draft-ietf-ippm-model-based-metrics-05.txt   draft-ietf-ippm-model-based-metrics-06.txt 
IP Performance Working Group M. Mathis IP Performance Working Group M. Mathis
Internet-Draft Google, Inc Internet-Draft Google, Inc
Intended status: Experimental A. Morton Intended status: Experimental A. Morton
Expires: December 15, 2015 AT&T Labs Expires: January 7, 2016 AT&T Labs
June 13, 2015 July 6, 2015
Model Based Metrics for Bulk Transport Capacity Model Based Metrics for Bulk Transport Capacity
draft-ietf-ippm-model-based-metrics-05.txt draft-ietf-ippm-model-based-metrics-06.txt
Abstract Abstract
We introduce a new class of model based metrics designed to determine We introduce a new class of Model Based Metrics designed to assess if
if a complete Internet path can meet predefined bulk transport a complete Internet path can be expected to meet a predefined Bulk
performance targets by applying a suite of IP diagnostic tests to Transport Performance target by applying a suite of IP diagnostic
successive subpaths. The subpath-at-a-time tests can be robustly tests to successive subpaths. The subpath-at-a-time tests can be
applied to key infrastructure, such as interconnects or even robustly applied to key infrastructure, such as interconnects or even
individual devices, to accurately detect if any part of the individual devices, to accurately detect if any part of the
infrastructure will prevent any path traversing it from meeting the infrastructure will prevent any path traversing it from meeting the
specified target performance. specified Target Transport Performance.
The diagnostic tests consist of precomputed traffic patterns and The IP diagnostic tests consist of precomputed traffic patterns and
statistical criteria for evaluating packet delivery. The traffic statistical criteria for evaluating packet delivery. The traffic
patterns are precomputed to mimic TCP or other transport protocol patterns are precomputed to mimic TCP or other transport protocol
over a long path but are constructed in such a way that they are over a long path but are constructed in such a way that they are
independent of the actual details of the subpath under test, end independent of the actual details of the subpath under test, end
systems or applications. Likewise the success criteria depends on systems or applications. Likewise the success criteria depends on
the packet delivery statistics of the subpath, as evaluated against a the packet delivery statistics of the subpath, as evaluated against a
protocol model applied to the target performance. The success protocol model applied to the Target Transport Performance. The
criteria also does not depend on the details of the subpath, end success criteria also does not depend on the details of the subpath,
systems or application. This makes the measurements open loop, end systems or application. This makes the measurements open loop,
eliminating most of the difficulties encountered by traditional bulk eliminating most of the difficulties encountered by traditional bulk
transport metrics. transport metrics.
Model based metrics exhibit several important new properties not Model based metrics exhibit several important new properties not
present in other Bulk Capacity Metrics, including the ability to present in other Bulk Capacity Metrics, including the ability to
reason about concatenated or overlapping subpaths. The results are reason about concatenated or overlapping subpaths. The results are
vantage independent which is critical for supporting independent vantage independent which is critical for supporting independent
validation of tests results from multiple Measurement Points. validation of tests results from multiple Measurement Points.
This document does not define diagnostic tests directly, but provides This document does not define IP diagnostic tests directly, but
a framework for designing suites of IP diagnostics tests that are provides a framework for designing suites of IP diagnostics tests
tailored to confirming that infrastructure can meet a predetermined that are tailored to confirming that infrastructure can meet a
target performance. predetermined Target Transport Performance.
Interim DRAFT Formatted: Sat Jun 13 16:25:01 PDT 2015
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
skipping to change at page 2, line 17 skipping to change at page 2, line 16
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 December 15, 2015. This Internet-Draft will expire on January 7, 2016.
Copyright Notice Copyright Notice
Copyright (c) 2015 IETF Trust and the persons identified as the Copyright (c) 2015 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 . . . . . . . . . . . . . . . . . . . . . . . . . 5 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.1. Version Control . . . . . . . . . . . . . . . . . . . . . 6 1.1. Version Control . . . . . . . . . . . . . . . . . . . . . 6
2. Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 2. Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
3. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 10 3. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 9
4. New requirements relative to RFC 2330 . . . . . . . . . . . . 14 4. Background . . . . . . . . . . . . . . . . . . . . . . . . . . 14
5. Background . . . . . . . . . . . . . . . . . . . . . . . . . . 15 4.1. TCP properties . . . . . . . . . . . . . . . . . . . . . . 16
5.1. TCP properties . . . . . . . . . . . . . . . . . . . . . . 16 4.2. Diagnostic Approach . . . . . . . . . . . . . . . . . . . 17
5.2. Diagnostic Approach . . . . . . . . . . . . . . . . . . . 17 4.3. New requirements relative to RFC 2330 . . . . . . . . . . 18
6. Common Models and Parameters . . . . . . . . . . . . . . . . . 19 5. Common Models and Parameters . . . . . . . . . . . . . . . . . 18
6.1. Target End-to-end parameters . . . . . . . . . . . . . . . 19 5.1. Target End-to-end parameters . . . . . . . . . . . . . . . 18
6.2. Common Model Calculations . . . . . . . . . . . . . . . . 19 5.2. Common Model Calculations . . . . . . . . . . . . . . . . 19
6.3. Parameter Derating . . . . . . . . . . . . . . . . . . . . 20 5.3. Parameter Derating . . . . . . . . . . . . . . . . . . . . 20
7. Traffic generating techniques . . . . . . . . . . . . . . . . 21 5.4. Test Preconditions . . . . . . . . . . . . . . . . . . . . 21
7.1. Paced transmission . . . . . . . . . . . . . . . . . . . . 21 6. Traffic generating techniques . . . . . . . . . . . . . . . . 21
7.2. Constant window pseudo CBR . . . . . . . . . . . . . . . . 22 6.1. Paced transmission . . . . . . . . . . . . . . . . . . . . 21
7.3. Scanned window pseudo CBR . . . . . . . . . . . . . . . . 23 6.2. Constant window pseudo CBR . . . . . . . . . . . . . . . . 23
7.4. Concurrent or channelized testing . . . . . . . . . . . . 23 6.3. Scanned window pseudo CBR . . . . . . . . . . . . . . . . 24
8. Interpreting the Results . . . . . . . . . . . . . . . . . . . 24 6.4. Concurrent or channelized testing . . . . . . . . . . . . 24
8.1. Test outcomes . . . . . . . . . . . . . . . . . . . . . . 24 7. Interpreting the Results . . . . . . . . . . . . . . . . . . . 25
8.2. Statistical criteria for estimating run_length . . . . . . 26 7.1. Test outcomes . . . . . . . . . . . . . . . . . . . . . . 25
8.3. Reordering Tolerance . . . . . . . . . . . . . . . . . . . 27 7.2. Statistical criteria for estimating run_length . . . . . . 27
9. Test Preconditions . . . . . . . . . . . . . . . . . . . . . . 28 7.3. Reordering Tolerance . . . . . . . . . . . . . . . . . . . 29
10. Diagnostic Tests . . . . . . . . . . . . . . . . . . . . . . . 29 8. Diagnostic Tests . . . . . . . . . . . . . . . . . . . . . . . 29
10.1. Basic Data Rate and Delivery Statistics Tests . . . . . . 29 8.1. Basic Data Rate and Packet Delivery Tests . . . . . . . . 30
10.1.1. Delivery Statistics at Paced Full Data Rate . . . . . 30 8.1.1. Delivery Statistics at Paced Full Data Rate . . . . . 30
10.1.2. Delivery Statistics at Full Data Windowed Rate . . . 30 8.1.2. Delivery Statistics at Full Data Windowed Rate . . . . 31
10.1.3. Background Delivery Statistics Tests . . . . . . . . 30 8.1.3. Background Packet Delivery Statistics Tests . . . . . 31
10.2. Standing Queue Tests . . . . . . . . . . . . . . . . . . . 31 8.2. Standing Queue Tests . . . . . . . . . . . . . . . . . . . 31
10.2.1. Congestion Avoidance . . . . . . . . . . . . . . . . 32 8.2.1. Congestion Avoidance . . . . . . . . . . . . . . . . . 33
10.2.2. Bufferbloat . . . . . . . . . . . . . . . . . . . . . 32 8.2.2. Bufferbloat . . . . . . . . . . . . . . . . . . . . . 33
10.2.3. Non excessive loss . . . . . . . . . . . . . . . . . 33 8.2.3. Non excessive loss . . . . . . . . . . . . . . . . . . 33
10.2.4. Duplex Self Interference . . . . . . . . . . . . . . 33 8.2.4. Duplex Self Interference . . . . . . . . . . . . . . . 34
10.3. Slowstart tests . . . . . . . . . . . . . . . . . . . . . 34 8.3. Slowstart tests . . . . . . . . . . . . . . . . . . . . . 34
10.3.1. Full Window slowstart test . . . . . . . . . . . . . 34 8.3.1. Full Window slowstart test . . . . . . . . . . . . . . 35
10.3.2. Slowstart AQM test . . . . . . . . . . . . . . . . . 34 8.3.2. Slowstart AQM test . . . . . . . . . . . . . . . . . . 35
10.4. Sender Rate Burst tests . . . . . . . . . . . . . . . . . 35 8.4. Sender Rate Burst tests . . . . . . . . . . . . . . . . . 35
10.5. Combined and Implicit Tests . . . . . . . . . . . . . . . 35 8.5. Combined and Implicit Tests . . . . . . . . . . . . . . . 36
10.5.1. Sustained Bursts Test . . . . . . . . . . . . . . . . 36 8.5.1. Sustained Bursts Test . . . . . . . . . . . . . . . . 36
10.5.2. Streaming Media . . . . . . . . . . . . . . . . . . . 37 8.5.2. Streaming Media . . . . . . . . . . . . . . . . . . . 37
11. An Example . . . . . . . . . . . . . . . . . . . . . . . . . . 37 9. An Example . . . . . . . . . . . . . . . . . . . . . . . . . . 38
12. Validation . . . . . . . . . . . . . . . . . . . . . . . . . . 39 10. Validation . . . . . . . . . . . . . . . . . . . . . . . . . . 40
13. Security Considerations . . . . . . . . . . . . . . . . . . . 40 11. Security Considerations . . . . . . . . . . . . . . . . . . . 41
14. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 41 12. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 41
15. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 41 13. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 42
16. References . . . . . . . . . . . . . . . . . . . . . . . . . . 41 14. References . . . . . . . . . . . . . . . . . . . . . . . . . . 42
16.1. Normative References . . . . . . . . . . . . . . . . . . . 41 14.1. Normative References . . . . . . . . . . . . . . . . . . . 42
16.2. Informative References . . . . . . . . . . . . . . . . . . 41 14.2. Informative References . . . . . . . . . . . . . . . . . . 42
Appendix A. Model Derivations . . . . . . . . . . . . . . . . . . 44 Appendix A. Model Derivations . . . . . . . . . . . . . . . . . . 44
A.1. Queueless Reno . . . . . . . . . . . . . . . . . . . . . . 44 A.1. Queueless Reno . . . . . . . . . . . . . . . . . . . . . . 45
Appendix B. Complex Queueing . . . . . . . . . . . . . . . . . . 45 Appendix B. Complex Queueing . . . . . . . . . . . . . . . . . . 46
Appendix C. Version Control . . . . . . . . . . . . . . . . . . . 46 Appendix C. Version Control . . . . . . . . . . . . . . . . . . . 47
Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 46 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 47
1. Introduction 1. Introduction
Model Based Metrics (MBM) rely on mathematical models to specify a Model Based Metrics (MBM) rely on mathematical models to specify a
targeted diagnostic suite of IP diagnostic tests, designed to verify targeted suite of IP diagnostic tests, designed to assess whether
that common transport protocols can meet a predetermined performance common transport protocols can be expected to meet a predetermined
target over an Internet path. Each diagnostic in the suite measures performance target over an Internet path. Each test in the Targeted
some aspect of IP delivery that is required to meet the performance Diagnostic Suite (TDS) measures some aspect of IP packet transfer
target. For example a TDS may have separate diagnostic tests to that is required to meet the Target Transport Performance. For
verify that there is sufficient data rate and sufficient queueing example a TDS may have separate diagnostic tests to verify that there
buffer space to deliver typical transport bursts, and that the is: sufficient IP capacity (rate); sufficient queue space to deliver
background packet loss is small enough not to interfere with typical transport bursts; and that the background packet loss ratio
congestion control. Unlike other metrics which yield measures of is small enough not to interfere with congestion control. Unlike
network properties, Model Based Metrics nominally yield pass/fail other metrics which yield measures of network properties, Model Based
evaluations of the ability of transport protocols to meet a Metrics nominally yield pass/fail evaluations of the ability of
performance objective as need by a user application over a particular standard transport protocols to meet a specific performance objective
network path. over some network path.
This note describes the modeling framework to derive the IP This note describes the modeling framework to derive the IP
diagnostic test parameters from the target performance specified for diagnostic test parameters from the Target Transport Performance
TCP bulk transport capacity. In the future, other Model Based specified for TCP Bulk Transport Capacity. Model Based Metrics is an
Metrics may cover other applications and transports, such as VoIP alternative to the approach described in [RFC3148]. In the future,
over RTP. In most cases the IP diagnostic tests can be implemented other Model Based Metrics may cover other applications and
by combining existing IPPM metrics with additional controls for transports, such as VoIP over RTP. In most cases the IP diagnostic
precomputed traffic patterns and statistical criteria for evaluating tests can be implemented by combining existing IPPM metrics with
packet delivery. additional controls for generating precomputed traffic patterns and
statistical criteria for evaluating packet delivery.
This approach, mapping transport performance targets to a targeted This approach, mapping Target Transport Performance to a targeted
diagnostic suite (TDS) of IP diagnostic tests, solves an intrinsic diagnostic suite (TDS) of IP tests, solves some intrinsic problems
problem with using TCP or other throughput maximizing protocols for with using TCP or other throughput maximizing protocols for
measurement. In particular all throughput maximizing protocols (and measurement. In particular all throughput maximizing protocols (and
TCP congestion control in particular) cause some level of congestion TCP congestion control in particular) cause some level of congestion
in order to fill the network. This self inflicted congestion in order to detect when they have filled the network. This self
obscures the network properties of interest and introduces non-linear inflicted congestion obscures the network properties of interest and
equilibrium behaviors that make any resulting measurements useless as introduces non-linear equilibrium behaviors that make any resulting
metrics because they have no predictive value for conditions or paths measurements useless as metrics because they have no predictive value
different than the measurement itself. This problem is discussed in for conditions or paths other than that of the measurement itself.
Section 5. These problems are discussed at length in Section 4.
A targeted suite of IP diagnostic tests do not have such A targeted suite of IP diagnostic tests does not have such
difficulties. They can be constructed to make strong statistical difficulties. They can be constructed such that they make strong
statements about path properties that are independent of the statistical statements about path properties that are independent of
measurement details, such as vantage and choice of measurement the measurement details, such as vantage and choice of measurement
points. Model Based Metrics bridge the gap between empirical IP points. Model Based Metrics bridge the gap between empirical IP
measurements and expected TCP performance. measurements and expected TCP performance.
1.1. Version Control 1.1. Version Control
RFC Editor: Please remove this entire subsection prior to RFC Editor: Please remove this entire subsection prior to
publication. publication.
Please send comments about this draft to ippm@ietf.org. See Please send comments about this draft to ippm@ietf.org. See
http://goo.gl/02tkD for more information including: interim drafts, http://goo.gl/02tkD for more information including: interim drafts,
an up to date todo list and information on contributing. an up to date todo list and information on contributing.
Formatted: Sat Jun 13 16:25:01 PDT 2015 Formatted: Mon Jul 6 13:49:30 PDT 2015
Changes since -05 draft:
o Wordsmithing on sections overhauled in -05 draft.
o Reorganized the document:
* Relocated subsection "Preconditions".
* Relocated subsection "New Requirements relative to RFC 2330".
o Addressed nits and not so nits by Ruediger Geib. (Thanks!)
o Substantially tightened the entire definitions section.
o Many terminology changes, to better conform to other docs :
* IP rate and IP capacity (following RFC 5136) replaces various
forms of link data rate.
* subpath replaces link.
* target_window_size replaces target_pipe_size.
* Implied Bottleneck IP Rate replaces effective bottleneck link
rate.
* Packet delivery statistics replaces delivery statistics.
Changes since -04 draft: Changes since -04 draft:
o The introduction was heavily overhauled: split into a separate o The introduction was heavily overhauled: split into a separate
introduction and overview. introduction and overview.
o The new shorter introduction: o The new shorter introduction:
* Is a problem statement; * Is a problem statement;
* This document provides a framework; * This document provides a framework;
* That it replaces TCP measurement by IP tests; * That it replaces TCP measurement by IP tests;
* That the results are pass/fail. * That the results are pass/fail.
o Added a diagram of the framework to the overview o Added a diagram of the framework to the overview
o and introduces all of the elements of the framework. o and introduces all of the elements of the framework.
o Renumbered sections, reducing the depth of some section numbers. o Renumbered sections, reducing the depth of some section numbers.
o Updated definitions to better agree with other documents: o Updated definitions to better agree with other documents:
* Reordered section 2 * Reordered section 2
* Bulk [data] performance -> Bulk Transport Capacity, everywhere * Bulk [data] performance -> Bulk Transport Capacity, everywhere
including the title. including the title.
* loss rate and loss probability -> loss ratio * loss rate and loss probability -> packet loss ratio
* end-to-end path -> complete path * end-to-end path -> complete path
* [end-to-end][target] performance -> target transport * [end-to-end][target] performance -> Target Transport
performance Performance
* load test -> capacity test
This interim draft is a partial update since the WGLC, to collect an * load test -> capacity test
additional round of feedback on the Introduction, overview, and
terminology sections. Note that some of the prior WGLC comments are
still pending. Later sections (4 and beyond) have only been updated
to track changes in the terminology section. We intend to produce an
additional draft prior to the IETF, incorporating still pending
comments from the WGLC and any additional comments on the
introduction and overview.
2. Overview 2. Overview
This document describes a modeling framework for deriving Target This document describes a modeling framework for deriving a Targeted
Diagnostic Suites to determine if an IP path can be expected to meet Diagnostic Suite from a predetermined Target Transport Performance.
a predetermined target performance. It relies on other standards It is not a complete specification, and relies on other standards
documents to define Important details such as packet type-p documents to define important details such as packet type-p
selection, sampling techniques, vantage selection, etc. which are not selection, sampling techniques, vantage selection, etc. We imagine
specified here. We imagine Fully Specified Targeted Diagnostic Fully Specified Targeted Diagnostic Suites (FSTDS), that define all
Suites (FSTDS), that define all of these details. We use TDS to of these details. We use Targeted Diagnostic Suite (TDS) to refer to
refer to the subset of such a specification that is in scope for this the subset of such a specification that is in scope for this
document. document. This terminology is defined in Section 3.
