draft-ietf-tcpm-hystartplusplus-02.txt   draft-ietf-tcpm-hystartplusplus-03.txt 
Network Working Group P. Balasubramanian Network Working Group P. Balasubramanian
Internet-Draft Y. Huang Internet-Draft Y. Huang
Intended status: Standards Track M. Olson Intended status: Standards Track M. Olson
Expires: January 13, 2022 Microsoft Expires: 26 January 2022 Microsoft
July 12, 2021 25 July 2021
HyStart++: Modified Slow Start for TCP HyStart++: Modified Slow Start for TCP
draft-ietf-tcpm-hystartplusplus-02 draft-ietf-tcpm-hystartplusplus-03
Abstract Abstract
This doument describes HyStart++, a simple modification to the slow This doument describes HyStart++, a simple modification to the slow
start phase of TCP congestion control algorithms. Traditional slow start phase of TCP congestion control algorithms. Traditional slow
start can cause overshotting of the ideal send rate and cause large start can cause overshooting of the ideal send rate and cause large
packet loss within a round-trip time which results in poor packet loss within a round-trip time which results in poor
performance. HyStart++ is composed of the delay increase variant of performance. HyStart++ uses a delay increase heuristic to exit slow
HyStart to prevent overshooting of the ideal sending rate, while also start early while also mitigating poor performance which can result
mitigating poor performance which can result from false positives. from false positives.
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 https://datatracker.ietf.org/drafts/current/. Drafts is at https://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 January 13, 2022. This Internet-Draft will expire on 26 January 2022.
Copyright Notice Copyright Notice
Copyright (c) 2021 IETF Trust and the persons identified as the Copyright (c) 2021 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 (https://trustee.ietf.org/
(https://trustee.ietf.org/license-info) in effect on the date of license-info) in effect on the date of publication of this document.
publication of this document. Please review these documents Please review these documents carefully, as they describe your rights
carefully, as they describe your rights and restrictions with respect and restrictions with respect to this document. Code Components
to this document. Code Components extracted from this document must extracted from this document must include Simplified BSD License text
include Simplified BSD License text as described in Section 4.e of as described in Section 4.e of the Trust Legal Provisions and are
the Trust Legal Provisions and are provided without warranty as provided without warranty as described in the Simplified BSD License.
described in the Simplified BSD License.
Table of Contents Table of Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2
2. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 3 2. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 3
3. Definitions . . . . . . . . . . . . . . . . . . . . . . . . . 3 3. Definitions . . . . . . . . . . . . . . . . . . . . . . . . . 3
4. HyStart++ Algorithm . . . . . . . . . . . . . . . . . . . . . 3 4. HyStart++ Algorithm . . . . . . . . . . . . . . . . . . . . . 3
4.1. Summary . . . . . . . . . . . . . . . . . . . . . . . . . 3 4.1. Summary . . . . . . . . . . . . . . . . . . . . . . . . . 3
4.2. Algorithm Details . . . . . . . . . . . . . . . . . . . . 4 4.2. Algorithm Details . . . . . . . . . . . . . . . . . . . . 4
4.3. Tuning constants . . . . . . . . . . . . . . . . . . . . 6 4.3. Tuning constants . . . . . . . . . . . . . . . . . . . . 6
skipping to change at page 2, line 30 skipping to change at page 2, line 36
8. References . . . . . . . . . . . . . . . . . . . . . . . . . 7 8. References . . . . . . . . . . . . . . . . . . . . . . . . . 7
8.1. Normative References . . . . . . . . . . . . . . . . . . 7 8.1. Normative References . . . . . . . . . . . . . . . . . . 7
8.2. Informative References . . . . . . . . . . . . . . . . . 8 8.2. Informative References . . . . . . . . . . . . . . . . . 8
Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 8 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 8
1. Introduction 1. Introduction
[RFC5681] describes the slow start congestion control algorithm for [RFC5681] describes the slow start congestion control algorithm for
TCP. The slow start algorithm is used when the congestion window TCP. The slow start algorithm is used when the congestion window
(cwnd) is less than the slow start threshold (ssthresh). During slow (cwnd) is less than the slow start threshold (ssthresh). During slow
start, in absence of packet loss signals, TCP sender increases cwnd start, in absence of packet loss signals, TCP increases cwnd
exponentially to probe the network capacity. Such a fast growth can exponentially to probe the network capacity. This fast growth can
lead to overshooting the ideal sending rate and cause significant overshoot the ideal sending rate and cause significant packet loss
packet loss. This is counter-productive for the TCP flow itself, and which cannot always be recovered efficiently, impairing flow
also impacts the rest of the traffic sharing the bottleneck link. completion time.
