueing. Details
TBD.
TBA 6.2.1. Congestion Avoidance
4.4.2. Queue Capacity Measurement Use the procedure in Section 5.1.3 to sweep the window (rate) from
below link_pipe up to beyond target_pipe+link_pipe. Depending on
events that happen during the scan, score the link. Identify the
power_point=MAX(rate/RTT) as the start of the test.
TBA Fail if first loss is too early (loss rate too high) on repeated
tests or if the losses are more than half of the outstanding data. (a
load test)
4.5. AQM Test 6.2.2. Buffer Bloat
Parameters: Use the procedure in Section 5.1.3 to sweep the window (rate) from
TBA TBA below link_pipe up to beyond target_pipe+link_pipe. Depending on
events that happen during the scan, score the link. Identify the
"power point:MAX(rate/RTT) as the start of the test (should be
window=target_pipe)
4.5.1. Model Calculation Fail if first loss is too late (insufficient AQM and subject to
bufferbloat - an engineering test). NO THEORY
TBA 6.2.3. Self Interference
4.5.2. AQM Measurement Use the procedure in Section 5.1.3 to sweep the window (rate) from
below link_pipe up to beyond target_pipe+link_pipe. Depending on
events that happen during the scan, score the link. Identify the
"power point:MAX(rate/RTT) as the start of the test (should be
window=target_pipe)
TBA Fail if RTT is non-monotonic by more than a small number of packet
times (channel allocation self interference - engineering) IS THIS
SUFFICIENT?
4.6. Reordering Test 6.3. Slow Start tests
Parameters: These tests mimic slow start: data is sent at twice subpath_rate.
TBA TBA They are deemed inconclusive if the elapsed time to send the data
burst is not less than half of the (extrapolated) time to receive the
ACKs. (i.e. sending data too fast is ok, but sending it slower than
twice the actual bottleneck rate is deemed inconclusive). Space the
bursts such that the average ACK rate matches the target_data_rate.
4.6.1. Model Calculation These tests are not useful at burst sizes smaller than the server
rate tests, since the server rate tests are more strenuous. If it is
necessary to derate the server rate tests, then the full window
slowstart test (un-derated) would be important.
TBA 6.3.1. Full Window slow start test
4.6.2. Reordering Measurement Send target_pipe_size*derate bursts must have fewer than one loss per
target_run_length*derate. Note that these are the same parameters as
the Server Full Window test, except the burst rate is at slowestart
rate, rather than server interface rate. SHOULD derate=1.
TBA Otherwise TCP will exit from slowstart prematurely, and only reach a
full target_pipe_size window by way of congestion avoidance.
5. Calibration This is a load test: cross traffic may cause premature losses.
6.3.2. Slowstart AQM test
Do a continuous slowstart (date rate = 2*subpath_rate), until first
loss, and repeat, gathering statistics on the last delivered packet's
RTT and window size. Fail if too large (NO THEORY for value).
This is an engineering test: It would be best performed on a
quiescent network or testbed, since cross traffic might cause a false
pass.
6.4. Server Rate tests
These tests us "server interface rate" bursts. Although this is not
well defined it should be assumed to be current state of the art
server grade hardware (probably 10Gb/s today). (load)
6.4.1. Server TCP Send Offload (TSO) tests
If MIN(target_pipe_size, 42) packet bursts meet target_run_lenght
(Not derated!).
Otherwise the link will interact badly with modern server NIC
implementations, which as an optimization to reduce host side
interactions (interrupts etc) accept up to 64kB super packets and
send them as 42 seperate packets on the wire side.cc (load)
6.4.2. Server Full Window test
target_pipe_size*derate bursts have fewer than one loss per
target_run_length*derate.
Otherwise application pauses will cause unwarranted losses. Current
standards permit TCP to send a full cwnd burst following an
application pause. (Cwnd validation in not required, but even so
does not take effect until the pause is longer than RTO).
NB: there is no model here for what is good enough. derate=1 is
safest, but may be unnecessarily conservative for some applications.
Some application, such as streaming video need derate=1 to be
efficient when the application pacing quanta is larger than cwnd.