Figure 1 shows the MBM modeling and measurement framework. (See Section 4 describes some key aspects of TCP behavior and what it
Section 3 for terminology used throughout this document). The target implies about the requirements for IP packet delivery. Most of the
transport performance is determined by the needs of the user or IP diagnostic tests needed to confirm that the path meets these
application, outside the scope of this document. For bulk transport properties can be built on existing IPPM metrics, with the addition
capacity, the performance parameter of interest is the target data of statistical criteria for evaluating packet delivery and in a few
rate. However, since TCP's ability to compensate for less than ideal cases, new mechanisms to implement precomputed traffic patterns.
network conditions is fundamentally affected by the Round Trip Time (One group of tests, the standing queue tests described in
(RTT) and the Maximum Transmission Unit (MTU) of the complete path, Section 8.2, don't correspond to existing IPPM metrics, but suitable
these parameters must also be specified in advance using knowledge metrics can be patterned after existing tools.)
about the intended application setting. Section 6 describes the
common parameters and models used to derive a targeted diagnostic
suite.
The target transport performance may reflect a specific application Figure 1 shows the MBM modeling and measurement framework. The
over real path through the Internet or an idealized application and Target Transport Performance, at the top of the figure, is determined
path representing a typical user community. by the needs of the user or application, outside the scope of this
document. For Bulk Transport Capacity, the main performance
parameter of interest is the target data rate. However, since TCP's
ability to compensate for less than ideal network conditions is
fundamentally affected by the Round Trip Time (RTT) and the Maximum
Transmission Unit (MTU) of the complete path, these parameters must
also be specified in advance based on knowledge about the intended
application setting. They may reflect a specific application over
real path through the Internet or an idealized application and
hypothetical path representing a typical user community. Section 5
describes the common parameters and models derived from the Target
Transport Performance.
target transport performance Target Transport Performance
(target data rate, target RTT and target MTU) (target data rate, target RTT and target MTU)
| |
________V_________ ________V_________
| mathematical | | mathematical |
| models | | models |
| | | |
------------------ ------------------
Traffic parameters | | Statistical criteria Traffic parameters | | Statistical criteria
| | | |
_______V____________V____Targeted_______ _______V____________V____Targeted_______
skipping to change at page 8, line 38 skipping to change at page 8, line 38
| | subpath under test | |- | | subpath under test | |-
----V----------------------------------V--- | ----V----------------------------------V--- |
| | | | | | | | | | | |
V V V V V V V V V V V V
fail/inconclusive pass/fail/inconclusive fail/inconclusive pass/fail/inconclusive
Overall Modeling Framework Overall Modeling Framework
Figure 1 Figure 1
Section 5 describes some key aspects of TCP behavior and what they The mathematical models are used to design traffic patterns that
imply about the requirements for IP packet delivery. Most of the IP mimic TCP or other bulk transport protocol operating at the target
diagnostic tests needed to confirm that the path meets these data rate, MTU and RTT over a full range of conditions, including
properties can be built on existing IPPM metrics, with the addition flows that are bursty at multiple time scales. The traffic patterns
of statistical criteria for evaluating packet delivery and in some are generated based on the three target parameters of complete path
cases new mechanisms to implement precomputed traffic patterns. One and independent of the properties of individual subpaths using the
group of tests, the standing queue tests described in section techniques described in Section 6. As much as possible the
Section 10.2, don't correspond to existing IPPM metrics, but suitable measurement traffic is generated deterministically (precomputed) to
metrics can be patterned after existing tools. minimize the extent to which test methodology, measurement points,
measurement vantage or path partitioning affect the details of the
Mathematical models are used to design traffic patterns that mimic measurement traffic.
TCP or other bulk transport protocol operating at the target data
rate, MTU and RTT over a full range of conditions, including flows
that are bursty at multiple time scales. The traffic patterns are
generated based on the three target parameters of complete path and
independent of the properties of individual subpaths as described in
Section 7. As much as possible the measurement traffic is generated
deterministically to that minimize the extent to which test
methodology, measurement points, measurement vantage or path
partitioning affect the details of the measurement traffic.
Section 8 describes packet delivery statistics and methods test them Section 7 describes packet delivery statistics and methods test them
against the bounds provided by the mathematical models. Since these against the bounds provided by the mathematical models. Since these
statistics are typically aggregated from all subpaths of the complete statistics are typically the composition of subpaths of the complete
path, in situ testing requires that the end-to-end statistical bounds path [RFC6049] , in situ testing requires that the end-to-end
be apportioned as a separate bound for each subpath. Links that are statistical bounds be apportioned as separate bounds for each
expected to be bottlenecks are expected to contribute a larger subpath. Subpaths that are expected to be bottlenecks may be
fraction of the total packet loss. In compensation, other links have expected to contribute a larger fraction of the total packet loss.
to be constrained to contribute less packet loss. The criteria for In compensation, non-bottlenecked subpaths have to be constrained to
passing each test of a TDS is an apportioned share of the total bound contribute less packet loss. The criteria for passing each test of a
determined by the mathematical model from the target transport TDS is an apportioned share of the total bound determined by the
performance . mathematical model from the Target Transport Performance.
Section 10 describes the suite of individual tests needed to verify Section 8 describes the suite of individual tests needed to verify
all of required IP delivery properties. A subpath passes if and only all of required IP delivery properties. A subpath passes if and only
if all of the individual IP diagnostics tests pass. Any subpath that if all of the individual IP diagnostics tests pass. Any subpath that
fails any test indicates that some users are likely fail to attain fails any test indicates that some users are likely fail to attain
their target transport performance under some conditions. In their Target Transport Performance under some conditions. In
addition to passing or failing, a test can be deemed to be addition to passing or failing, a test can be deemed to be
inconclusive for a number of reasons including: the precomputed inconclusive for a number of reasons including: the precomputed
traffic pattern was not accurately generated; the measurement results traffic pattern was not accurately generated; the measurement results
were not statistically significant; and others such as failing to were not statistically significant; and others such as failing to
meet some required test preconditions. If all test pass, except some meet some required test preconditions. If all tests pass, except
are inconclusive then the entire suite is deemed to be inconclusive. some are inconclusive then the entire suite is deemed to be
inconclusive.
Since there is some uncertainty in this process, Section 12,
describes a validation procedure to diagnose and minimize false
positive and false negative results.
In Section 11 we present an example TDS that might be representative In Section 9 we present an example TDS that might be representative
of HD video, and illustrate how Model Based Metrics can be used to of HD video, and illustrate how Model Based Metrics can be used to
address difficult measurement situations, such as confirming that address difficult measurement situations, such as confirming that
intercarrier exchanges have sufficient performance and capacity to intercarrier exchanges have sufficient performance and capacity to
deliver HD video between ISPs. deliver HD video between ISPs.
A TDS includes the target parameters, documentation of the models and Since there is some uncertainty in the modeling process, Section 10
assumptions used to derive the IP diagnostic test parameters, describes a validation procedure to diagnose and minimize false
specifications for the traffic and delivery statistics for the tests positive and false negative results.
themselves, and a description of a test setup that can be used to
validate the tests and models.
3. Terminology 3. Terminology
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
"SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
document are to be interpreted as described in [RFC2119]. document are to be interpreted as described in [RFC2119].
Note that terms containing underscores (rather than spaces) appear in
equations in the modeling sections. In some cases both forms are
used for aesthetic reasons, they do not have different meanings.
General Terminology: General Terminology:
Target: A general term for any parameter specified by or derived Target: A general term for any parameter specified by or derived
from the user's application or transport performance requirements. from the user's application or transport performance requirements.
Complete Path: From RFC 5835 Target Transport Performance: Application or transport performance
target transport performance: Application or transport performance goals for the complete path. For Bulk Transport Capacity defined
goals for the complete path. For bulk transport capacity defined in this note the Target Transport Performance includes the target
in this note the target transport performance includes the target
data rate, target RTT and target MTU as described below. data rate, target RTT and target MTU as described below.
Target Data Rate: The specified application data rate required for Target Data Rate: The specified application data rate required for
an application's proper operation. This is typically the an application's proper operation. Conventional BTC metrics are
performance goal as needed by the ultimate user. focused on the target data rate, however these metrics had little
Target RTT (Round Trip Time): The baseline (minimum) RTT of the or no predictive value because they do not consider the effects of
longest complete path over which the application expects to be the other two parameters of the Target Transport Performance, the
able meet the target performance. TCP and other transport RTT and MTU of the complete paths.
Target RTT (Round Trip Time): The specified baseline (minimum) RTT
of the longest complete path over which the application expects to
be able meet the target performance. TCP and other transport
protocol's ability to compensate for path problems is generally protocol's ability to compensate for path problems is generally
proportional to the number of round trips per second. The Target proportional to the number of round trips per second. The Target
RTT determines both key parameters of the traffic patterns (e.g. RTT determines both key parameters of the traffic patterns (e.g.
burst sizes) and the thresholds on acceptable traffic statistics. burst sizes) and the thresholds on acceptable traffic statistics.
The Target RTT must be specified considering authentic packets The Target RTT must be specified considering appropriate packets
sizes: MTU sized packets on the forward path, ACK sized packets sizes: MTU sized packets on the forward path, ACK sized packets
(typically header_overhead) on the return path. (typically header_overhead) on the return path. Note that target
Target MTU (Maximum Transmission Unit): The maximum MTU supported by RTT is specified and not measured, it determines the applicability
the complete path the over which the application expects to meet MBM evaluations for paths that are different than the measured
the target performance. Assume 1500 Byte MTU unless otherwise path.
specified. If some subpath forces a smaller MTU, then it becomes Target MTU (Maximum Transmission Unit): The specified maximum MTU
the target MTU, and all model calculations and subpath tests must supported by the complete path the over which the application
use the same smaller MTU. expects to meet the target performance. Assume 1500 Byte MTU
Targeted Diagnostic Suite (TDS): A set of IP Diagnostics designed to unless otherwise specified. If some subpath forces a smaller MTU,
determine if an otherwise ideal complete path containing the then it becomes the target MTU for the complete path, and all
subpath under test can sustain flows at a specific model calculations and subpath tests must use the same smaller
MTU.
Targeted Diagnostic Suite (TDS): A set of IP diagnostic tests
designed to determine if an otherwise ideal complete path
containing the subpath under test can sustain flows at a specific
target_data_rate using target_MTU sized packets when the RTT of target_data_rate using target_MTU sized packets when the RTT of
the complete path is target_RTT. the complete path is target_RTT.
Fully Specified Targeted Diagnostic Suite: A TDS together with Fully Specified Targeted Diagnostic Suite: A TDS together with
additional specification such as "type-p", etc which are out of additional specification such as "type-p", etc which are out of
scope for this document, but need to be drawn from other standards scope for this document, but need to be drawn from other standards
documents. documents.
loss ratio: See "Packet Loss Ratio in [RFC2680bis]
apportioned: To divide and allocate, for example budgeting packet
loss ratio across multiple subpaths such that they will accumulate
to less than a specified end-to-end loss ratio.
open loop: A control theory term used to describe a class of
techniques where systems that naturally exhibit circular
dependencies can be analyzed by suppressing some of the
dependences, such that the resulting dependency graph is acyclic.
Bulk Transport Capacity: Bulk Transport Capacity Metrics evaluate an Bulk Transport Capacity: Bulk Transport Capacity Metrics evaluate an
Internet path's ability to carry bulk data, such as large files, Internet path's ability to carry bulk data, such as large files,
streaming (non-real time) video, and under some conditions, web streaming (non-real time) video, and under some conditions, web
images and other content. Prior efforts to define BTC metrics images and other content. Prior efforts to define BTC metrics
have been based on [RFC3148], which never succeeded due to some have been based on [RFC3148], which predates our understanding of
overlooked requirements described in Section 4 and problems TCP ant the requirements described in Section 4
described in The metrics presented in this document reflect an
entirely different approach to the problem outlined in [RFC3148]. IP diagnostic tests: Measurements or diagnostic tests to determine
if packet delivery statistics meet some precomputed target.
traffic patterns: The temporal patterns or statistics of traffic traffic patterns: The temporal patterns or statistics of traffic
generated by applications over transport protocols such as TCP. generated by applications over transport protocols such as TCP.
There are several mechanisms that cause bursts at various time There are several mechanisms that cause bursts at various time
scales. Our goal here is to mimic the range of common patterns scales as described in Section 4.1. Our goal here is to mimic the
(burst sizes and rates, etc), without tieing our applicability to range of common patterns (burst sizes and rates, etc), without
specific applications, implementations or technologies, which are tying our applicability to specific applications, implementations
sure to become stale. or technologies, which are sure to become stale.
delivery Statistics: Raw or summary statistics about packet delivery packet delivery statistics: Raw, detailed or summary statistics
properties of the IP layer including packet losses, ECN marks, about packet delivery properties of the IP layer including packet
reordering, or any other properties that may be germane to losses, ECN marks, reordering, or any other properties that may be
transport performance. germane to transport performance.
IP performance tests: Measurements or diagnostic tests to determine packet loss ratio: As defined in [I-D.ietf-ippm-2680-bis].
delivery statistics. apportioned: To divide and allocate, for example budgeting packet
loss across multiple subpaths such that they will accumulate to
less than a specified end-to-end loss ratio.
open loop: A control theory term used to describe a class of
techniques where systems that naturally exhibit circular
dependencies can be analyzed by suppressing some of the
dependencies, such that the resulting dependency graph is acyclic.
Terminology about paths, etc. See [RFC2330] and [RFC7398]. Terminology about paths, etc. See [RFC2330] and [RFC7398].
[data] sender: Host sending data and receiving ACKs. [data] sender: Host sending data and receiving ACKs.
[data] receiver: Host receiving data and sending ACKs. [data] receiver: Host receiving data and sending ACKs.
subpath: A portion of the full path. Note that there is no complete path: The end-to-end path from the data sender to the data
requirement that subpaths be non-overlapping. receiver.
subpath: A portion of the complete path. Note that there is no
requirement that subpaths be non-overlapping. A subpath can be a
small as a single device, link or interface.
Measurement Point: Measurement points as described in [RFC7398]. Measurement Point: Measurement points as described in [RFC7398].
test path: A path between two measurement points that includes a test path: A path between two measurement points that includes a
subpath of the complete path under test, and could include subpath of the complete path under test, and if the measurement
infrastructure between the measurement points and the subpath. points are off path, may include "test leads" between the
measurement points and the subpath.
[Dominant] Bottleneck: The Bottleneck that generally dominates [Dominant] Bottleneck: The Bottleneck that generally dominates
traffic statistics for the entire path. It typically determines a packet delivery statistics for the entire path. It typically
flow's self clock timing, packet loss and ECN marking rate. See determines a flow's self clock timing, packet loss and ECN marking
Section 5.1. rate. See Section 4.1.
front path: The subpath from the data sender to the dominant front path: The subpath from the data sender to the dominant
bottleneck. bottleneck.
back path: The subpath from the dominant bottleneck to the receiver. back path: The subpath from the dominant bottleneck to the receiver.
return path: The path taken by the ACKs from the data receiver to return path: The path taken by the ACKs from the data receiver to
the data sender. the data sender.
cross traffic: Other, potentially interfering, traffic competing for cross traffic: Other, potentially interfering, traffic competing for
network resources (bandwidth and/or queue capacity). network resources (bandwidth and/or queue capacity).
Properties determined by the complet path and application. They are Properties determined by the complete path and application. They are
described in more detail in Section 6.1. described in more detail in Section 5.1.
Application Data Rate: General term for the data rate as seen by the Application Data Rate: General term for the data rate as seen by the
application above the transport layer. This is the payload data application above the transport layer in bytes per second. This
rate, and explicitly excludes transport and lower level headers is the payload data rate, and explicitly excludes transport and
(TCP/IP or other protocols), retransmissions and other overhead lower level headers (TCP/IP or other protocols), retransmissions
that is not part to the total quantity of data delivered to the and other overhead that is not part to the total quantity of data
application. delivered to the application.
Link Data Rate: General term for the data rate as seen by the link IP Rate: The actual number of IP-layer bytes delivered through a
or lower layers. The link data rate includes transport and IP subpath, per unit time, including TCP and IP headers, retransmits
headers, retransmissions and other transport layer overhead. This and other TCP/IP overhead. Follows from IP-type-P Link Usage
document is agnostic as to whether the link data rate includes or [RFC5136].
excludes framing, MAC, or other lower layer overheads, except that IP Capacity: The maximum number of IP-layer bytes that can be
they must be treated uniformly. transmitted through a subpath, per unit time, including TCP and IP
Effective Bottleneck Data Rate: This is the bottleneck data rate headers, retransmits and other TCP/IP overhead. Follows from IP-
implied by the returning ACKs, by looking at how much application type-P Link Capacity [RFC5136].
data the ACK stream reports delivered per unit time. If the path Bottleneck IP Rate: This is the IP rate of the data flowing through
is thinning ACKs or batching ACKs the effective bottleneck rate the dominant bottleneck in the forward path. TCP and other
can be much higher than the average link rate. See Section 5.1 protocols normally derive their self clocks from the timing of
and Appendix B for more details. this data. See Section 4.1 and Appendix B for more details.
[sender | interface] rate: The burst data rate, constrained by the Implied Bottleneck IP Rate: This is the bottleneck IP rate implied
data sender's interface. Today 1 or 10 Gb/s are typical. by the returning ACKs from the receiver. It is determined by
looking at how much application data the ACK stream reports
delivered per unit time. If the return path is thinning, batching
or otherwise altering ACK timing TCP will derive its clock from
the the implied bottleneck IP rate of the ACK stream, which in the
short term, might be much different than the actual bottleneck IP
rate. In the case of thinned or batched ACKs front path must have
sufficient buffering to smooth any data bursts to the IP capacity
of the bottleneck. If the return path is not altering the ACK
stream, then the Implied Bottleneck IP Rate will be the same as
the Bottleneck IP Rate. See Section 4.1 and Appendix B for more
details.
[sender | interface] rate: The IP rate which corresponds to the IP
Capacity of the data sender's interface. Due to issues of sender
efficiency and technologies such as TCP offload engines, nearly
all moderns servers deliver data in bursts at full interface link
rate. Today 1 or 10 Gb/s are typical.
Header_overhead: The IP and TCP header sizes, which are the portion Header_overhead: The IP and TCP header sizes, which are the portion
of each MTU not available for carrying application payload. of each MTU not available for carrying application payload.
Without loss of generality this is assumed to be the size for Without loss of generality this is assumed to be the size for
returning acknowledgements (ACKs). For TCP, the Maximum Segment returning acknowledgements (ACKs). For TCP, the Maximum Segment
Size (MSS) is the Target MTU minus the header_overhead. Size (MSS) is the Target MTU minus the header_overhead.
Basic parameters common to models and subpath tests are defined here Basic parameters common to models and subpath tests are defined here
are described in more detail in Section 6.2. Note that these are are described in more detail in Section 5.2. Note that these are
mixed between application transport performance (excludes headers) mixed between application transport performance (excludes headers)
and link IP performance (includes headers). and IP performance (which include TCP headers and retransmissions as
part of the payload).