TCP has several mechanisms for loss recovery, but they are only
effective for moderate loss. When these techniques are unable to
recover lost packets, a last-resort retransmission timeout (RTO) is
used to trigger packet recovery. In most operating systems, the
minimum RTO is set to a large value (200 msec or 300 msec) to prevent
spurious timeouts. This results in a long idle time which
drastically impairs flow completion times.
HyStart++ adds delay increase as a signal to exit slow start before HyStart++ first uses delay increase as a signal to exit slow start
any packet loss occurs. This is one of two algorithms specified in before any packet loss occurs. This is one of two algorithms
[HyStart]. After the HyStart delay algorithm finds an exit point, a specified in [HyStart]. After the HyStart delay algorithm finds an
Conservative Slow Start (CSS) phase is used to determine if the slow exit point, a novel Conservative Slow Start (CSS) phase is used to
start exit was spurious. This provides protection against jitter and determine whether the slow start exit was spurious. This provides
prevents pefrormance problems that result from early slow start exit protection against jitter and prevents performance problems that
due to false positives. HyStart++ reduces packet loss and result from early slow start exit due to false positives. HyStart++
retransmissions, and improves goodput in lab measurements as well as reduces packet loss and retransmissions, and improves goodput in lab
real world deployments. measurements as well as real world deployments.
2. Terminology 2. 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].
3. Definitions 3. Definitions
We repeat here some definition from [RFC5681] to aid the reader. We repeat here some definition from [RFC5681] to aid the reader.
skipping to change at page 3, line 49 skipping to change at page 3, line 49
[HyStart] specifies two algorithms (a "Delay Increase" algorithm and [HyStart] specifies two algorithms (a "Delay Increase" algorithm and
an "Inter-Packet Arrival" algorithm) to be run in parallel to detect an "Inter-Packet Arrival" algorithm) to be run in parallel to detect
that the sending rate has reached capacity. In practice, the Inter- that the sending rate has reached capacity. In practice, the Inter-
Packet Arrival algorithm does not perform well and is not able to Packet Arrival algorithm does not perform well and is not able to
detect congestion early, primarily due to ACK compression. The idea detect congestion early, primarily due to ACK compression. The idea
of the Delay Increase algorithm is to look for RTT spikes, which of the Delay Increase algorithm is to look for RTT spikes, which
suggest that the bottleneck buffer is filling up. suggest that the bottleneck buffer is filling up.
In HyStart++, a TCP sender uses traditional slow start and then uses In HyStart++, a TCP sender uses traditional slow start and then uses
the "Delay Increase" algorithm to trigger an exit from slow start. the "Delay Increase" algorithm to trigger an exit from slow start.
But instead of using a congestion avoidance algorithm, the sender But instead of going straight from slow start to congestion
uses a Conservative Slow Start (CSS) algorithm to determine if the avoidance, the sender spends a number of RTTs in a Conservative Slow
exit was spurious. If the exit is determined to be spurious, slow Start (CSS) phase to determine whether the exit was spurious. During
start is resumed. If the exit is determined to be not spurious, the CSS, the congestion window is grown exponentially like in regular
sender enters congestion avoidance. slow start, but with a smaller exponential base, resulting in less
aggressive growth. If the RTT shrinks at any time during CSS, it's
concluded that the RTT spike was not related to congestion caused by
the connection sending too fast (i.e. the exit was spurious), and the
connection resumes slow start. If the RTT inflation persists
throughout CSS, the connection enters congestion avoidance.
4.2. Algorithm Details 4.2. Algorithm Details
We assume that Appropriate Byte Counting (as described in [RFC3465]) We assume that Appropriate Byte Counting (as described in [RFC3465])
is in use and L is the cwnd increase limit. The choice of value of L is in use and L is the cwnd increase limit as discussed in RFC 3465.
is up to the implementation.