(load)
7. Combined Tests
These tests are more efficient from a deployment/operational
perspective, but may not be possible to diagnose if they fail.
7.1. Sustained burst test
Send target_pipe_size server rate bursts every target_RTT, verify
that the observed run length meets target_run_length. Key
observations:
o This test is RTT invariant, as long as the tester can generate the
required pattern.
o The subpath under test is expected to go idle for some fraction of
the time: (link_rate-target_rate)/link_rate. Failing to do so
suggests a problem with the procedure.
o This test is more strenuous than the slow start tests: they are
not needed if the link passes underated server rate burst tests.
o This test could be derated by reducing both the burst size and
headway (same average data rate).
o A link that passes this test is likely to be able to sustain
higher rates (close to link_rate) for paths with RTTs smaller than
the target_RTT. Offsetting this performance underestimation is
the rationale behind permitting derating in general.
o This test should be implementable with standard instrumented TCP,
[RFC 4898] using a specialized measurement application at one end
and a minimal service at the other end [RFC 863, RFC 864]. It may
require tweaks to the TCP implementation.
o This test is efficient to implement, since it does not require
per-packet timers, and can make maximal use of TSO in modern NIC
hardware.
o This test is not totally sufficient: the standing window
engineering tests are also needed to be sure that the link is well
behaved at and beyond the onset of congestion.
o I believe that this test can be proven to be the one load test to
supplant them all.
Example
To confirm that a 100 Mb/s link can reliably deliver single 10
MByte/s stream at a distance of 50 mS, test the link by sending 346
packet bursts every 50 mS (10 MByte/s payload rate, assuming a 1500
Byte IP MTU and 52 Byte TCP/IP headers). These bursts are 4196288
bits on the wire (assuming 16 bytes of link overhead and framing) for
an aggregate test data rate of 8.4 Mb/s.
To pass the test using the most conservative TCP model for a single
stream the observed run length must be larger than 179574 packets.
This is the same as less than one loss per 519 bursts (1.5*346) or
every 26 seconds.
Note that this test potentially cause transient 346 packet queues at
the bottleneck.
8. Calibration
If using derated metrics, or when something goes wrong, the results If using derated metrics, or when something goes wrong, the results
must be calibrated against a traditional BTC...... must be calibrated against a traditional BTC. The preferred
diagnostic follow-up to calibration issues is to run open end-to-end
measurements on an open platform, such as Measurement Lab
[http://www.measurementlab.net/]
6. References 9. Acknowledgements
6.1. Normative References
Ganga Maguluri suggested the statistical test for measuring loss
probability in the target run length.
Meredith Whittaker for improving the clarity of the communications.
10. References
10.1. Normative References
[RFC2119] Bradner, S., "Key words for use in RFCs to Indicate [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate
Requirement Levels", BCP 14, RFC 2119, March 1997. Requirement Levels", BCP 14, RFC 2119, March 1997.
[RFC2026] Bradner, S., "The Internet Standards Process -- Revision [RFC2026] Bradner, S., "The Internet Standards Process -- Revision
3", BCP 9, RFC 2026, October 1996. 3", BCP 9, RFC 2026, October 1996.
6.2. Informative References 10.2. Informative References
[RFC2330] Paxson, V., Almes, G., Mahdavi, J., and M. Mathis, [RFC2330] Paxson, V., Almes, G., Mahdavi, J., and M. Mathis,
"Framework for IP Performance Metrics", RFC 2330, "Framework for IP Performance Metrics", RFC 2330,
May 1998. May 1998.
[RFC4737] Morton, A., Ciavattone, L., Ramachandran, G., Shalunov,
S., and J. Perser, "Packet Reordering Metrics", RFC 4737,
November 2006.
[RFC5681] Allman, M., Paxson, V., and E. Blanton, "TCP Congestion [RFC5681] Allman, M., Paxson, V., and E. Blanton, "TCP Congestion
Control", RFC 5681, September 2009. Control", RFC 5681, September 2009.