Window: The total quantity of data plus the data represented by ACKs Window: The total quantity of data plus the data represented by ACKs
circulating in the network is referred to as the window. See circulating in the network is referred to as the window. See
Section 5.1 Section 4.1. Sometimes used with other qualifiers (congestion
window, cwnd or receiver window) to indicate which mechanism is
controlling the window.
pipe size: A general term for number of packets needed in flight pipe size: A general term for number of packets needed in flight
(the window size) to exactly fill some network path or subpath. (the window size) to exactly fill some network path or subpath.
It corresponds to the window size which maximizes network power, It corresponds to the window size which maximizes network power,
the observed data rate divided by the observed RTT. Often used the observed data rate divided by the observed RTT. Often used
with additional qualifies to specify which path, etc. with additional qualifiers to specify which path, or under what
conditions, etc.
target_pipe_size: The number of packets in flight (the window size) target_window_size: The average number of packets in flight (the
needed to exactly meet the target rate, with a single stream and window size) needed to meet the target data rate, for the
no cross traffic for the specified application target data rate, specified target RTT, and MTU. It implies the scale of the bursts
RTT, and MTU. It is the amount of circulating data required to
meet the target data rate, and implies the scale of the bursts
that the network might experience. that the network might experience.
run length: A general term for the observed, measured, or specified run length: A general term for the observed, measured, or specified
number of packets that are (to be) delivered between losses or ECN number of packets that are (expected to be) delivered between
marks. Nominally one over the sum of the loss and ECN marking losses or ECN marks. Nominally one over the sum of the loss and
probabilities, if there are independently and identically ECN marking probabilities, if there are independently and
distributed. identically distributed.
target_run_length: The target_run_length is an estimate of the target_run_length: The target_run_length is an estimate of the
minimum number of non-congestion marked packets needed between minimum number of non-congestion marked packets needed between
losses or ECN marks necessary to attain the target_data_rate over losses or ECN marks necessary to attain the target_data_rate over
a path with the specified target_RTT and target_MTU, as computed a path with the specified target_RTT and target_MTU, as computed
by a mathematical model of TCP congestion control. A reference by a mathematical model of TCP congestion control. A reference
calculation is shown in Section 6.2 and alternatives in Appendix A calculation is shown in Section 5.2 and alternatives in Appendix A
reference target_run_length: target_run_length computed precisely by reference target_run_length: target_run_length computed precisely by
the method in Section 6.2. This is likely to be more slightly the method in Section 5.2. This is likely to be more slightly
conservative than required by modern TCP algorithms. conservative than required by modern TCP implementations.
Ancillary parameters used for some tests Ancillary parameters used for some tests:
derating: Under some conditions the standard models are too derating: Under some conditions the standard models are too
conservative. The modeling framework permits some latitude in conservative. The modeling framework permits some latitude in
relaxing or "derating" some test parameters as described in relaxing or "derating" some test parameters as described in
Section 6.3 in exchange for a more stringent TDS validation Section 5.3 in exchange for a more stringent TDS validation
procedures, described in Section 12. procedures, described in Section 10.
subpath_data_rate: The maximum data rate supported by a subpath. subpath_IP_capacity: The IP capacity of a specific subpath.
This typically includes TCP/IP overhead, including all headers and
retransmits, etc. test path: A subpath of a complete path under test.
test_path_RTT: The RTT observed between two measurement points using test_path_RTT: The RTT observed between two measurement points using
packet sizes that are consistent with the transport protocol. packet sizes that are consistent with the transport protocol.
Generally MTU sized packets of the forward path, header_overhead Generally MTU sized packets of the forward path, header_overhead
sized packets on the return path. sized packets on the return path.
test_path_pipe: The amount of data necessary to fill a test path. test_path_pipe: The pipe size of a test path. Nominally the test
Nominally the test path RTT times the subpath_data_rate. path RTT times the test path IP_capacity.
test_window: The window necessary to meet the target_rate over a test_window: The window necessary to meet the target_rate over a
subpath. Typically test_window=target_data_rate*test_RTT/ test path. Typically test_window=target_data_rate*test_path_RTT/
(target_MTU - header_overhead). (target_MTU - header_overhead).
Tests can be grouped according to their applicability. The tests described in this note can be grouped according to their
applicability.
Capacity tests: determine if a network subpath has sufficient capacity tests: determine if a network subpath has sufficient
capacity to deliver the target performance. As long as the test capacity to deliver the Target Transport Performance. As long as
traffic is within the proper envelope for the target performance, the test traffic is within the proper envelope for the Target
the average packet losses or ECN marks must be below the threshold Transport Performance, the average packet losses or ECN marks must
computed by the model. As such, capacity tests reflect parameters be below the threshold computed by the model. As such, capacity
that can transition from passing to failing as a consequence of tests reflect parameters that can transition from passing to
cross traffic, additional presented load or the actions of other failing as a consequence of cross traffic, additional presented
network users. By definition, capacity tests also consume load or the actions of other network users. By definition,
significant network resources (data capacity and/or buffer space), capacity tests also consume significant network resources (data
and the test schedules must be balanced by their cost. capacity and/or queue buffer space), and the test schedules must
be balanced by their cost.
Monitoring tests: are designed to capture the most important aspects Monitoring tests: are designed to capture the most important aspects
of a capacity test, but without presenting excessive ongoing load of a capacity test, but without presenting excessive ongoing load
themselves. As such they may miss some details of the network's themselves. As such they may miss some details of the network's
performance, but can serve as a useful reduced-cost proxy for a performance, but can serve as a useful reduced-cost proxy for a
capacity test, for example to support ongoing monitoring. capacity test, for example to support ongoing monitoring.
Engineering tests: evaluate how network algorithms (such as AQM and Engineering tests: evaluate how network algorithms (such as AQM and
channel allocation) interact with TCP-style self clocked protocols channel allocation) interact with TCP-style self clocked protocols
and adaptive congestion control based on packet loss and ECN and adaptive congestion control based on packet loss and ECN
marks. These tests are likely to have complicated interactions marks. These tests are likely to have complicated interactions
with cross traffic and under some conditions can be inversely with cross traffic and under some conditions can be inversely
sensitive to load. For example a test to verify that an AQM sensitive to load. For example a test to verify that an AQM
algorithm causes ECN marks or packet drops early enough to limit algorithm causes ECN marks or packet drops early enough to limit
queue occupancy may experience a false pass result in the presence queue occupancy may experience a false pass result in the presence
of cross traffic. It is important that engineering tests be of cross traffic. It is important that engineering tests be
performed under a wide range of conditions, including both in situ performed under a wide range of conditions, including both in situ
and bench testing, and over a wide variety of load conditions. and bench testing, and over a wide variety of load conditions.
Ongoing monitoring is less likely to be useful for engineering Ongoing monitoring is less likely to be useful for engineering
tests, although sparse in situ testing might be appropriate. tests, although sparse in situ testing might be appropriate.
4. New requirements relative to RFC 2330 4. Background
Model Based Metrics are designed to fulfill some additional
requirement that were not recognized at the time RFC 2330 was written
[RFC2330]. These missing requirements may have significantly
contributed to policy difficulties in the IP measurement space. Some
additional requirements are:
o IP 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 should be spatially composable, such that measures of
concatenated paths should be predictable from subpaths.
o Metrics must be vantage point invariant over a significant range
of measurement point choices, including off path measurement
points. The only requirements on MP selection should be that the
portion of the test path that is not under test between the MP and
the part that is under test is effectively ideal, or is non ideal
in ways that can be calibrated out of the measurements and the
test RTT between the MPs is below some reasonable bound.
o Metric measurements must be repeatable by multiple parties with no
specialized access to MPs or diagnostic infrastructure. 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. Note
that vantage independence is key to this requirement.
5. Background
At the time the IPPM WG was chartered, sound Bulk Transport Capacity At the time the IPPM WG was chartered, sound Bulk Transport Capacity
measurement was known to be well beyond our capabilities. Even at measurement was known to be well beyond our capabilities. Even at
the time [RFC3148] was written we knew that we didn't fully the time [RFC3148] was written we knew that we didn't fully
understand the problem. Now, by hindsight we understand why BTC is understand the problem. Now, by hindsight we understand why BTC is
such a hard problem: such a hard problem:
o TCP is a control system with circular dependencies - everything o TCP is a control system with circular dependencies - everything
affects performance, including components that are explicitly not affects performance, including components that are explicitly not
part of the test. part of the test.
o Congestion control is an equilibrium process, such that transport o Congestion control is an equilibrium process, such that transport
protocols change the network (raise the loss ratio and/or RTT) to protocols change the network statistics (raise the packet loss
conform to their behavior. By design TCP congestion control keep ratio and/or RTT) to conform to their behavior. By design TCP
raising the data rate until the network give some indication that congestion control keep raising the data rate until the network
it is full by delaying, dropping or ECN marking packets. gives some indication that it is full by dropping or ECN marking
packets. If TCP successfully fills the network the packet loss
and ECN marks are mostly determined by TCP and how hard TCP drives
the network and not by the network itself.
o 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 roundtrips per second (i.e. proportional to the number of roundtrips per second (i.e.
inversely proportional to the RTT). As a consequence a flawed inversely proportional to the RTT). As a consequence a flawed
link may pass a short RTT local test even though it fails when the subpath may pass a short RTT local test even though it fails when
path is extended by a perfect network to some larger RTT. the path is extended by a perfect network to some larger RTT.
o TCP has a meta Heisenberg problem - Measurement and cross traffic o TCP has an extreme form of the Heisenberg problem - Measurement
interact in unknown and ill defined ways. The situation is and cross traffic interact in unknown and ill defined ways. The
actually worse than the traditional physics problem where you can situation is actually worse than the traditional physics problem
at least estimate bounds on the relative momentum of the where you can at least estimate bounds on the relative momentum of
measurement and measured particles. For network measurement you the measurement and measured particles. For network measurement
can not in general determine the relative "mass" of the you can not in general determine the relative "mass" of either the
measurement traffic and cross traffic, so you can not even gauge measurement traffic or the cross traffic, so you can not gauge the
the relative magnitude of their effects on each other. relative magnitude of the uncertainty that might be introduced by
any interaction.
These properties are a consequence of the equilibrium behavior These properties are a consequence of the equilibrium behavior
intrinsic to how all throughput optimizing protocols interact with intrinsic to how all throughput maximizing protocols interact with
the Internet. The protocols rely on control systems based on the Internet. These protocols rely on control systems based on
multiple network estimators to regulate the quantity of data traffic estimated network parameters to regulate the quantity of data traffic
sent into the network. The data traffic in turn alters network and sent into the network. The data traffic in turn alters network and
the properties observed by the estimators, such that there are the properties observed by the estimators, such that there are
circular dependencies between every component and every property. circular dependencies between every component and every property.
Since some of these properties are nonlinear, the entire system is Since some of these properties are nonlinear, the entire system is
nonlinear, and any change anywhere causes difficult to predict nonlinear, and any change anywhere causes difficult to predict
changes in every parameter. changes in every parameter.
Model Based Metrics overcome these problems by forcing the Model Based Metrics overcome these problems by forcing the
measurement system to be open loop: the delivery statistics (akin to measurement system to be open loop: the packet delivery statistics
the network estimators) do not affect the traffic or traffic patterns (akin to the network estimators) do not affect the traffic or traffic
(bursts), which computed on the basis of the target performance. In patterns (bursts), which computed on the basis of the Target
order for a network to pass, the resulting delivery statistics and Transport Performance. In order for a network to pass, the resulting
corresponding network estimators have to be such that they would not packet delivery statistics and corresponding network estimators have
cause the control systems slow the traffic below the target rate. to be such that they would not cause the control systems slow the
traffic below the target data rate.
5.1. TCP properties 4.1. TCP properties
TCP and SCTP are self clocked protocols. The dominant steady state TCP and SCTP are self clocked protocols. The dominant steady state
behavior is to have an approximately fixed quantity of data and behavior is to have an approximately fixed quantity of data and
acknowledgements (ACKs) circulating in the network. The receiver acknowledgements (ACKs) circulating in the network. The receiver
reports arriving data by returning ACKs to the data sender, the data reports arriving data by returning ACKs to the data sender, the data
sender typically responds by sending exactly the same quantity of sender typically responds by sending exactly the same quantity of
data back into the network. The total quantity of data plus the data data back into the network. The total quantity of data plus the data
represented by ACKs circulating in the network is referred to as the represented by ACKs circulating in the network is referred to as the
window. The mandatory congestion control algorithms incrementally window. The mandatory congestion control algorithms incrementally
adjust the window by sending slightly more or less data in response adjust the window by sending slightly more or less data in response
to each ACK. The fundamentally important property of this systems is to each ACK. The fundamentally important property of this system is
that it is entirely self clocked: The data transmissions are a that it is self clocked: The data transmissions are a reflection of
reflection of the ACKs that were delivered by the network, the ACKs the ACKs that were delivered by the network, the ACKs are a
are a reflection of the data arriving from the network. reflection of the data arriving from the network.
A number of phenomena can cause bursts of data, even in idealized A number of phenomena can cause bursts of data, even in idealized
networks that are modeled as simple queueing systems. networks that can be modeled as simple queueing systems.
During slowstart the data rate is doubled on each RTT by sending During slowstart the data rate is doubled on each RTT by sending
twice as much data as was delivered to the receiver on the prior RTT. twice as much data as was delivered to the receiver on the prior RTT.
For slowstart to be able to fill such a network the network must be For slowstart to be able to fill such a network the network must be
able to tolerate slowstart bursts up to the full pipe size inflated able to tolerate slowstart bursts up to the full pipe size inflated
by the anticipated window reduction on the first loss or ECN mark. by the anticipated window reduction on the first loss or ECN mark.
For example, with classic Reno congestion control, an optimal For example, with classic Reno congestion control, an optimal
slowstart has to end with a burst that is twice the bottleneck rate slowstart has to end with a burst that is twice the bottleneck rate
for exactly one RTT in duration. This burst causes a queue which is for exactly one RTT in duration. This burst causes a queue which is
exactly equal to the pipe size (i.e. the window is exactly twice the exactly equal to the pipe size (i.e. the window is exactly twice the
skipping to change at page 17, line 7 skipping to change at page 17, line 5
Other sources of bursts include application pauses and channel Other sources of bursts include application pauses and channel
allocation mechanisms. Appendix B describes the treatment of channel allocation mechanisms. Appendix B describes the treatment of channel
allocation systems. If the application pauses (stops reading or allocation systems. If the application pauses (stops reading or
writing data) for some fraction of one RTT, state-of-the-art TCP writing data) for some fraction of one RTT, state-of-the-art TCP
catches up to the earlier window size by sending a burst of data at catches up to the earlier window size by sending a burst of data at
the full sender interface rate. To fill such a network with a the full sender interface rate. To fill such a network with a
realistic application, the network has to be able to tolerate realistic application, the network has to be able to tolerate
interface rate bursts from the data sender large enough to cover interface rate bursts from the data sender large enough to cover
application pauses. application pauses.
Although the interface rate bursts are typically smaller than last Although the interface rate bursts are typically smaller than the
burst of a slowstart, they are at a higher data rate so they last burst of a slowstart, they are at a higher data rate so they
potentially exercise queues at arbitrary points along the front path potentially exercise queues at arbitrary points along the front path
from the data sender up to and including the queue at the dominant from the data sender up to and including the queue at the dominant
bottleneck. There is no model for how frequent or what sizes of bottleneck. There is no model for how frequent or what sizes of
sender rate bursts should be tolerated. sender rate bursts should be tolerated.
To verify that a path can meet a performance target, it is necessary To verify that a path can meet a Target Transport Performance, it is
to independently confirm that the path can tolerate bursts in the necessary to independently confirm that the path can tolerate bursts
dimensions that can be caused by these mechanisms. Three cases are in the dimensions that can be caused by these mechanisms. Three
likely to be sufficient: cases are likely to be sufficient:
o Slowstart bursts sufficient to get connections started properly. o Slowstart bursts sufficient to get connections started properly.
o Frequent sender interface rate bursts that are small enough where o Frequent sender interface rate bursts that are small enough where
they can be assumed not to significantly affect delivery they can be assumed not to significantly affect packet delivery
statistics. (Implicitly derated by selecting the burst size). statistics. (Implicitly derated by limiting the burst size).
o Infrequent sender interface rate full target_pipe_size bursts that o Infrequent sender interface rate full target_window_size bursts
do affect the delivery statistics. (Target_run_length may be that might affect the packet delivery statistics.
derated). (Target_run_length may be derated).
5.2. Diagnostic Approach
The MBM approach is to open loop TCP by precomputing traffic patterns
that are typically generated by TCP operating at the given target
parameters, and evaluating delivery statistics (packet loss, ECN
marks and delay). In this approach the measurement software
explicitly controls the data rate, transmission pattern or cwnd
(TCP's primary congestion control state variables) to create
repeatable traffic patterns that mimic TCP behavior but are
independent of the actual behavior of the subpath under test. 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.
By opening the protocol control loops, we remove most sources of
temporal and spatial correlation in the traffic delivery statistics,
such that each subpath's contribution to the end-to-end delivery
statistics can be assumed to be independent and stationary (The
delivery statistics depend on the fine structure of the data
transmissions, but not on long time scale state imbedded in the
sender, receiver or other network components.) Therefore each
subpath's contribution to the end-to-end delivery statistics can be
assumed to be independent, and spatial composition techniques such as
[RFC5835] and [RFC6049] apply.
In typical networks, the dominant bottleneck contributes the majority
of the packet loss and ECN marks. Often the rest of the path makes
insignificant contribution to these properties. A TDS should
apportion the end-to-end budget for the specified parameters
(primarily packet loss and ECN marks) to each subpath or group of
subpaths. For example the dominant bottleneck may be permitted to
contribute 90% of the loss budget, while the rest of the path is only
permitted to contribute 10%.
A TDS or FSTDS MUST apportion all relevant packet delivery statistics 4.2. Diagnostic Approach
between successive subpaths, such that the spatial composition of the
apportioned metrics will yield end-to-end delivery statistics which
are within the bounds determined by the models.
A network is expected to be able to sustain a Bulk TCP flow of a A complete path is expected to be able to sustain a Bulk TCP flow of
given data rate, MTU and RTT when all of the following conditions are a given data rate, MTU and RTT when all of the following conditions
met: are met:
1. The raw link rate is higher than the target data rate. See 1. The IP capacity is above the target data rate by sufficient
Section 10.1 or any number of data rate tests outside of MBM. margin to cover all TCP/IP overheads. See Section 8.1 or any
number of data rate tests outside of MBM.
2. The observed packet delivery statistics are better than required 2. The observed packet delivery statistics are better than required
by a suitable TCP performance model (e.g. fewer losses or ECN by a suitable TCP performance model (e.g. fewer losses or ECN
marks). See Section 10.1 or any number of low rate packet loss marks). See Section 8.1 or any number of low rate packet loss
tests outside of MBM. tests outside of MBM.
3. There is sufficient buffering at the dominant bottleneck to 3. There is sufficient buffering at the dominant bottleneck to
absorb a slowstart rate burst large enough to get the flow out of absorb a slowstart rate burst large enough to get the flow out of
slowstart at a suitable window size. See Section 10.3. slowstart at a suitable window size. See Section 8.3.
4. There is sufficient buffering in the front path to absorb and 4. There is sufficient buffering in the front path to absorb and
smooth sender interface rate bursts at all scales that are likely smooth sender interface rate bursts at all scales that are likely
to be generated by the application, any channel arbitration in to be generated by the application, any channel arbitration in
the ACK path or any other mechanisms. See Section 10.4. the ACK path or any other mechanisms. See Section 8.4.
5. When there is a standing queue at a bottleneck for a shared media 5. When there is a slowly rising standing queue at the bottleneck
the onset of packet loss has to be at an appropriate point (time
or queue depth) and progressive. See Section 8.2.
6. When there is a standing queue at a bottleneck for a shared media
subpath (e.g. half duplex), there are suitable bounds on how the subpath (e.g. half duplex), there are suitable bounds on how the
data and ACKs interact, for example due to the channel data and ACKs interact, for example due to the channel
arbitration mechanism. See Section 10.2.4. arbitration mechanism. See Section 8.2.4.
6. When there is a slowly rising standing queue at the bottleneck
the onset of packet loss has to be at an appropriate point (time
or queue depth) and progressive. See Section 10.2.
Note that conditions 1 through 4 require capacity tests for Note that conditions 1 through 4 require capacity tests for
confirmation, and thus need to be monitored on an ongoing basis. validation, and thus may need to be monitored on an ongoing basis.
Conditions 5 and 6 require engineering tests. They won't generally Conditions 5 and 6 require engineering tests best performed in
controlled environments such as a bench test. They won't generally
fail due to load, but may fail in the field due to configuration fail due to load, but may fail in the field due to configuration
errors, etc. and should be spot checked. errors, etc. and should be spot checked.
We are developing a tool that can perform many of the tests described We are developing a tool that can perform many of the tests described
here[MBMSource]. here [MBMSource].
6. Common Models and Parameters 4.3. New requirements relative to RFC 2330
6.1. Target End-to-end parameters Model Based Metrics are designed to fulfill some additional
requirements that were not recognized at the time RFC 2330 was
written [RFC2330]. These missing requirements may have significantly
contributed to policy difficulties in the IP measurement space. Some
additional requirements are:
o IP 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 should be spatially composable, such that measures of
concatenated paths should be predictable from subpaths.
o Metrics must be vantage point invariant over a significant range
of measurement point choices, including off path measurement
points. The only requirements on MP selection should be that the
RTT between the MPs is below some reasonable bound, and that the
effects of the "test leads" connecting MPs to the subpath under
test can be can be calibrated out of the measurements. The latter
might be be accomplished if the test leads are effectively ideal
or their properties can be deducted from the measurements between
the MPs. While many of tests require that the test leads have at
least as much IP capacity as the subpath under test, some do not,
for example Background Packet Delivery Tests described in
Section 8.1.3.
o Metric measurements must be repeatable by multiple parties with no
specialized access to MPs or diagnostic infrastructure. 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. Note
that vantage independence is key to meeting this requirement.
5. Common Models and Parameters
5.1. Target End-to-end parameters
The target end-to-end parameters are the target data rate, target RTT The target end-to-end parameters are the target data rate, target RTT
and target MTU as defined in Section 3. These parameters are and target MTU as defined in Section 3. These parameters are
determined by the needs of the application or the ultimate end user determined by the needs of the application or the ultimate end user
and the complete Internet path over which the application is expected and the complete Internet path over which the application is expected
to operate. The target parameters are in units that make sense to to operate. The target parameters are in units that make sense to
upper layers: payload bytes delivered to the application, above TCP. upper layers: payload bytes delivered to the application, above TCP.
They exclude overheads associated with TCP and IP headers, They exclude overheads associated with TCP and IP headers,
retransmits and other protocols (e.g. DNS). retransmits and other protocols (e.g. DNS).
Other end-to-end parameters defined in Section 3 include the Other end-to-end parameters defined in Section 3 include the
effective bottleneck data rate, the sender interface data rate and effective bottleneck data rate, the sender interface data rate and
the TCP/IP header sizes (overhead). the TCP and IP header sizes.
The target data rate must be smaller than all link data rates by The target_data_rate must be smaller than all subpath IP capacities
enough headroom to carry the transport protocol overhead, explicitly by enough headroom to carry the transport protocol overhead,
including retransmissions and an allowance for fluctuations in the explicitly including retransmissions and an allowance for
actual data rate, needed to meet the specified average rate. fluctuations in TCP's actual data rate. Specifying a
Specifying a target rate with insufficient headroom is likely to target_data_rate with insufficient headroom is likely to result in
result in brittle measurements having little predictive value. brittle measurements having little predictive value.
Note that the target parameters can be specified for a hypothetical Note that the target parameters can be specified for a hypothetical
path, for example to construct TDS designed for bench testing in the path, for example to construct TDS designed for bench testing in the
absence of a real application, or for a real physical test for in absence of a real application; or for a live in situ test of
situ testing of production infrastructure. production infrastructure.
The number of concurrent connections is explicitly not a parameter to The number of concurrent connections is explicitly not a parameter to
this model. If a subpath requires multiple connections in order to this model. If a subpath requires multiple connections in order to
meet the specified performance, that must be stated explicitly and meet the specified performance, that must be stated explicitly and
the procedure described in Section 7.4 applies. the procedure described in Section 6.4 applies.
6.2. Common Model Calculations 5.2. Common Model Calculations
The target transport performance is used to derive the The Target Transport Performance is used to derive the
target_pipe_size and the reference target_run_length. target_window_size and the reference target_run_length.
The target_pipe_size, is the average window size in packets needed to The target_window_size, is the average window size in packets needed
meet the target rate, for the specified target RTT and MTU. It is to meet the target_rate, for the specified target_RTT and target_MTU.
given by: It is given by:
target_pipe_size = ceiling( target_rate * target_RTT / ( target_MTU - target_window_size = ceiling( target_rate * target_RTT / ( target_MTU
header_overhead ) ) - header_overhead ) )
Target_run_length is an estimate of the minimum required number of Target_run_length is an estimate of the minimum required number of
unmarked packets that must be delivered between losses or ECN marks, unmarked packets that must be delivered between losses or ECN marks,
as computed by a mathematical model of TCP congestion control. The as computed by a mathematical model of TCP congestion control. The
derivation here follows [MSMO97], and by design is quite derivation here follows [MSMO97], and by design is quite
conservative. The alternate models described in Appendix A generally conservative.
yield smaller run_lengths (higher acceptable loss or ECN marking
rates), but may not apply in all situations. A FSTDS that uses an
alternate model MUST compare it to the reference target_run_length
computed here.
Reference target_run_length is derived as follows: assume the Reference target_run_length is derived as follows: assume the
subpath_data_rate is infinitesimally larger than the target_data_rate subpath_IP_capacity is infinitesimally larger than the
plus the required header_overhead. Then target_pipe_size also target_data_rate plus the required header_overhead. Then
predicts the onset of queueing. A larger window will cause a target_window_size also predicts the onset of queueing. A larger
standing queue at the bottleneck. window will cause a standing queue at the bottleneck.
Assume the transport protocol is using standard Reno style Additive Assume the transport protocol is using standard Reno style Additive
Increase, Multiplicative Decrease (AIMD) congestion control [RFC5681] Increase, Multiplicative Decrease (AIMD) congestion control [RFC5681]
(but not Appropriate Byte Counting [RFC3465]) and the receiver is (but not Appropriate Byte Counting [RFC3465]) and the receiver is
using standard delayed ACKs. Reno increases the window by one packet using standard delayed ACKs. Reno increases the window by one packet
every pipe_size worth of ACKs. With delayed ACKs this takes 2 Round every pipe_size worth of ACKs. With delayed ACKs this takes 2 Round
Trip Times per increase. To exactly fill the pipe, losses must be no Trip Times per increase. To exactly fill the pipe, losses must be no
closer than when the peak of the AIMD sawtooth reached exactly twice closer than when the peak of the AIMD sawtooth reached exactly twice
the target_pipe_size otherwise the multiplicative window reduction the target_window_size otherwise the multiplicative window reduction
triggered by the loss would cause the network to be underfilled. triggered by the loss would cause the network to be underfilled.
Following [MSMO97] the number of packets between losses must be the Following [MSMO97] the number of packets between losses must be the
area under the AIMD sawtooth. They must be no more frequent than area under the AIMD sawtooth. They must be no more frequent than
every 1 in ((3/2)*target_pipe_size)*(2*target_pipe_size) packets, every 1 in ((3/2)*target_window_size)*(2*target_window_size) packets,
which simplifies to: which simplifies to:
target_run_length = 3*(target_pipe_size^2) target_run_length = 3*(target_window_size^2)
Note that this calculation is very conservative and is based on a Note that this calculation is very conservative and is based on a
number of assumptions that may not apply. Appendix A discusses these number of assumptions that may not apply. Appendix A discusses these
assumptions and provides some alternative models. If a different assumptions and provides some alternative models. If a different
model is used, a fully specified TDS or FSTDS MUST document the model is used, a fully specified TDS or FSTDS MUST document the
actual method for computing target_run_length and ratio between actual method for computing target_run_length and ratio between
alternate target_run_length and the reference target_run_length alternate target_run_length and the reference target_run_length
calculated above, along with a discussion of the rationale for the calculated above, along with a discussion of the rationale for the
underlying assumptions. underlying assumptions.
These two parameters, target_pipe_size and target_run_length, These two parameters, target_window_size and target_run_length,
directly imply most of the individual parameters for the tests in directly imply most of the individual parameters for the tests in
Section 10. Section 8.
6.3. Parameter Derating 5.3. Parameter Derating
Since some aspects of the models are very conservative, the MBM Since some aspects of the models are very conservative, the MBM
framework permits some latitude in derating test parameters. Rather framework permits some latitude in derating test parameters. Rather
than trying to formalize more complicated models we permit some test than trying to formalize more complicated models we permit some test
parameters to be relaxed as long as they meet some additional parameters to be relaxed as long as they meet some additional
procedural constraints: procedural constraints:
o The TDS or FSTDS MUST document and justify the actual method used o The TDS or FSTDS MUST document and justify the actual method used
to compute the derated metric parameters. to compute the derated metric parameters.
o The validation procedures described in Section 12 must be used to o The validation procedures described in Section 10 must be used to
demonstrate the feasibility of meeting the performance targets demonstrate the feasibility of meeting the Target Transport
with infrastructure that infinitesimally passes the derated tests. Performance with infrastructure that infinitesimally passes the
o The validation process itself must be documented is such a way derated tests.
that other researchers can duplicate the validation experiments.
o The validation process for a FSTDS itself must be documented is
such a way that other researchers can duplicate the validation
experiments.
Except as noted, all tests below assume no derating. Tests where Except as noted, all tests below assume no derating. Tests where
there is not currently a well established model for the required there is not currently a well established model for the required
parameters explicitly include derating as a way to indicate parameters explicitly include derating as a way to indicate
flexibility in the parameters. flexibility in the parameters.
7. Traffic generating techniques 5.4. Test Preconditions
7.1. Paced transmission Many tests have preconditions which are required to assure their
validity. For example the presence or nonpresence of cross traffic
on specific subpaths, or appropriate preloading to put reactive
network elements into the proper states [RFC7312]. If preconditions
are not properly satisfied for some reason, the tests should be
considered to be inconclusive. In general it is useful to preserve
diagnostic information about why the preconditions were not met, and
any test data that was collected even if it is not useful for the
intended test. Such diagnostic information and partial test data may
be useful for improving the test in the future.
It is important to preserve the record that a test was scheduled,
because otherwise precondition enforcement mechanisms can introduce
sampling bias. For example, canceling tests due to cross traffic on
subscriber access links might introduce sampling bias in tests of the
rest of the network by reducing the number of tests during peak
network load.
Test preconditions and failure actions MUST be specified in a FSTDS.
6. Traffic generating techniques
6.1. Paced transmission
Paced (burst) transmissions: send bursts of data on a timer to meet a Paced (burst) transmissions: send bursts of data on a timer to meet a
particular target rate and pattern. In all cases the specified data particular target rate and pattern. In all cases the specified data
rate can either be the application or link rates. Header overheads rate can either be the application or IP rates. Header overheads
must be included in the calculations as appropriate. must be included in the calculations as appropriate.
Packet Headway: Time interval between packets, specified from the Packet Headway: Time interval between packets, specified from the
start of one to the start of the next. e.g. If packets are sent start of one to the start of the next. e.g. If packets are sent
with a 1 mS headway, there will be exactly 1000 packets per with a 1 mS headway, there will be exactly 1000 packets per
second. second.
Burst Headway: Time interval between bursts, specified from the Burst Headway: Time interval between bursts, specified from the
start of the first packet one burst to the start of the first start of the first packet one burst to the start of the first
packet of the next burst. e.g. If 4 packet bursts are sent with a packet of the next burst. e.g. If 4 packet bursts are sent with a
1 mS headway, there will be exactly 4000 packets per second. 1 mS burst headway, there will be exactly 4000 packets per second.
Paced single packets: Send individual packets at the specified rate Paced single packets: Send individual packets at the specified rate
or packet headway. [@@@@ Site RFC 3432, update definition?] or packet headway.
Paced Bursts: Send sender interface rate bursts on a timer. Specify Paced Bursts: Send sender interface rate bursts on a timer. Specify
any 3 of: average rate, packet size, burst size (number of any 3 of: average rate, packet size, burst size (number of
packets) and burst headway (burst start to start). The packet packets) and burst headway (burst start to start). The packet
headway within a burst is typically assumed to be the minimum headway within a burst is typically assumed to be the minimum
supported by the tester's interface. i.e. Bursts are normally supported by the tester's interface. i.e. Bursts are normally
sent as back-to-back packets. The packet headway within the sent as back-to-back packets. The packet headway within the
bursts can be explicitly specified. bursts can also be explicitly specified.
Slowstart bursts: Send 4 packet paced bursts at an average data rate Slowstart burst: Mimic TCP slowstart by sending 4 packet paced
equal to twice effective bottleneck link rate (but not more than bursts at an average data rate equal to twice the implied
the sender interface rate). This corresponds to the average rate bottleneck IP rate (but not more than the sender interface rate).
during a TCP slowstart when Appropriate Byte Counting [RFC3465] is If the implied bottleneck IP rate is more than half of the sender
present or delayed ack is disabled. Note that if the effective interface rate, slowstart rate bursts become sender interface rate
bottleneck link rate is more than half of the sender interface bursts. See the discussion and figure below.
rate, slowstart rate bursts become sender interface rate bursts. Repeated Slowstart bursts: Repeat Slowstart bursts once per
target_RTT. For TCP each burst would be twice as large as the
prior burst, and the sequence would end at the first ECN mark or
lost packet. For measurement, all slowstart bursts would be the
same size (nominally target_window_size but other sizes might be
specified). See the discussion and figure below.
[@@@@ Add figure --MM]. The slowstart bursts mimic TCP slowstart under a particular set of
Repeated Slowstart bursts: Slowstart bursts are typically part of implementation assumptions. The burst headway shown in Figure 2
larger scale pattern of repeated bursts, such as sending reflects the TCP self clock derived from the data passing through the
target_pipe_size packets as slowstart bursts on a target_RTT dominant bottleneck. The slow start burst size is nominally
headway (burst start to burst start). Such a stream has three target_window_size (so it might end with a bust that is less than 4
different average rates, depending on the averaging interval. At packets). The slowstart bursts are repeated every target_RTT. Note
the finest time scale the average rate is the same as the sender that a stream of repeated slowstart bursts has three different
interface rate, at a medium scale the average rate is twice the average rates, depending on the averaging interval. At the finest
effective bottleneck link rate and at the longest time scales the time scale (a few packet times at the sender interface) the peak of
average rate is equal to the target data rate. the average rate is the same as the sender interface rate; at a
medium scale (a few packet times at the dominant bottleneck) the peak
of the average rate is twice the implied bottleneck IP rate; and at
time scales longer than the target_RTT and when the burst size is
equal to the target_window_size the average rate is equal to the
target_data_rate. This pattern corresponds to repeating the last RTT
of TCP slowstart when delayed ACK and sender side byte counting are
present but without the limits specified in Appropriate Byte Counting
[RFC3465].
Note that in conventional measurement theory, exponential time --> ( - = one packet)
distributions are often used to eliminate many sorts of correlations. Packet stream:
For the procedures above, the correlations are created by the network
elements and accurately reflect their behavior. At some point in the
future, it will be desirable to introduce noise sources into the
above pacing models, but they are not warranted at this time.
7.2. Constant window pseudo CBR ---- ---- ---- ---- ---- ---- ---- ...
|<>| 4 packet sender interface rate bursts
|<--->| Burst headway
|<------------------------>| slowstart burst size
|<---------------------------------------------->| slowstart headway
\____________ _____________/ \______ __ ...
V V
One slowstart burst Repeated slowstart bursts
Slowstart Burst Structure
Figure 2
Note that in conventional measurement practice, exponentially
distributed intervals are often used to eliminate many sorts of
correlations. For the procedures above, the correlations are created
by the network or protocol elements and accurately reflect their
behavior. At some point in the future, it will be desirable to
introduce noise sources into the above pacing models, but they are
not warranted at this time.
6.2. Constant window pseudo CBR
Implement pseudo constant bit rate by running a standard protocol Implement pseudo constant bit rate by running a standard protocol
such as TCP with a fixed window size, such that it is self clocked. such as TCP with a fixed window size, such that it is self clocked.
Data packets arriving at the receiver trigger acknowledgements (ACKs) Data packets arriving at the receiver trigger acknowledgements (ACKs)
which travel back to the sender where they trigger additional which travel back to the sender where they trigger additional
transmissions. The window size is computed from the target_data_rate transmissions. The window size is computed from the target_data_rate
and the actual RTT of the test path. The rate is only maintained in and the actual RTT of the test path. The rate is only maintained in
average over each RTT, and is subject to limitations of the transport average over each RTT, and is subject to limitations of the transport
protocol. protocol.
Since the window size is constrained to be an integer number of Since the window size is constrained to be an integer number of
packets, for small RTTs or low data rates there may not be packets, for small RTTs or low data rates there may not be
sufficiently precise control over the data rate. Rounding the window sufficiently precise control over the data rate. Rounding the window
size up (the default) is likely to be result in data rates that are size up (the default) is likely to be result in data rates that are
higher than the target rate, but reducing the window by one packet higher than the target rate, but reducing the window by one packet
may result in data rates that are too small. Also cross traffic may result in data rates that are too small. Also cross traffic
potentially raises the RTT, implicitly reducing the rate. Cross potentially raises the RTT, implicitly reducing the rate. Cross
traffic that raises the RTT nearly always makes the test more traffic that raises the RTT nearly always makes the test more
strenuous. A FSTDS specifying a constant window CBR tests MUST strenuous. A FSTDS specifying a constant window CBR tests MUST
explicitly indicate under what conditions errors in the data cause explicitly indicate under what conditions errors in the data cause
tests to inconclusive. See the discussion of test outcomes in tests to inconclusive.