[RFC5835] Morton, A. and S. Van den Berghe, "Framework for Metric [RFC5835] Morton, A. and S. Van den Berghe, "Framework for Metric
Composition", RFC 5835, April 2010. Composition", RFC 5835, April 2010.
[RFC6049] Morton, A. and E. Stephan, "Spatial Composition of
Metrics", RFC 6049, January 2011.
[MSMO97] Mathis, M., Semke, J., Mahdavi, J., and T. Ott, "The [MSMO97] Mathis, M., Semke, J., Mahdavi, J., and T. Ott, "The
Macroscopic Behavior of the TCP Congestion Avoidance Macroscopic Behavior of the TCP Congestion Avoidance
Algorithm", Computer Communications Review volume 27, Algorithm", Computer Communications Review volume 27,
number3, July 1997. number3, July 1997.
[BScope] Broswerscope, "Browserscope Network tests", Sept 2012, [BScope] Broswerscope, "Browserscope Network tests", Sept 2012,
<http://www.browserscope.org/?category=network>. <http://www.browserscope.org/?category=network>.
See Max Connections column See Max Connections column
[I-D.morton-ippm-lmap-path]
Bagnulo, M., Burbridge, T., Crawford, S., Eardley, P., and
A. Morton, "A Reference Path and Measurement Points for
LMAP", draft-morton-ippm-lmap-path-00 (work in progress),
January 2013.
[Rtool] R Development Core Team, "R: A language and environment
for statistical computing. R Foundation for Statistical
Computing, Vienna, Austria. ISBN 3-900051-07-0, URL
http://www.R-project.org/", , 2011.
[StatQC] Montgomery, D., "Introduction to Statistical Quality
Control - 2nd ed.", ISBN 0-471-51988-X, 1990.
[CVST] Krueger, T. and M. Braun, "R package: Fast Cross-
Validation via Sequential Testing", version 0.1, 11 2012.
Appendix A. Model Derivations Appendix A. Model Derivations
This appendix describes several different ways to calculate This appendix describes several different ways to calculate
target_run_length and the implication of the chosen calculation. target_run_length and the implication of the chosen calculation.
Rederive MSMO97 under two different assumptions: target_rate = Rederive MSMO97 under two different assumptions: target_rate =
link_rate and target_rate < 2 * link_rate. link_rate and target_rate < 2 * link_rate.
Show equivalent derivation for CUBIC. Show equivalent derivation for CUBIC.
skipping to change at page 15, line 40 skipping to change at page 29, line 36
and more frequent, less disruptive tests with the same target rate and more frequent, less disruptive tests with the same target rate
but the test rate equal to 1% of the target rate. To observe the but the test rate equal to 1% of the target rate. To observe the
required run length, the low rate test would take 100 times longer to required run length, the low rate test would take 100 times longer to
run. run.
Returning to the example: a full rate test would entail sending 690 Returning to the example: a full rate test would entail sending 690
pps (1 MByte/s) for several tens of seconds (e.g. 50k packets), and pps (1 MByte/s) for several tens of seconds (e.g. 50k packets), and
observing that the total loss rate is below 1:1e4. A less disruptive observing that the total loss rate is below 1:1e4. A less disruptive
test might be to send at 6.9 pps for 100 times longer, and observing test might be to send at 6.9 pps for 100 times longer, and observing
Formatted: Mon Oct 15 16:00:51 PDT 2012 Formatted: Mon Feb 25 15:01:45 PST 2013
Author's Address Authors' Addresses
Matt Mathis Matt Mathis
Google, Inc Google, Inc
1600 Amphitheater Parkway 1600 Amphitheater Parkway
Mountain View, California 93117 Mountain View, California 93117
USA USA
Email: mattmathis@google.com Email: mattmathis@google.com
Al Morton
AT&T Labs
200 Laurel Avenue South
Middletown, NJ 07748
USA
Phone: +1 732 420 1571
Email: acmorton@att.com
URI: http://home.comcast.net/~acmacm/
 End of changes. 103 change blocks. 
278 lines changed or deleted 873 lines changed or added

This html diff was produced by rfcdiff 1.48. The latest version is available from http://tools.ietf.org/tools/rfcdiff/