Section 8.1.
Since constant window pseudo CBR testing is sensitive to RTT Since constant window pseudo CBR testing is sensitive to RTT
fluctuations it can not accurately control the data rate in fluctuations it is less accurate at control the data rate in
environments with fluctuating delays. environments with fluctuating delays.
7.3. Scanned window pseudo CBR 6.3. Scanned window pseudo CBR
Scanned window pseudo CBR is similar to the constant window CBR Scanned window pseudo CBR is similar to the constant window CBR
described above, except the window is scanned across a range of sizes described above, except the window is scanned across a range of sizes
designed to include two key events, the onset of queueing and the designed to include two key events, the onset of queueing and the
onset of packet loss or ECN marks. The window is scanned by onset of packet loss or ECN marks. The window is scanned by
incrementing it by one packet every 2*target_pipe_size delivered incrementing it by one packet every 2*target_window_size delivered
packets. This mimics the additive increase phase of standard TCP packets. This mimics the additive increase phase of standard TCP
congestion avoidance when delayed ACKs are in effect. It normally congestion avoidance when delayed ACKs are in effect. Normally the
separates the the window increases by approximately twice the window increases separated by intervals slightly longer than twice
target_RTT. the target_RTT.
There are two ways to implement this test: one built by applying a There are two ways to implement this test: one built by applying a
window clamp to standard congestion control in a standard protocol window clamp to standard congestion control in a standard protocol
such as TCP and the other built by stiffening a non-standard such as TCP and the other built by stiffening a non-standard
transport protocol. When standard congestion control is in effect, transport protocol. When standard congestion control is in effect,
any losses or ECN marks cause the transport to revert to a window any losses or ECN marks cause the transport to revert to a window
smaller than the clamp such that the scanning clamp loses control the smaller than the clamp such that the scanning clamp loses control the
window size. The NPAD pathdiag tool is an example of this class of window size. The NPAD pathdiag tool is an example of this class of
algorithms [Pathdiag]. algorithms [Pathdiag].
Alternatively a non-standard congestion control algorithm can respond Alternatively a non-standard congestion control algorithm can respond
to losses by transmitting extra data, such that it maintains the to losses by transmitting extra data, such that it maintains the
specified window size independent of losses or ECN marks. Such a specified window size independent of losses or ECN marks. Such a
stiffened transport explicitly violates mandatory Internet congestion stiffened transport explicitly violates mandatory Internet congestion
control and is not suitable for in situ testing. [RFC5681] It is control and is not suitable for in situ testing. [RFC5681] It is
only appropriate for engineering testing under laboratory conditions. only appropriate for engineering testing under laboratory conditions.
The Windowed Ping tool implements such a test [WPING]. The tool The Windowed Ping tool implements such a test [WPING]. The tool
described in the paper has been updated.[mpingSource] described in the paper has been updated.[mpingSource]
The test procedures in Section 10.2 describe how to the partition the The test procedures in Section 8.2 describe how to the partition the
scans into regions and how to interpret the results. scans into regions and how to interpret the results.
7.4. Concurrent or channelized testing 6.4. Concurrent or channelized testing
The procedures described in this document are only directly The procedures described in this document are only directly
applicable to single stream performance measurement, e.g. one TCP applicable to single stream measurement, e.g. one TCP connection or
connection. In an ideal world, we would disallow all performance measurement stream. In an ideal world, we would disallow all
claims based multiple concurrent streams, but this is not practical performance claims based multiple concurrent streams, but this is not
due to at least two different issues. First, many very high rate practical due to at least two different issues. First, many very
link technologies are channelized and pin individual flows to high rate link technologies are channelized and at last partially pin
specific channels to minimize reordering or other problems and the flow to channel mapping to minimize packet reordering within
second, TCP itself has scaling limits. Although the former problem flows. Second, TCP itself has scaling limits. Although the former
might be overcome through different design decisions, the later problem might be overcome through different design decisions, the
problem is more deeply rooted. later problem is more deeply rooted.
All congestion control algorithms that are philosophically aligned All congestion control algorithms that are philosophically aligned
with the standard [RFC5681] (e.g. claim some level of TCP with the standard [RFC5681] (e.g. claim some level of TCP
friendliness) have scaling limits, in the sense that as a long fast compatibility, friendliness or fairness) have scaling limits, in the
network (LFN) with a fixed RTT and MTU gets faster, these congestion sense that as a long fast network (LFN) with a fixed RTT and MTU gets
control algorithms get less accurate and as a consequence have faster, these congestion control algorithms get less accurate and as
difficulty filling the network[CCscaling]. These properties are a a consequence have difficulty filling the network[CCscaling]. These
consequence of the original Reno AIMD congestion control design and properties are a consequence of the original Reno AIMD congestion
the requirement in [RFC5681] that all transport protocols have control design and the requirement in [RFC5681] that all transport
uniform response to congestion. protocols have similar responses to congestion.
There are a number of reasons to want to specify performance in term There are a number of reasons to want to specify performance in term
of multiple concurrent flows, however this approach is not of multiple concurrent flows, however this approach is not
recommended for data rates below several megabits per second, which recommended for data rates below several megabits per second, which
can be attained with run lengths under 10000 packets. Since the can be attained with run lengths under 10000 packets on many paths.
required run length goes as the square of the data rate, at higher Since the required run length goes as the square of the data rate, at
rates the run lengths can be unreasonably large, and multiple higher rates the run lengths can be unreasonably large, and multiple
connection might be the only feasible approach. flows might be the only feasible approach.
If multiple connections are deemed necessary to meet aggregate If multiple flows are deemed necessary to meet aggregate performance
performance targets then this MUST be stated both the design of the targets then this MUST be stated both the design of the TDS and in
TDS and in any claims about network performance. The tests MUST be any claims about network performance. The IP diagnostic tests MUST
performed concurrently with the specified number of connections. For be performed concurrently with the specified number of connections.
the the tests that use bursty traffic, the bursts should be For the the tests that use bursty traffic, the bursts should be
synchronized across flows. synchronized across flows.
8. Interpreting the Results 7. Interpreting the Results
8.1. Test outcomes 7.1. Test outcomes
To perform an exhaustive test of a complete network path, each test To perform an exhaustive test of a complete network path, each test
of the TDS is applied to each subpath of the complete path. If any of the TDS is applied to each subpath of the complete path. If any
subpath fails any test then an application running over the complete subpath fails any test then a standard transport protocol running
path can also be expected to fail to attain the target performance over the complete path can also be expected to fail to attain the
under some conditions. Target Transport Performance under some conditions.
In addition to passing or failing, a test can be deemed to be In addition to passing or failing, a test can be deemed to be
inconclusive for a number of reasons. Proper instrumentation and inconclusive for a number of reasons. Proper instrumentation and
treatment of inconclusive outcomes is critical to the accuracy and treatment of inconclusive outcomes is critical to the accuracy and
robustness of Model Based Metrics. Tests can be inconclusive if the robustness of Model Based Metrics. Tests can be inconclusive if the
precomputed traffic pattern or data rates were not accurately precomputed traffic pattern or data rates were not accurately
generated; the measurement results were not statistically generated; the measurement results were not statistically
significant; and others causes such as failing to meet some required significant; and others causes such as failing to meet some required
preconditions for the test. preconditions for the test. See Section 5.4
For example consider a test that implements Constant Window Pseudo For example consider a test that implements Constant Window Pseudo
CBR (Section 7.2) by adding rate controls and detailed traffic CBR (Section 6.2) by adding rate controls and detailed traffic
instrumentation to TCP (e.g. [RFC4898]). TCP includes built in instrumentation to TCP (e.g. [RFC4898]). TCP includes built in
control systems which might interfere with the sending data rate. If control systems which might interfere with the sending data rate. If
such a test meets the required delivery statistics (e.g. run length) such a test meets the required packet delivery statistics (e.g. run
while failing to attain the specified data rate it must be treated as length) while failing to attain the specified data rate it must be
an inconclusive result, because we can not a priori determine if the treated as an inconclusive result, because we can not a priori
reduced data rate was caused by a TCP problem or a network problem, determine if the reduced data rate was caused by a TCP problem or a
or if the reduced data rate had a material effect on the observed network problem, or if the reduced data rate had a material effect on
delivery statistics. the observed packet delivery statistics.
Note that for capacity tests, if the observed delivery statistics Note that for capacity tests, if the observed packet delivery
fail to meet the targets, the test can can be considered to have statistics meet the statistical criteria for failing (accepting
hypnosis H1 in Section 7.2), the test can can be considered to have
failed because it doesn't really matter that the test didn't attain failed because it doesn't really matter that the test didn't attain
the required data rate. the required data rate.
The really important new properties of MBM, such as vantage The really important new properties of MBM, such as vantage
independence, are a direct consequence of opening the control loops independence, are a direct consequence of opening the control loops
in the protocols, such that the test traffic does not depend on in the protocols, such that the test traffic does not depend on
network conditions or traffic received. Any mechanism that network conditions or traffic received. Any mechanism that
introduces feedback between the paths measurements and the traffic introduces feedback between the paths measurements and the traffic
generation is at risk of introducing nonlinearities that spoil these generation is at risk of introducing nonlinearities that spoil these
properties. Any exceptional event that indicates that such feedback properties. Any exceptional event that indicates that such feedback
has happened should cause the test to be considered inconclusive. has happened should cause the test to be considered inconclusive.
One way to view inconclusive tests is that they reflect situations One way to view inconclusive tests is that they reflect situations
where a test outcome is ambiguous between limitations of the network where a test outcome is ambiguous between limitations of the network
and some unknown limitation of the diagnostic test itself, which may and some unknown limitation of the IP diagnostic test itself, which
have been caused by some uncontrolled feedback from the network. may have been caused by some uncontrolled feedback from the network.
Note that procedures that attempt to sweep the target parameter space Note that procedures that attempt to sweep the target parameter space
to find the limits on some parameter such as target_data_rate are at to find the limits on some parameter such as target_data_rate are at
risk of breaking the location independent properties of Model Based risk of breaking the location independent properties of Model Based
Metrics, if the boundary between passing and inconclusive is at all Metrics, if any part of the boundary between passing and inconclusive
sensitive to RTT. is sensitive to RTT (which is normally the case).
One of the goals for evolving TDS designs will be to keep sharpening One of the goals for evolving TDS designs will be to keep sharpening
distinction between inconclusive, passing and failing tests. The distinction between inconclusive, passing and failing tests. The
criteria for for passing, failing and inconclusive tests MUST be criteria for for passing, failing and inconclusive tests MUST be
explicitly stated for every test in the TDS or FSTDS. explicitly stated for every test in the TDS or FSTDS.
One of the goals of evolving the testing process, procedures, tools One of the goals of evolving the testing process, procedures, tools
and measurement point selection should be to minimize the number of and measurement point selection should be to minimize the number of
inconclusive tests. inconclusive tests.
It may be useful to keep raw data delivery statistics for deeper It may be useful to keep raw packet delivery statistics and ancillary
study of the behavior of the network path and to measure the tools metrics [RFC3148] for deeper study of the behavior of the network
themselves. Raw delivery statistics can help to drive tool path and to measure the tools themselves. Raw packet delivery
evolution. Under some conditions it might be possible to reevaluate statistics can help to drive tool evolution. Under some conditions
the raw data for satisfying alternate performance targets. However it might be possible to reevaluate the raw data for satisfying
it is important to guard against sampling bias and other implicit alternate Target Transport Performance. However it is important to
feedback which can cause false results and exhibit measurement point guard against sampling bias and other implicit feedback which can
vantage sensitivity. cause false results and exhibit measurement point vantage
sensitivity. Simply applying different delivery criteria based on a
different Target Transport Performance is insufficient if the test
traffic patterns (bursts, etc) does not match the alternate Target
Transport Performance.
8.2. Statistical criteria for estimating run_length 7.2. Statistical criteria for estimating run_length
When evaluating the observed run_length, we need to determine When evaluating the observed run_length, we need to determine
appropriate packet stream sizes and acceptable error levels for appropriate packet stream sizes and acceptable error levels for
efficient measurement. In practice, can we compare the empirically efficient measurement. In practice, can we compare the empirically
estimated packet loss and ECN marking ratios with the targets as the estimated packet loss and ECN marking ratios with the targets as the
sample size grows? How large a sample is needed to say that the sample size grows? How large a sample is needed to say that the
measurements of packet transfer indicate a particular run length is measurements of packet transfer indicate a particular run length is
present? present?
The generalized measurement can be described as recursive testing: The generalized measurement can be described as recursive testing:
send packets (individually or in patterns) and observe the packet send packets (individually or in patterns) and observe the packet
delivery performance (loss ratio or other metric, any marking we delivery performance (packet loss ratio or other metric, any marking
define). we define).
As each packet is sent and measured, we have an ongoing estimate of As each packet is sent and measured, we have an ongoing estimate of
the performance in terms of the ratio of packet loss or ECN mark to the performance in terms of the ratio of packet loss or ECN mark to
total packets (i.e. an empirical probability). We continue to send total packets (i.e. an empirical probability). We continue to send
until conditions support a conclusion or a maximum sending limit has until conditions support a conclusion or a maximum sending limit has
been reached. been reached.
We have a target_mark_probability, 1 mark per target_run_length, We have a target_mark_probability, 1 mark per target_run_length,
where a "mark" is defined as a lost packet, a packet with ECN mark, where a "mark" is defined as a lost packet, a packet with ECN mark,
or other signal. This constitutes the null Hypothesis: or other signal. This constitutes the null Hypothesis:
H0: no more than one mark in target_run_length = H0: no more than one mark in target_run_length =
3*(target_pipe_size)^2 packets 3*(target_window_size)^2 packets
and we can stop sending packets if on-going measurements support and we can stop sending packets if on-going measurements support
accepting H0 with the specified Type I error = alpha (= 0.05 for accepting H0 with the specified Type I error = alpha (= 0.05 for
example). example).
We also have an alternative Hypothesis to evaluate: if performance is We also have an alternative Hypothesis to evaluate: if performance is
significantly lower than the target_mark_probability. Based on significantly lower than the target_mark_probability. Based on
analysis of typical values and practical limits on measurement analysis of typical values and practical limits on measurement
duration, we choose four times the H0 probability: duration, we choose four times the H0 probability:
skipping to change at page 27, line 49 skipping to change at page 29, line 12
Analysis [Rtool] , in the add-on package for Cross-Validation via Analysis [Rtool] , in the add-on package for Cross-Validation via
Sequential Testing (CVST) [CVST] . Sequential Testing (CVST) [CVST] .
Using the equations above, we can calculate the minimum number of Using the equations above, we can calculate the minimum number of
packets (n) needed to accept H0 when x defects are observed. For packets (n) needed to accept H0 when x defects are observed. For
example, when x = 0: example, when x = 0:
Xa = 0 = -h1 + s*n Xa = 0 = -h1 + s*n
and n = h1 / s and n = h1 / s
8.3. Reordering Tolerance 7.3. Reordering Tolerance
All tests must be instrumented for packet level reordering [RFC4737]. All tests must be instrumented for packet level reordering [RFC4737].
However, there is no consensus for how much reordering should be However, there is no consensus for how much reordering should be
acceptable. Over the last two decades the general trend has been to acceptable. Over the last two decades the general trend has been to
make protocols and applications more tolerant to reordering (see for make protocols and applications more tolerant to reordering (see for
example [RFC4015]), in response to the gradual increase in reordering example [RFC4015]), in response to the gradual increase in reordering
in the network. This increase has been due to the deployment of in the network. This increase has been due to the deployment of
technologies such as multi threaded routing lookups and Equal Cost technologies such as multi threaded routing lookups and Equal Cost
MultiPath (ECMP) routing. These techniques increase parallelism in MultiPath (ECMP) routing. These techniques increase parallelism in
network and are critical to enabling overall Internet growth to network and are critical to enabling overall Internet growth to
skipping to change at page 28, line 31 skipping to change at page 29, line 43
By implication, recording which is less than these bounds should not By implication, recording which is less than these bounds should not
be treated as a network impairment. However [RFC4737] still applies: be treated as a network impairment. However [RFC4737] still applies:
reordering should be instrumented and the maximum reordering that can reordering should be instrumented and the maximum reordering that can
be properly characterized by the test (e.g. bound on history buffers) be properly characterized by the test (e.g. bound on history buffers)
should be recorded with the measurement results. should be recorded with the measurement results.
Reordering tolerance and diagnostic limitations, such as history Reordering tolerance and diagnostic limitations, such as history
buffer size, MUST be specified in a FSTDS. buffer size, MUST be specified in a FSTDS.
9. Test Preconditions 8. Diagnostic Tests
Many tests have preconditions which are required to assure their
validity. For example the presence or nonpresence of cross traffic
on specific subpaths, or appropriate preloading to put reactive
network elements into the proper states[RFC7312]). If preconditions
are not properly satisfied for some reason, the tests should be
considered to be inconclusive. In general it is useful to preserve
diagnostic information about why the preconditions were not met, and
any test data that was collected even if it is not useful for the
intended test. Such diagnostic information and partial test data may
be useful for improving the test in the future.
It is important to preserve the record that a test was scheduled,
because otherwise precondition enforcement mechanisms can introduce
sampling bias. For example, canceling tests due to cross traffic on
subscriber access links might introduce sampling bias of tests of the
rest of the network by reducing the number of tests during peak
network load.
Test preconditions and failure actions MUST be specified in a FSTDS.
10. Diagnostic Tests
The diagnostic tests below are organized by traffic pattern: basic The IP diagnostic tests below are organized by traffic pattern: basic
data rate and delivery statistics, standing queues, slowstart bursts, data rate and packet delivery statistics, standing queues, slowstart
and sender rate bursts. We also introduce some combined tests which bursts, and sender rate bursts. We also introduce some combined
are more efficient when networks are expected to pass, but conflate tests which are more efficient when networks are expected to pass,
diagnostic signatures when they fail. but conflate diagnostic signatures when they fail.
There are a number of test details which are not fully defined here. There are a number of test details which are not fully defined here.
They must be fully specified in a FSTDS. From a standardization They must be fully specified in a FSTDS. From a standardization
perspective, this lack of specificity will weaken this version of perspective, this lack of specificity will weaken this version of
Model Based Metrics, however it is anticipated that this it be more Model Based Metrics, however it is anticipated that this it be more
than offset by the extent to which MBM suppresses the problems caused than offset by the extent to which MBM suppresses the problems caused
by using transport protocols for measurement. e.g. non-specific MBM by using transport protocols for measurement. e.g. non-specific MBM
metrics are likely to have better repeatability than many existing metrics are likely to have better repeatability than many existing
BTC like metrics. Once we have good field experience, the missing BTC like metrics. Once we have good field experience, the missing
details can be fully specified. details can be fully specified.
10.1. Basic Data Rate and Delivery Statistics Tests 8.1. Basic Data Rate and Packet Delivery Tests
We propose several versions of the basic data rate and delivery We propose several versions of the basic data rate and packet
statistics test. All measure the number of packets delivered between delivery statistics test. All measure the number of packets
losses or ECN marks, using a data stream that is rate controlled at delivered between losses or ECN marks, using a data stream that is
or below the target_data_rate. rate controlled at or below the target_data_rate.
The tests below differ in how the data rate is controlled. The data The tests below differ in how the data rate is controlled. The data
can be paced on a timer, or window controlled at full target data can be paced on a timer, or window controlled at full target data
rate. The first two tests implicitly confirm that sub_path has rate. The first two tests implicitly confirm that sub_path has
sufficient raw capacity to carry the target_data_rate. They are sufficient raw capacity to carry the target_data_rate. They are
recommend for relatively infrequent testing, such as an installation recommend for relatively infrequent testing, such as an installation
or periodic auditing process. The third, background delivery or periodic auditing process. The third, background packet delivery
statistics, is a low rate test designed for ongoing monitoring for statistics, is a low rate test designed for ongoing monitoring for
changes in subpath quality. changes in subpath quality.
All rely on the receiver accumulating packet delivery statistics as All rely on the receiver accumulating packet delivery statistics as
described in Section 8.2 to score the outcome: described in Section 7.2 to score the outcome:
Pass: it is statistically significant that the observed interval Pass: it is statistically significant that the observed interval
between losses or ECN marks is larger than the target_run_length. between losses or ECN marks is larger than the target_run_length.
Fail: it is statistically significant that the observed interval Fail: it is statistically significant that the observed interval
between losses or ECN marks is smaller than the target_run_length. between losses or ECN marks is smaller than the target_run_length.
A test is considered to be inconclusive if it failed to meet the data A test is considered to be inconclusive if it failed to meet the data
rate as specified below, meet the qualifications defined in Section 9 rate as specified below, meet the qualifications defined in
or neither run length statistical hypothesis was confirmed in the Section 5.4 or neither run length statistical hypothesis was
allotted test duration. confirmed in the allotted test duration.
10.1.1. Delivery Statistics at Paced Full Data Rate 8.1.1. Delivery Statistics at Paced Full Data Rate
Confirm that the observed run length is at least the Confirm that the observed run length is at least the
target_run_length while relying on timer to send data at the target_run_length while relying on timer to send data at the
target_rate using the procedure described in in Section 7.1 with a target_rate using the procedure described in in Section 6.1 with a
burst size of 1 (single packets) or 2 (packet pairs). burst size of 1 (single packets) or 2 (packet pairs).
The test is considered to be inconclusive if the packet transmission The test is considered to be inconclusive if the packet transmission
can not be accurately controlled for any reason. can not be accurately controlled for any reason.
RFC 6673 [RFC6673] is appropriate for measuring delivery statistics RFC 6673 [RFC6673] is appropriate for measuring packet delivery
at full data rate. statistics at full data rate.
10.1.2. Delivery Statistics at Full Data Windowed Rate 8.1.2. Delivery Statistics at Full Data Windowed Rate
Confirm that the observed run length is at least the Confirm that the observed run length is at least the
target_run_length while sending at an average rate approximately target_run_length while sending at an average rate approximately
equal to the target_data_rate, by controlling (or clamping) the equal to the target_data_rate, by controlling (or clamping) the
window size of a conventional transport protocol to a fixed value window size of a conventional transport protocol to a fixed value
computed from the properties of the test path, typically computed from the properties of the test path, typically
test_window=target_data_rate*test_RTT/target_MTU. Note that if there test_window=target_data_rate*test_path_RTT/target_MTU. Note that if
is any interaction between the forward and return path, test_window there is any interaction between the forward and return path,
may need to be adjusted slightly to compensate for the resulting test_window may need to be adjusted slightly to compensate for the
inflated RTT. resulting inflated RTT.
Since losses and ECN marks generally cause transport protocols to at Since losses and ECN marks generally cause transport protocols to at
least temporarily reduce their data rates, this test is expected to least temporarily reduce their data rates, this test is expected to
be less precise about controlling its data rate. It should not be be less precise about controlling its data rate. It should not be
considered inconclusive as long as at least some of the round trips considered inconclusive as long as at least some of the round trips
reached the full target_data_rate without incurring losses or ECN reached the full target_data_rate without incurring losses or ECN
marks. To pass this test the network MUST deliver target_pipe_size marks. To pass this test the network MUST deliver target_window_size
packets in target_RTT time without any losses or ECN marks at least packets in target_RTT time without any losses or ECN marks at least
once per two target_pipe_size round trips, in addition to meeting the once per two target_window_size round trips, in addition to meeting
run length statistical test. the run length statistical test.
10.1.3. Background Delivery Statistics Tests 8.1.3. Background Packet Delivery Statistics Tests
The background run length is a low rate version of the target target The background run length is a low rate version of the target target
rate test above, designed for ongoing lightweight monitoring for rate test above, designed for ongoing lightweight monitoring for
changes in the observed subpath run length without disrupting users. changes in the observed subpath run length without disrupting users.
It should be used in conjunction with one of the above full rate It should be used in conjunction with one of the above full rate
tests because it does not confirm that the subpath can support raw tests because it does not confirm that the subpath can support raw
data rate. data rate.
RFC 6673 [RFC6673] is appropriate for measuring background delivery RFC 6673 [RFC6673] is appropriate for measuring background packet
statistics. delivery statistics.
10.2. Standing Queue Tests 8.2. Standing Queue Tests
These engineering tests confirm that the bottleneck is well behaved These engineering tests confirm that the bottleneck is well behaved
across the onset of packet loss, which typically follows after the across the onset of packet loss, which typically follows after the
onset of queueing. Well behaved generally means lossless for onset of queueing. Well behaved generally means lossless for
transient queues, but once the queue has been sustained for a transient queues, but once the queue has been sustained for a
sufficient period of time (or reaches a sufficient queue depth) there sufficient period of time (or reaches a sufficient queue depth) there
should be a small number of losses to signal to the transport should be a small number of losses to signal to the transport
protocol that it should reduce its window. Losses that are too early protocol that it should reduce its window. Losses that are too early
can prevent the transport from averaging at the target_data_rate. can prevent the transport from averaging at the target_data_rate.
Losses that are too late indicate that the queue might be subject to Losses that are too late indicate that the queue might be subject to
bufferbloat [wikiBloat] and inflict excess queuing delays on all bufferbloat [wikiBloat] and inflict excess queuing delays on all
flows sharing the bottleneck queue. Excess losses (more than half of flows sharing the bottleneck queue. Excess losses (more than half of
the window) at the onset of congestion make loss recovery problematic the window) at the onset of congestion make loss recovery problematic
for the transport protocol. Non-linear, erratic or excessive RTT for the transport protocol. Non-linear, erratic or excessive RTT
increases suggest poor interactions between the channel acquisition increases suggest poor interactions between the channel acquisition
algorithms and the transport self clock. All of the tests in this algorithms and the transport self clock. All of the tests in this
section use the same basic scanning algorithm, described here, but section use the same basic scanning algorithm, described here, but
score the link on the basis of how well it avoids each of these score the link or subpath on the basis of how well it avoids each of
problems. these problems.
For some technologies the data might not be subject to increasing For some technologies the data might not be subject to increasing
delays, in which case the data rate will vary with the window size delays, in which case the data rate will vary with the window size
all the way up to the onset of load induced losses or ECN marks. For all the way up to the onset of load induced losses or ECN marks. For
theses technologies, the discussion of queueing does not apply, but theses technologies, the discussion of queueing does not apply, but
it is still required that the onset of losses or ECN marks be at an it is still required that the onset of losses or ECN marks be at an
appropriate point and progressive. appropriate point and progressive.
Use the procedure in Section 7.3 to sweep the window across the onset Use the procedure in Section 6.3 to sweep the window across the onset
of queueing and the onset of loss. The tests below all assume that of queueing and the onset of loss. The tests below all assume that
the scan emulates standard additive increase and delayed ACK by the scan emulates standard additive increase and delayed ACK by
incrementing the window by one packet for every 2*target_pipe_size incrementing the window by one packet for every 2*target_window_size
packets delivered. A scan can typically be divided into three packets delivered. A scan can typically be divided into three
regions: below the onset of queueing, a standing queue, and at or regions: below the onset of queueing, a standing queue, and at or
beyond the onset of loss. beyond the onset of loss.
Below the onset of queueing the RTT is typically fairly constant, and Below the onset of queueing the RTT is typically fairly constant, and
the data rate varies in proportion to the window size. Once the data the data rate varies in proportion to the window size. Once the data
rate reaches the link rate, the data rate becomes fairly constant, rate reaches the subpath IP rate, the data rate becomes fairly
and the RTT increases in proportion to the increase in window size. constant, and the RTT increases in proportion to the increase in
The precise transition across the start of queueing can be identified window size. The precise transition across the start of queueing can
by the maximum network power, defined to be the ratio data rate over be identified by the maximum network power, defined to be the ratio
the RTT. The network power can be computed at each window size, and data rate over the RTT. The network power can be computed at each
the window with the maximum are taken as the start of the queueing window size, and the window with the maximum are taken as the start
region. of the queueing region.
For technologies that do not have conventional queues, start the scan For technologies that do not have conventional queues, start the scan
at a window equal to the test_window=target_data_rate*test_RTT/ at a window equal to the test_window=target_data_rate*test_path_RTT/
target_MTU, i.e. starting at the target rate, instead of the power target_MTU, i.e. starting at the target rate, instead of the power
point. point.
If there is random background loss (e.g. bit errors, etc), precise If there is random background loss (e.g. bit errors, etc), precise
determination of the onset of queue induced packet loss may require determination of the onset of queue induced packet loss may require
multiple scans. Above the onset of queuing loss, all transport multiple scans. Above the onset of queuing loss, all transport
protocols are expected to experience periodic losses determined by protocols are expected to experience periodic losses determined by
the interaction between the congestion control and AQM algorithms. the interaction between the congestion control and AQM algorithms.
For standard congestion control algorithms the periodic losses are For standard congestion control algorithms the periodic losses are
likely to be relatively widely spaced and the details are typically likely to be relatively widely spaced and the details are typically
dominated by the behavior of the transport protocol itself. For the dominated by the behavior of the transport protocol itself. For the
stiffened transport protocols case (with non-standard, aggressive stiffened transport protocols case (with non-standard, aggressive
congestion control algorithms) the details of periodic losses will be congestion control algorithms) the details of periodic losses will be
dominated by how the the window increase function responds to loss. dominated by how the the window increase function responds to loss.
10.2.1. Congestion Avoidance 8.2.1. Congestion Avoidance
A link passes the congestion avoidance standing queue test if more A subpath passes the congestion avoidance standing queue test if more
than target_run_length packets are delivered between the onset of than target_run_length packets are delivered between the onset of
queueing (as determined by the window with the maximum network power) queueing (as determined by the window with the maximum network power)
and the first loss or ECN mark. If this test is implemented using a and the first loss or ECN mark. If this test is implemented using a
standards congestion control algorithm with a clamp, it can be standards congestion control algorithm with a clamp, it can be
performed in situ in the production internet as a capacity test. For performed in situ in the production internet as a capacity test. For
an example of such a test see [Pathdiag]. an example of such a test see [Pathdiag].
For technologies that do not have conventional queues, use the For technologies that do not have conventional queues, use the
test_window inplace of the onset of queueing. i.e. A link passes the test_window inplace of the onset of queueing. i.e. A subpath passes
congestion avoidance standing queue test if more than the congestion avoidance standing queue test if more than
target_run_length packets are delivered between start of the scan at target_run_length packets are delivered between start of the scan at
test_window and the first loss or ECN mark. test_window and the first loss or ECN mark.
10.2.2. Bufferbloat 8.2.2. Bufferbloat
This test confirms that there is some mechanism to limit buffer This test confirms that there is some mechanism to limit buffer
occupancy (e.g. that prevents bufferbloat). Note that this is not occupancy (e.g. that prevents bufferbloat). Note that this is not
strictly a requirement for single stream bulk performance, however if strictly a requirement for single stream bulk transport capacity,
there is no mechanism to limit buffer queue occupancy then a single however if there is no mechanism to limit buffer queue occupancy then
stream with sufficient data to deliver is likely to cause the a single stream with sufficient data to deliver is likely to cause
problems described in [RFC2309], [I-D.ietf-aqm-recommendation] and the problems described in [RFC2309], [I-D.ietf-aqm-recommendation]
[wikiBloat]. This may cause only minor symptoms for the dominant and [wikiBloat]. This may cause only minor symptoms for the dominant
flow, but has the potential to make the link unusable for other flows flow, but has the potential to make the subpath unusable for other
and applications. flows and applications.
Pass if the onset of loss occurs before a standing queue has Pass if the onset of loss occurs before a standing queue has
introduced more delay than than twice target_RTT, or other well introduced more delay than than twice target_RTT, or other well
defined and specified limit. Note that there is not yet a model for defined and specified limit. Note that there is not yet a model for
how much standing queue is acceptable. The factor of two chosen here how much standing queue is acceptable. The factor of two chosen here
reflects a rule of thumb. In conjunction with the previous test, reflects a rule of thumb. In conjunction with the previous test,
this test implies that the first loss should occur at a queueing this test implies that the first loss should occur at a queueing
delay which is between one and two times the target_RTT. delay which is between one and two times the target_RTT.
Specified RTT limits that are larger than twice the target_RTT must Specified RTT limits that are larger than twice the target_RTT must
be fully justified in the FSTDS. be fully justified in the FSTDS.
10.2.3. Non excessive loss 8.2.3. Non excessive loss
This test confirm that the onset of loss is not excessive. Pass if This test confirm that the onset of loss is not excessive. Pass if
losses are equal or less than the increase in the cross traffic plus losses are equal or less than the increase in the cross traffic plus
the test traffic window increase on the previous RTT. This could be the test traffic window increase on the previous RTT. This could be
restated as non-decreasing link throughput at the onset of loss, restated as non-decreasing subpath throughput at the onset of loss,
which is easy to meet as long as discarding packets in not more which is easy to meet as long as discarding packets is not more
expensive than delivering them. (Note when there is a transient drop expensive than delivering them. (Note when there is a transient drop
in link throughput, outside of a standing queue test, a link that in subpath throughput, outside of a standing queue test, a subpath
passes other queue tests in this document will have sufficient queue that passes other queue tests in this document will have sufficient
space to hold one RTT worth of data). queue space to hold one RTT worth of data).
Note that conventional Internet traffic policers will not pass this Note that conventional Internet traffic policers will not pass this
test, which is correct. TCP often fails to come into equilibrium at test, which is correct. TCP often fails to come into equilibrium at
more than a small fraction of the available capacity, if the capacity more than a small fraction of the available capacity, if the capacity
is enforced by a policer. [Citation Pending]. is enforced by a policer. [Citation Pending].
10.2.4. Duplex Self Interference 8.2.4. Duplex Self Interference
This engineering test confirms a bound on the interactions between This engineering test confirms a bound on the interactions between
the forward data path and the ACK return path. the forward data path and the ACK return path.
Some historical half duplex technologies had the property that each Some historical half duplex technologies had the property that each
direction held the channel until it completely drains its queue. direction held the channel until it completely drained its queue.
When a self clocked transport protocol, such as TCP, has data and When a self clocked transport protocol, such as TCP, has data and
acks passing in opposite directions through such a link, the behavior ACKs passing in opposite directions through such a link, the behavior
often reverts to stop-and-wait. Each additional packet added to the often reverts to stop-and-wait. Each additional packet added to the
window raises the observed RTT by two forward path packet times, once window raises the observed RTT by two forward path packet times, once
as it passes through the data path, and once for the additional delay as it passes through the data path, and once for the additional delay
incurred by the ACK waiting on the return path. incurred by the ACK waiting on the return path.
The duplex self interference test fails if the RTT rises by more than The duplex self interference test fails if the RTT rises by more than
some fixed bound above the expected queueing time computed from trom some fixed bound above the expected queueing time computed from trom
the excess window divided by the link data rate. This bound must be the excess window divided by the subpath IP Capacity. This bound
smaller than target_RTT/2 to avoid reverting to stop and wait must be smaller than target_RTT/2 to avoid reverting to stop and wait
behavior. (e.g. Packets have to be released at least twice per RTT, behavior. (e.g. Data packets and ACKs have to be released at least
to avoid stop and wait behavior.) twice per RTT.)
10.3. Slowstart tests 8.3. Slowstart tests
These tests mimic slowstart: data is sent at twice the effective These tests mimic slowstart: data is sent at twice the effective
bottleneck rate to exercise the queue at the dominant bottleneck. bottleneck rate to exercise the queue at the dominant bottleneck.
In general they are deemed inconclusive if the elapsed time to send In general they are deemed inconclusive if the elapsed time to send
the data burst is not less than half of the time to receive the ACKs. the data burst is not less than half of the time to receive the ACKs.
(i.e. sending data too fast is ok, but sending it slower than twice (i.e. sending data too fast is ok, but sending it slower than twice
the actual bottleneck rate as indicated by the ACKs is deemed the actual bottleneck rate as indicated by the ACKs is deemed
inconclusive). Space the bursts such that the average data rate is inconclusive). Space the bursts such that the average data rate is
equal to the target_data_rate. equal to the target_data_rate.
10.3.1. Full Window slowstart test 8.3.1. Full Window slowstart test
This is a capacity test to confirm that slowstart is not likely to This is a capacity test to confirm that slowstart is not likely to
exit prematurely. Send slowstart bursts that are target_pipe_size exit prematurely. Send slowstart bursts that are target_window_size
total packets. total packets.
Accumulate packet delivery statistics as described in Section 8.2 to Accumulate packet delivery statistics as described in Section 7.2 to
score the outcome. Pass if it is statistically significant that the score the outcome. Pass if it is statistically significant that the
observed number of good packets delivered between losses or ECN marks observed number of good packets delivered between losses or ECN marks
is larger than the target_run_length. Fail if it is statistically is larger than the target_run_length. Fail if it is statistically
significant that the observed interval between losses or ECN marks is significant that the observed interval between losses or ECN marks is
smaller than the target_run_length. smaller than the target_run_length.
Note that these are the same parameters as the Sender Full Window Note that these are the same parameters as the Sender Full Window
burst test, except the burst rate is at slowestart rate, rather than burst test, except the burst rate is at slowestart rate, rather than
sender interface rate. sender interface rate.
10.3.2. Slowstart AQM test 8.3.2. Slowstart AQM test
Do a continuous slowstart (send data continuously at slowstart_rate), Do a continuous slowstart (send data continuously at slowstart_rate),
until the first loss, stop, allow the network to drain and repeat, until the first loss, stop, allow the network to drain and repeat,
gathering statistics on the last packet delivered before the loss, gathering statistics on the last packet delivered before the loss,
the loss pattern, maximum observed RTT and window size. Justify the the loss pattern, maximum observed RTT and window size. Justify the
results. There is not currently sufficient theory justifying results. There is not currently sufficient theory justifying
requiring any particular result, however design decisions that affect requiring any particular result, however design decisions that affect
the outcome of this tests also affect how the network balances the outcome of this tests also affect how the network balances
between long and short flows (the "mice and elephants" problem). The between long and short flows (the "mice and elephants" problem). The
queue at the time of the first loss should be at least one half of queue at the time of the first loss should be at least one half of
the target_RTT. the target_RTT.
This is an engineering test: It would be best performed on a This is an engineering test: It would be best performed on a
quiescent network or testbed, since cross traffic has the potential quiescent network or testbed, since cross traffic has the potential
to change the results. to change the results.
10.4. Sender Rate Burst tests 8.4. Sender Rate Burst tests
These tests determine how well the network can deliver bursts sent at These tests determine how well the network can deliver bursts sent at
sender's interface rate. Note that this test most heavily exercises sender's interface rate. Note that this test most heavily exercises
the front path, and is likely to include infrastructure may be out of the front path, and is likely to include infrastructure may be out of
scope for an access ISP, even though the bursts might be caused by scope for an access ISP, even though the bursts might be caused by
ACK compression, thinning or channel arbitration in the access ISP. ACK compression, thinning or channel arbitration in the access ISP.
See Appendix B. See Appendix B.
Also, there are a several details that are not precisely defined. Also, there are a several details that are not precisely defined.
For starters there is not a standard server interface rate. 1 Gb/s For starters there is not a standard server interface rate. 1 Gb/s
and 10 Gb/s are very common today, but higher rates will become cost and 10 Gb/s are very common today, but higher rates will become cost
effective and can be expected to be dominant some time in the future. effective and can be expected to be dominant some time in the future.
Current standards permit TCP to send a full window bursts following Current standards permit TCP to send a full window bursts following
an application pause. (Congestion Window Validation [RFC2861], is an application pause. (Congestion Window Validation [RFC2861], is
not required, but even if was, it does not take effect until an not required, but even if was, it does not take effect until an
application pause is longer than an RTO.) Since full window bursts application pause is longer than an RTO.) Since full window bursts
are consistent with standard behavior, it is desirable that the are consistent with standard behavior, it is desirable that the
network be able to deliver such bursts, otherwise application pauses network be able to deliver such bursts, otherwise application pauses
will cause unwarranted losses. Note that the AIMD sawtooth requires will cause unwarranted losses. Note that the AIMD sawtooth requires
a peak window that is twice target_pipe_size, so the worst case burst a peak window that is twice target_window_size, so the worst case
may be 2*target_pipe_size. burst may be 2*target_window_size.
It is also understood in the application and serving community that It is also understood in the application and serving community that
interface rate bursts have a cost to the network that has to be interface rate bursts have a cost to the network that has to be
balanced against other costs in the servers themselves. For example balanced against other costs in the servers themselves. For example
TCP Segmentation Offload (TSO) reduces server CPU in exchange for TCP Segmentation Offload (TSO) reduces server CPU in exchange for
larger network bursts, which increase the stress on network buffer larger network bursts, which increase the stress on network buffer
memory. memory.
There is not yet theory to unify these costs or to provide a There is not yet theory to unify these costs or to provide a
framework for trying to optimize global efficiency. We do not yet framework for trying to optimize global efficiency. We do not yet
have a model for how much the network should tolerate server rate have a model for how much the network should tolerate server rate
bursts. Some bursts must be tolerated by the network, but it is bursts. Some bursts must be tolerated by the network, but it is
probably unreasonable to expect the network to be able to efficiently probably unreasonable to expect the network to be able to efficiently
deliver all data as a series of bursts. deliver all data as a series of bursts.
For this reason, this is the only test for which we encourage For this reason, this is the only test for which we encourage
derating. A TDS could include a table of pairs of derating derating. A TDS could include a table of pairs of derating
parameters: what burst size to use as a fraction of the parameters: what burst size to use as a fraction of the
target_pipe_size, and how much each burst size is permitted to reduce target_window_size, and how much each burst size is permitted to
the run length, relative to to the target_run_length. reduce the run length, relative to to the target_run_length.
10.5. Combined and Implicit Tests 8.5. Combined and Implicit Tests
Combined tests efficiently confirm multiple network properties in a Combined tests efficiently confirm multiple network properties in a
single test, possibly as a side effect of normal content delivery. single test, possibly as a side effect of normal content delivery.
They require less measurement traffic than other testing strategies They require less measurement traffic than other testing strategies
at the cost of conflating diagnostic signatures when they fail. at the cost of conflating diagnostic signatures when they fail.
These are by far the most efficient for monitoring networks that are These are by far the most efficient for monitoring networks that are
nominally expected to pass all tests. nominally expected to pass all tests.
10.5.1. Sustained Bursts Test 8.5.1. Sustained Bursts Test
The sustained burst test implements a combined worst case version of The sustained burst test implements a combined worst case version of
all of the capacity tests above. It is simply: all of the capacity tests above. It is simply:
Send target_pipe_size bursts of packets at server interface rate with Send target_window_size bursts of packets at server interface rate
target_RTT burst headway (burst start to burst start). Verify that with target_RTT burst headway (burst start to burst start). Verify
the observed delivery statistics meets the target_run_length. that the observed packet delivery statistics meets the
target_run_length.
Key observations: Key observations:
o The subpath under test is expected to go idle for some fraction of o The subpath under test is expected to go idle for some fraction of
the time: (subpath_data_rate-target_rate)/subpath_data_rate. the time: (subpath_IP_capacity-target_rate/
(target_MTU-header_overhead)*target_MTU)/subpath_IP_capacity.
Failing to do so indicates a problem with the procedure and an Failing to do so indicates a problem with the procedure and an
inconclusive test result. inconclusive test result.
o The burst sensitivity can be derated by sending smaller bursts o The burst sensitivity can be derated by sending smaller bursts
more frequently. E.g. send target_pipe_size*derate packet bursts more frequently. E.g. send target_window_size*derate packet
every target_RTT*derate. bursts every target_RTT*derate.
o When not derated, this test is the most strenuous capacity test. o When not derated, this test is the most strenuous capacity test.
o A link that passes this test is likely to be able to sustain o A subpath that passes this test is likely to be able to sustain
higher rates (close to subpath_data_rate) for paths with RTTs higher rates (close to subpath_IP_capacity) for paths with RTTs
significantly smaller than the target_RTT. significantly smaller than the target_RTT.
o This test can be implemented with instrumented TCP [RFC4898], o This test can be implemented with instrumented TCP [RFC4898],
using a specialized measurement application at one end [MBMSource] using a specialized measurement application at one end [MBMSource]
and a minimal service at the other end [RFC0863] [RFC0864]. and a minimal service at the other end [RFC0863] [RFC0864].
o This test is efficient to implement, since it does not require o This test is efficient to implement, since it does not require
per-packet timers, and can make use of TSO in modern NIC hardware. per-packet timers, and can make use of TSO in modern NIC hardware.
o This test by itself is not sufficient: the standing window o This test by itself is not sufficient: the standing window
engineering tests are also needed to ensure that the link is well engineering tests are also needed to ensure that the subpath is
behaved at and beyond the onset of congestion. well behaved at and beyond the onset of congestion.
o Assuming the link passes relevant standing window engineering o Assuming the subpath passes relevant standing window engineering
tests (particularly that it has a progressive onset of loss at an tests (particularly that it has a progressive onset of loss at an
appropriate queue depth) the passing sustained burst test is appropriate queue depth) the passing sustained burst test is
(believed to be) a sufficient verify that the subpath will not (believed to be) a sufficient verify that the subpath will not
impair stream at the target performance under all conditions. impair stream at the target performance under all conditions.
Proving this statement will be subject of ongoing research. Proving this statement will be subject of ongoing research.
Note that this test is clearly independent of the subpath RTT, or Note that this test is clearly independent of the subpath RTT, or
other details of the measurement infrastructure, as long as the other details of the measurement infrastructure, as long as the
measurement infrastructure can accurately and reliably deliver the measurement infrastructure can accurately and reliably deliver the
required bursts to the subpath under test. required bursts to the subpath under test.
10.5.2. Streaming Media 8.5.2. Streaming Media
Model Based Metrics can be implicitly implemented as a side effect of Model Based Metrics can be implicitly implemented as a side effect of
serving any non-throughput maximizing traffic, such as streaming serving any non-throughput maximizing traffic, such as streaming
media, with some additional controls and instrumentation in the media, with some additional controls and instrumentation in the
servers. The essential requirement is that the traffic be servers. The essential requirement is that the traffic be
constrained such that even with arbitrary application pauses, bursts constrained such that even with arbitrary application pauses, bursts
and data rate fluctuations, the traffic stays within the envelope and data rate fluctuations, the traffic stays within the envelope
defined by the individual tests described above. defined by the individual tests described above.
If the application's serving_data_rate is less than or equal to the If the application's serving_data_rate is less than or equal to the
target_data_rate and the serving_RTT (the RTT between the sender and target_data_rate and the serving_RTT (the RTT between the sender and
client) is less than the target_RTT, this constraint is most easily client) is less than the target_RTT, this constraint is most easily
implemented by clamping the transport window size to be no larger implemented by clamping the transport window size to be no larger
than: than:
serving_window_clamp=target_data_rate*serving_RTT/ serving_window_clamp=target_data_rate*serving_RTT/
(target_MTU-header_overhead) (target_MTU-header_overhead)
Under the above constraints the serving_window_clamp will limit the Under the above constraints the serving_window_clamp will limit the
both the serving data rate and burst sizes to be no larger than the both the serving data rate and burst sizes to be no larger than the
procedures in Section 10.1.2 and Section 10.4 or Section 10.5.1. procedures in Section 8.1.2 and Section 8.4 or Section 8.5.1. Since
Since the serving RTT is smaller than the target_RTT, the worst case the serving RTT is smaller than the target_RTT, the worst case bursts
bursts that might be generated under these conditions will be smaller that might be generated under these conditions will be smaller than
than called for by Section 10.4 and the sender rate burst sizes are called for by Section 8.4 and the sender rate burst sizes are
implicitly derated by the serving_window_clamp divided by the implicitly derated by the serving_window_clamp divided by the
target_pipe_size at the very least. (Depending on the application target_window_size at the very least. (Depending on the application
behavior, the data traffic might be significantly smoother than behavior, the data traffic might be significantly smoother than
specified by any of the burst tests.) specified by any of the burst tests.)
Note that it is important that the target_data_rate be above the
actual average rate needed by the application so it can recover after
transient pauses caused by congestion or the application itself.
In an alternative implementation the data rate and bursts might be In an alternative implementation the data rate and bursts might be
explicitly controlled by a host shaper or pacing at the sender. This explicitly controlled by a host shaper or pacing at the sender. This
would provide better control over transmissions but it is would provide better control over transmissions but it is
substantially more complicated to implement and would be likely to substantially more complicated to implement and would be likely to
have a higher CPU overhead. have a higher CPU overhead.
Note that these techniques can be applied to any content delivery Note that these techniques can be applied to any content delivery
that can be subjected to a reduced data rate in order to inhibit TCP that can be subjected to a reduced data rate in order to inhibit TCP
equilibrium behavior. equilibrium behavior.
11. An Example 9. An Example
In this section a we illustrate a TDS designed to confirm that an In this section a we illustrate a TDS designed to confirm that an
access ISP can reliably deliver HD video from multiple content access ISP can reliably deliver HD video from multiple content
providers to all of their customers. With modern codecs, minimal HD providers to all of their customers. With modern codecs, minimal HD
video (720p) generally fits in 2.5 Mb/s. Due to their geographical video (720p) generally fits in 2.5 Mb/s. Due to their geographical
size, network topology and modem designs the ISP determines that most size, network topology and modem designs the ISP determines that most
content is within a 50 mS RTT from their users (This is a sufficient content is within a 50 mS RTT from their users (This is a sufficient
to cover continental Europe or either US coast from a single serving to cover continental Europe or either US coast from a single serving
site.) site.)
2.5 Mb/s over a 50 ms path 2.5 Mb/s over a 50 ms path
+----------------------+-------+---------+ +----------------------+-------+---------+
| End-to-End Parameter | value | units | | End-to-End Parameter | value | units |
+----------------------+-------+---------+ +----------------------+-------+---------+
| target_rate | 2.5 | Mb/s | | target_rate | 2.5 | Mb/s |
| target_RTT | 50 | ms | | target_RTT | 50 | ms |
| target_MTU | 1500 | bytes | | target_MTU | 1500 | bytes |
| header_overhead | 64 | bytes | | header_overhead | 64 | bytes |
| target_pipe_size | 11 | packets | | target_window_size | 11 | packets |
| target_run_length | 363 | packets | | target_run_length | 363 | packets |
+----------------------+-------+---------+ +----------------------+-------+---------+
Table 1 Table 1
Table 1 shows the default TCP model with no derating, and as such is Table 1 shows the default TCP model with no derating, and as such is
quite conservative. The simplest TDS would be to use the sustained quite conservative. The simplest TDS would be to use the sustained
burst test, described in Section 10.5.1. Such a test would send 11 burst test, described in Section 8.5.1. Such a test would send 11
packet bursts every 50mS, and confirming that there was no more than packet bursts every 50mS, and confirming that there was no more than
1 packet loss per 33 bursts (363 total packets in 1.650 seconds). 1 packet loss per 33 bursts (363 total packets in 1.650 seconds).
Since this number represents is the entire end-to-end loss budget, Since this number represents is the entire end-to-end loss budget,
independent subpath tests could be implemented by apportioning the independent subpath tests could be implemented by apportioning the
loss ratio across subpaths. For example 50% of the losses might be packet loss ratio across subpaths. For example 50% of the losses
allocated to the access or last mile link to the user, 40% to the might be allocated to the access or last mile link to the user, 40%
interconnects with other ISPs and 1% to each internal hop (assuming to the interconnects with other ISPs and 1% to each internal hop
no more than 10 internal hops). Then all of the subpaths can be (assuming no more than 10 internal hops). Then all of the subpaths
tested independently, and the spatial composition of passing subpaths can be tested independently, and the spatial composition of passing
would be expected to be within the end-to-end loss budget. subpaths would be expected to be within the end-to-end loss budget.
Testing interconnects has generally been problematic: conventional Testing interconnects has generally been problematic: conventional
performance tests run between Measurement Points adjacent to either performance tests run between Measurement Points adjacent to either
side of the interconnect, are not generally useful. Unconstrained side of the interconnect, are not generally useful. Unconstrained
TCP tests, such as iperf [iperf] are usually overly aggressive TCP tests, such as iperf [iperf] are usually overly aggressive
because the RTT is so small (often less than 1 mS). With a short RTT because the RTT is so small (often less than 1 mS). With a short RTT
these tools are likely to report inflated numbers because for short these tools are likely to report inflated numbers because for short
RTTs these tools can tolerate very high loss ratio and can push other RTTs these tools can tolerate very high packet loss ratios and can
cross traffic off of the network. As a consequence they are useless push other cross traffic off of the network. As a consequence they
for predicting actual user performance, and may themselves be quite are useless for predicting actual user performance, and may
disruptive. Model Based Metrics solves this problem. The same test themselves be quite disruptive. Model Based Metrics solves this
pattern as used on other links can be applied to the interconnect. problem. The same test pattern as used on other subpaths can be
For our example, when apportioned 40% of the losses, 11 packet bursts applied to the interconnect. For our example, when apportioned 40%
sent every 50mS should have fewer than one loss per 82 bursts (902 of the losses, 11 packet bursts sent every 50mS should have fewer
packets). than one loss per 82 bursts (902 packets).
12. Validation 10. Validation
Since some aspects of the models are likely to be too conservative, Since some aspects of the models are likely to be too conservative,
Section 6.2 permits alternate protocol models and Section 6.3 permits Section 5.2 permits alternate protocol models and Section 5.3 permits
test parameter derating. If either of these techniques are used, we test parameter derating. If either of these techniques are used, we
require demonstrations that such a TDS can robustly detect links that require demonstrations that such a TDS can robustly detect subpaths
will prevent authentic applications using state-of-the-art protocol that will prevent authentic applications using state-of-the-art
implementations from meeting the specified performance targets. This protocol implementations from meeting the specified Target Transport
correctness criteria is potentially difficult to prove, because it Performance. This correctness criteria is potentially difficult to
implicitly requires validating a TDS against all possible links and prove, because it implicitly requires validating a TDS against all
subpaths. The procedures described here are still experimental. possible subpaths and subpaths. The procedures described here are
still experimental.
We suggest two approaches, both of which should be applied: first, We suggest two approaches, both of which should be applied: first,
publish a fully open description of the TDS, including what publish a fully open description of the TDS, including what
assumptions were used and and how it was derived, such that the assumptions were used and and how it was derived, such that the
research community can evaluate the design decisions, test them and research community can evaluate the design decisions, test them and
comment on their applicability; and second, demonstrate that an comment on their applicability; and second, demonstrate that an
applications running over an infinitessimally passing testbed do meet applications running over an infinitessimally passing testbed do meet
the performance targets. the performance targets.
An infinitessimally passing testbed resembles a epsilon-delta proof An infinitessimally passing testbed resembles a epsilon-delta proof
in calculus. Construct a test network such that all of the in calculus. Construct a test network such that all of the
individual tests of the TDS pass by only small (infinitesimal) individual tests of the TDS pass by only small (infinitesimal)
margins, and demonstrate that a variety of authentic applications margins, and demonstrate that a variety of authentic applications
running over real TCP implementations (or other protocol as running over real TCP implementations (or other protocol as
appropriate) meets the target transport performance over such a appropriate) meets the Target Transport Performance over such a
network. The workloads should include multiple types of streaming network. The workloads should include multiple types of streaming
media and transaction oriented short flows (e.g. synthetic web media and transaction oriented short flows (e.g. synthetic web
traffic ). traffic).
For example, for the HD streaming video TDS described in Section 11, For example, for the HD streaming video TDS described in Section 9,
the link layer bottleneck data rate should be exactly the header the IP capacity should be exactly the header overhead above 2.5 Mb/s,
overhead above 2.5 Mb/s, the per packet random background loss ratio the per packet random background loss ratio should be 1/363, for a
should be 1/363, for a run length of 363 packets, the bottleneck run length of 363 packets, the bottleneck queue should be 11 packets
queue should be 11 packets and the front path should have just enough and the front path should have just enough buffering to withstand 11
buffering to withstand 11 packet interface rate bursts. We want packet interface rate bursts. We want every one of the TDS tests to
every one of the TDS tests to fail if we slightly increase the fail if we slightly increase the relevant test parameter, so for
relevant test parameter, so for example sending a 12 packet bursts example sending a 12 packet bursts should cause excess (possibly
should cause excess (possibly deterministic) packet drops at the deterministic) packet drops at the dominant queue at the bottleneck.
dominant queue at the bottleneck. On this infinitessimally passing On this infinitessimally passing network it should be possible for a
network it should be possible for a real application using a stock real application using a stock TCP implementation in the vendor's
TCP implementation in the vendor's default configuration to attain default configuration to attain 2.5 Mb/s over an 50 mS path.
2.5 Mb/s over an 50 mS path.
The most difficult part of setting up such a testbed is arranging for The most difficult part of setting up such a testbed is arranging for
it to infinitesimally pass the individual tests. Two approaches: it to infinitesimally pass the individual tests. Two approaches:
constraining the network devices not to use all available resources constraining the network devices not to use all available resources
(e.g. by limiting available buffer space or data rate); and (e.g. by limiting available buffer space or data rate); and
preloading subpaths with cross traffic. Note that is it important preloading subpaths with cross traffic. Note that is it important
that a single environment be constructed which infinitessimally that a single environment be constructed which infinitessimally
passes all tests at the same time, otherwise there is a chance that passes all tests at the same time, otherwise there is a chance that
TCP can exploit extra latitude in some parameters (such as data rate) TCP can exploit extra latitude in some parameters (such as data rate)
to partially compensate for constraints in other parameters (queue to partially compensate for constraints in other parameters (queue
skipping to change at page 40, line 26 skipping to change at page 41, line 20
To the extent that a TDS is used to inform public dialog it should be To the extent that a TDS is used to inform public dialog it should be
fully publicly documented, including the details of the tests, what fully publicly documented, including the details of the tests, what
assumptions were used and how it was derived. All of the details of assumptions were used and how it was derived. All of the details of
the validation experiment should also be published with sufficient the validation experiment should also be published with sufficient
detail for the experiments to be replicated by other researchers. detail for the experiments to be replicated by other researchers.
All components should either be open source of fully described All components should either be open source of fully described
proprietary implementations that are available to the research proprietary implementations that are available to the research
community. community.
13. Security Considerations 11. Security Considerations
Measurement is often used to inform business and policy decisions, Measurement is often used to inform business and policy decisions,
and as a consequence is potentially subject to manipulation for and as a consequence is potentially subject to manipulation. Model
illicit gains. Model Based Metrics are expected to be a huge step Based Metrics are expected to be a huge step forward because
forward because equivalent measurements can be performed from equivalent measurements can be performed from multiple vantage
multiple vantage points, such that performance claims can be points, such that performance claims can be independently validated
independently validated by multiple parties. by multiple parties.
Much of the acrimony in the Net Neutrality debate is due by the Much of the acrimony in the Net Neutrality debate is due by the
historical lack of any effective vantage independent tools to historical lack of any effective vantage independent tools to
characterize network performance. Traditional methods for measuring characterize network performance. Traditional methods for measuring
Bulk Transport Capacity are sensitive to RTT and as a consequence Bulk Transport Capacity are sensitive to RTT and as a consequence
often yield very different results local to an ISP and when run over often yield very different results when run local to an ISP or
a customer's complete path. Neither the ISP nor customer can repeat internconnect and when run over a customer's complete path. Neither
the other's measurements, leading to high levels of distrust and the ISP nor customer can repeat the other's measurements, leading to
acrimony. Model Based Metrics are expected to greatly improve this high levels of distrust and acrimony. Model Based Metrics are
situation. expected to greatly improve this situation.
This document only describes a framework for designing Fully This document only describes a framework for designing Fully
Specified Targeted Diagnostic Suite. Each FSTDS MUST include its own Specified Targeted Diagnostic Suite. Each FSTDS MUST include its own
security section. security section.
14. Acknowledgements 12. Acknowledgements
Ganga Maguluri suggested the statistical test for measuring loss Ganga Maguluri suggested the statistical test for measuring loss
probability in the target run length. Alex Gilgur for helping with probability in the target run length. Alex Gilgur for helping with
the statistics. the statistics.
Meredith Whittaker for improving the clarity of the communications. Meredith Whittaker for improving the clarity of the communications.
Ruediger Geib provided feedback which greatly improved the document.
This work was inspired by Measurement Lab: open tools running on an This work was inspired by Measurement Lab: open tools running on an
open platform, using open tools to collect open data. See open platform, using open tools to collect open data. See
http://www.measurementlab.net/ http://www.measurementlab.net/
15. IANA Considerations 13. IANA Considerations
This document has no actions for IANA. This document has no actions for IANA.
16. References 14. References
16.1. Normative References 14.1. 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.
16.2. Informative References 14.2. Informative References
[RFC0863] Postel, J., "Discard Protocol", STD 21, RFC 863, May 1983. [RFC0863] Postel, J., "Discard Protocol", STD 21, RFC 863, May 1983.
[RFC0864] Postel, J., "Character Generator Protocol", STD 22, [RFC0864] Postel, J., "Character Generator Protocol", STD 22,
RFC 864, May 1983. RFC 864, May 1983.
[RFC2309] Braden, B., Clark, D., Crowcroft, J., Davie, B., Deering, [RFC2309] Braden, B., Clark, D., Crowcroft, J., Davie, B., Deering,
S., Estrin, D., Floyd, S., Jacobson, V., Minshall, G., S., Estrin, D., Floyd, S., Jacobson, V., Minshall, G.,
Partridge, C., Peterson, L., Ramakrishnan, K., Shenker, Partridge, C., Peterson, L., Ramakrishnan, K., Shenker,
S., Wroclawski, J., and L. Zhang, "Recommendations on S., Wroclawski, J., and L. Zhang, "Recommendations on
skipping to change at page 42, line 19 skipping to change at page 43, line 13
[RFC4015] Ludwig, R. and A. Gurtov, "The Eifel Response Algorithm [RFC4015] Ludwig, R. and A. Gurtov, "The Eifel Response Algorithm
for TCP", RFC 4015, February 2005. for TCP", RFC 4015, February 2005.
[RFC4737] Morton, A., Ciavattone, L., Ramachandran, G., Shalunov, [RFC4737] Morton, A., Ciavattone, L., Ramachandran, G., Shalunov,
S., and J. Perser, "Packet Reordering Metrics", RFC 4737, S., and J. Perser, "Packet Reordering Metrics", RFC 4737,
November 2006. November 2006.
[RFC4898] Mathis, M., Heffner, J., and R. Raghunarayan, "TCP [RFC4898] Mathis, M., Heffner, J., and R. Raghunarayan, "TCP
Extended Statistics MIB", RFC 4898, May 2007. Extended Statistics MIB", RFC 4898, May 2007.
[RFC5136] Chimento, P. and J. Ishac, "Defining Network Capacity",
RFC 5136, February 2008.
[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
Composition", RFC 5835, April 2010.
[RFC6049] Morton, A. and E. Stephan, "Spatial Composition of [RFC6049] Morton, A. and E. Stephan, "Spatial Composition of
Metrics", RFC 6049, January 2011. Metrics", RFC 6049, January 2011.
[RFC6673] Morton, A., "Round-Trip Packet Loss Metrics", RFC 6673, [RFC6673] Morton, A., "Round-Trip Packet Loss Metrics", RFC 6673,
August 2012. August 2012.
[RFC7312] Fabini, J. and A. Morton, "Advanced Stream and Sampling [RFC7312] Fabini, J. and A. Morton, "Advanced Stream and Sampling
Framework for IP Performance Metrics (IPPM)", RFC 7312, Framework for IP Performance Metrics (IPPM)", RFC 7312,
August 2014. August 2014.
[RFC7398] Bagnulo, M., Burbridge, T., Crawford, S., Eardley, P., and [RFC7398] Bagnulo, M., Burbridge, T., Crawford, S., Eardley, P., and
A. Morton, "A Reference Path and Measurement Points for A. Morton, "A Reference Path and Measurement Points for
Large-Scale Measurement of Broadband Performance", Large-Scale Measurement of Broadband Performance",
RFC 7398, February 2015. RFC 7398, February 2015.
[I-D.ietf-ippm-2680-bis]
Almes, G., Kalidindi, S., Zekauskas, M., and A. Morton, "A
One-Way Loss Metric for IPPM", draft-ietf-ippm-2680-bis-02
(work in progress), June 2015.
[I-D.ietf-aqm-recommendation] [I-D.ietf-aqm-recommendation]
Baker, F. and G. Fairhurst, "IETF Recommendations Baker, F. and G. Fairhurst, "IETF Recommendations
Regarding Active Queue Management", Regarding Active Queue Management",
draft-ietf-aqm-recommendation-11 (work in progress), draft-ietf-aqm-recommendation-11 (work in progress),
February 2015. February 2015.
[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.
skipping to change at page 44, line 7 skipping to change at page 44, line 51
index.php?title=Bufferbloat&oldid=608805474, March 2015. index.php?title=Bufferbloat&oldid=608805474, March 2015.
[CCscaling] [CCscaling]
Fernando, F., Doyle, J., and S. Steven, "Scalable laws for Fernando, F., Doyle, J., and S. Steven, "Scalable laws for
stable network congestion control", Proceedings of stable network congestion control", Proceedings of
Conference on Decision and Conference on Decision and
Control, http://www.ee.ucla.edu/~paganini, December 2001. Control, http://www.ee.ucla.edu/~paganini, December 2001.
Appendix A. Model Derivations Appendix A. Model Derivations
The reference target_run_length described in Section 6.2 is based on The reference target_run_length described in Section 5.2 is based on
very conservative assumptions: that all window above target_pipe_size very conservative assumptions: that all window above
contributes to a standing queue that raises the RTT, and that classic target_window_size contributes to a standing queue that raises the
Reno congestion control with delayed ACKs are in effect. In this RTT, and that classic Reno congestion control with delayed ACKs are
section we provide two alternative calculations using different in effect. In this section we provide two alternative calculations
assumptions. using different assumptions.
It may seem out of place to allow such latitude in a measurement It may seem out of place to allow such latitude in a measurement
standard, but this section provides offsetting requirements. standard, but this section provides offsetting requirements.
The estimates provided by these models make the most sense if network The estimates provided by these models make the most sense if network
performance is viewed logarithmically. In the operational Internet, performance is viewed logarithmically. In the operational Internet,
data rates span more than 8 orders of magnitude, RTT spans more than data rates span more than 8 orders of magnitude, RTT spans more than
3 orders of magnitude, and loss ratio spans at least 8 orders of 3 orders of magnitude, and packet loss ratio spans at least 8 orders
magnitude. When viewed logarithmically (as in decibels), these of magnitude if not more. When viewed logarithmically (as in
correspond to 80 dB of dynamic range. On an 80 db scale, a 3 dB decibels), these correspond to 80 dB of dynamic range. On an 80 dB
error is less than 4% of the scale, even though it might represent a scale, a 3 dB error is less than 4% of the scale, even though it
factor of 2 in untransformed parameter. represents a factor of 2 in untransformed parameter.
This document gives a lot of latitude for calculating This document gives a lot of latitude for calculating
target_run_length, however people designing a TDS should consider the target_run_length, however people designing a TDS should consider the
effect of their choices on the ongoing tussle about the relevance of effect of their choices on the ongoing tussle about the relevance of
"TCP friendliness" as an appropriate model for Internet capacity "TCP friendliness" as an appropriate model for Internet capacity
allocation. Choosing a target_run_length that is substantially allocation. Choosing a target_run_length that is substantially
smaller than the reference target_run_length specified in Section 6.2 smaller than the reference target_run_length specified in Section 5.2
strengthens the argument that it may be appropriate to abandon "TCP strengthens the argument that it may be appropriate to abandon "TCP
friendliness" as the Internet fairness model. This gives developers friendliness" as the Internet fairness model. This gives developers
incentive and permission to develop even more aggressive applications incentive and permission to develop even more aggressive applications
and protocols, for example by increasing the number of connections and protocols, for example by increasing the number of connections
that they open concurrently. that they open concurrently.
A.1. Queueless Reno A.1. Queueless Reno
In Section 6.2 it was assumed that the link rate matches the target In Section 5.2 it was assumed that the subpath IP rate matches the
rate plus overhead, such that the excess window needed for the AIMD target rate plus overhead, such that the excess window needed for the
sawtooth causes a fluctuating queue at the bottleneck. AIMD sawtooth causes a fluctuating queue at the bottleneck.
An alternate situation would be bottleneck where there is no An alternate situation would be bottleneck where there is no
significant queue and losses are caused by some mechanism that does significant queue and losses are caused by some mechanism that does
not involve extra delay, for example by the use of a virtual queue as not involve extra delay, for example by the use of a virtual queue as
in Approximate Fair Dropping[AFD]. A flow controlled by such a in Approximate Fair Dropping [AFD]. A flow controlled by such a
bottleneck would have a constant RTT and a data rate that fluctuates bottleneck would have a constant RTT and a data rate that fluctuates
in a sawtooth due to AIMD congestion control. Assume the losses are in a sawtooth due to AIMD congestion control. Assume the losses are
being controlled to make the average data rate meet some goal which being controlled to make the average data rate meet some goal which
is equal or greater than the target_rate. The necessary run length is equal or greater than the target_rate. The necessary run length
can be computed as follows: can be computed as follows:
For some value of Wmin, the window will sweep from Wmin packets to For some value of Wmin, the window will sweep from Wmin packets to
2*Wmin packets in 2*Wmin RTT (due to delayed ACK). Unlike the 2*Wmin packets in 2*Wmin RTT (due to delayed ACK). Unlike the
queueing case where Wmin = Target_pipe_size, we want the average of queueing case where Wmin = target_window_size, we want the average of
Wmin and 2*Wmin to be the target_pipe_size, so the average rate is Wmin and 2*Wmin to be the target_window_size, so the average rate is
the target rate. Thus we want Wmin = (2/3)*target_pipe_size. the target rate. Thus we want Wmin = (2/3)*target_window_size.
Between losses each sawtooth delivers (1/2)(Wmin+2*Wmin)(2Wmin) Between losses each sawtooth delivers (1/2)(Wmin+2*Wmin)(2Wmin)
packets in 2*Wmin round trip times. packets in 2*Wmin round trip times.
Substituting these together we get: Substituting these together we get:
target_run_length = (4/3)(target_pipe_size^2) target_run_length = (4/3)(target_window_size^2)
Note that this is 44% of the reference_run_length computed earlier. Note that this is 44% of the reference_run_length computed earlier.
This makes sense because under the assumptions in Section 6.2 the This makes sense because under the assumptions in Section 5.2 the
AMID sawtooth caused a queue at the bottleneck, which raised the AMID sawtooth caused a queue at the bottleneck, which raised the
effective RTT by 50%. effective RTT by 50%.
Appendix B. Complex Queueing Appendix B. Complex Queueing
For many network technologies simple queueing models don't apply: the For many network technologies simple queueing models don't apply: the
network schedules, thins or otherwise alters the timing of ACKs and network schedules, thins or otherwise alters the timing of ACKs and
data, generally to raise the efficiency of the channel allocation data, generally to raise the efficiency of the channel allocation
when confronted with relatively widely spaced small ACKs. These when confronted with relatively widely spaced small ACKs. These
efficiency strategies are ubiquitous for half duplex, wireless and efficiency strategies are ubiquitous for half duplex, wireless and
broadcast media. broadcast media.
Altering the ACK stream generally has two consequences: it raises the Altering the ACK stream generally has two consequences: it raises the
effective bottleneck data rate, making slowstart burst at higher implied bottleneck IP capacity, making slowstart burst at higher
rates (possibly as high as the sender's interface rate) and it rates (possibly as high as the sender's interface rate) and it
effectively raises the RTT by the average time that the ACKs and data effectively raises the RTT by the average time that the ACKs and data
were delayed. The first effect can be partially mitigated by were delayed. The first effect can be partially mitigated by
reclocking ACKs once they are beyond the bottleneck on the return reclocking ACKs once they are beyond the bottleneck on the return
path to the sender, however this further raises the effective RTT. path to the sender, however this further raises the effective RTT.
The most extreme example of this sort of behavior would be a half The most extreme example of this sort of behavior would be a half
duplex channel that is not released as long as end point currently duplex channel that is not released as long as end point currently
holding the channel has more traffic (data or ACKs) to send. Such holding the channel has more traffic (data or ACKs) to send. Such
environments cause self clocked protocols under full load to revert environments cause self clocked protocols under full load to revert
to extremely inefficient stop and wait behavior, where they send an to extremely inefficient stop and wait behavior, where they send an
entire window of data as a single burst of the forward path, followed entire window of data as a single burst of the forward path, followed
by the entire window of ACKs on the return path. It is important to by the entire window of ACKs on the return path. It is important to
note that due to self clocking, ill conceived channel allocation note that due to self clocking, ill conceived channel allocation
mechanisms can increase the stress on upstream links in a long path: mechanisms can increase the stress on upstream subpaths in a long
they cause large and faster bursts. path: they cause large and faster bursts.
If a particular return path contains a link or device that alters the If a particular return path contains a subpath or device that alters
ACK stream, then the entire path from the sender up to the bottleneck the ACK stream, then the entire path from the sender up to the
must be tested at the burst parameters implied by the ACK scheduling bottleneck must be tested at the burst parameters implied by the ACK
algorithm. The most important parameter is the Effective Bottleneck scheduling algorithm. The most important parameter is the Implied
Data Rate, which is the average rate at which the ACKs advance Bottleneck IP Capacity, which is the average rate at which the ACKs
snd.una. Note that thinning the ACKs (relying on the cumulative advance snd.una. Note that thinning the ACKs (relying on the
nature of seg.ack to permit discarding some ACKs) is implies an cumulative nature of seg.ack to permit discarding some ACKs) is
effectively infinite bottleneck data rate. implies an effectively infinite Implied Bottleneck IP Capacity.
Holding data or ACKs for channel allocation or other reasons (such as Holding data or ACKs for channel allocation or other reasons (such as
forward error correction) always raises the effective RTT relative to forward error correction) always raises the effective RTT relative to
the minimum delay for the path. Therefore it may be necessary to the minimum delay for the path. Therefore it may be necessary to
replace target_RTT in the calculation in Section 6.2 by an replace target_RTT in the calculation in Section 5.2 by an
effective_RTT, which includes the target_RTT plus a term to account effective_RTT, which includes the target_RTT plus a term to account
for the extra delays introduced by these mechanisms. for the extra delays introduced by these mechanisms.
Appendix C. Version Control Appendix C. Version Control
This section to be removed prior to publication. This section to be removed prior to publication.
Formatted: Sat Jun 13 16:25:01 PDT 2015 Formatted: Mon Jul 6 13:49:30 PDT 2015
Authors' Addresses Authors' Addresses
Matt Mathis Matt Mathis
Google, Inc Google, Inc
1600 Amphitheater Parkway 1600 Amphitheater Parkway
Mountain View, California 94043 Mountain View, California 94043
USA USA
Email: mattmathis@google.com Email: mattmathis@google.com
 End of changes. 225 change blocks. 
770 lines changed or deleted 818 lines changed or added

This html diff was produced by rfcdiff 1.42. The latest version is available from http://tools.ietf.org/tools/rfcdiff/