A round is chosen to be approximately the Round-Trip Time (RTT). A round is chosen to be approximately the Round-Trip Time (RTT). We
Round can be approximated using sequence numbers as follows: recommend that rounds be measured using sequence numbers. Round can
be approximated using sequence numbers as follows:
Define windowEnd as a sequence number initialize to SND.UNA Define windowEnd as a sequence number initialize to SND.UNA
When windowEnd is ACKed, the current round ends and windowEnd is When windowEnd is ACKed, the current round ends and windowEnd is
set to SND.NXT set to SND.NXT
At the start of each round during normal slow start and CSS: At the start of each round during standard slow start ([RFC5681]) and
CSS:
lastRoundMinRTT = currentRoundMinRTT lastRoundMinRTT = currentRoundMinRTT
currentRoundMinRTT = infinity currentRoundMinRTT = infinity
rttSampleCount = 0 rttSampleCount = 0
For each arriving ACK in slow start, where N is the number of For each arriving ACK in slow start, where N is the number of
previously unacknowledged bytes acknowledged in the arriving ACK: previously unacknowledged bytes acknowledged in the arriving ACK:
Update the cwnd Update the cwnd
cwnd = cwnd + min (N, L * SMSS) - cwnd = cwnd + min (N, L * SMSS)
Keep track of minimum observed RTT Keep track of minimum observed RTT
currentRoundMinRTT = min(currentRoundMinRTT, currRTT) - currentRoundMinRTT = min(currentRoundMinRTT, currRTT)
where currRTT is the RTT sampled from the incoming ACK - where currRTT is the RTT sampled from the latest incoming ACK
rttSampleCount += 1 - rttSampleCount += 1
For rounds where cwnd is at or higher than LOW_CWND and For rounds where cwnd is at or higher than LOW_CWND and
N_RTT_SAMPLE RTT samples have been obtained, check if delay N_RTT_SAMPLE RTT samples have been obtained, check if delay
increase triggers slow start exit increase triggers slow start exit
- if (cwnd >= (LOW_CWND * SMSS) AND rttSampleCount >=
if (cwnd >= (LOW_CWND * SMSS) AND rttSampleCount >=
N_RTT_SAMPLE) N_RTT_SAMPLE)
RttThresh = clamp(MIN_RTT_THRESH, lastRoundMinRTT / 8,
o RttThresh = clamp(MIN_RTT_THRESH, lastRoundMinRTT / 8,
MAX_RTT_THRESH) MAX_RTT_THRESH)
if (currentRoundMinRTT >= (lastRoundMinRTT + RttThresh)) o if (currentRoundMinRTT >= (lastRoundMinRTT + RttThresh))
cssBaselineMinRtt = currentRoundMinRTT + cssBaselineMinRtt = currentRoundMinRTT
exit slow start and enter CSS + exit slow start and enter CSS
CSS lasts CSS_ROUNDS rounds. If the transition into CSS happens in CSS lasts at most CSS_ROUNDS rounds. If the transition into CSS
the middle of a round, that partial round counts towards the limit. happens in the middle of a round, that partial round counts towards
the limit.
For each arriving ACK in CSS, where N is the number of previously For each arriving ACK in CSS, where N is the number of previously
unacknowledged bytes acknowledged in the arriving ACK: unacknowledged bytes acknowledged in the arriving ACK:
Update the cwnd Update the cwnd
cwnd = cwnd + (min (N, L * SMSS) / CSS_GROWTH_DIVISOR) - cwnd = cwnd + (min (N, L * SMSS) / CSS_GROWTH_DIVISOR)
Keep track of minimum observed RTT Keep track of minimum observed RTT
currentRoundMinRTT = min(currentRoundMinRTT, currRTT) - currentRoundMinRTT = min(currentRoundMinRTT, currRTT)
where currRTT is the sampled RTT from the incoming ACK - where currRTT is the sampled RTT from the incoming ACK
rttSampleCount += 1 - rttSampleCount += 1
For CSS rounds where N_RTT_SAMPLE RTT samples have been obtained, For CSS rounds where N_RTT_SAMPLE RTT samples have been obtained,
check if current round's minRTT drops below baseline indicating check if current round's minRTT drops below baseline indicating
that HyStart exit was spurious. that HyStart exit was spurious.
if (currentRoundMinRTT < cssBaselineMinRtt) - if (currentRoundMinRTT < cssBaselineMinRtt)
cssBaselineMinRtt = infinity o cssBaselineMinRtt = infinity
resume slow start including HyStart++ o resume slow start including HyStart++
If CSS_ROUNDS rounds are complete, enter congestion avoidance. If CSS_ROUNDS rounds are complete, enter congestion avoidance.
ssthresh = cwnd * ssthresh = cwnd
If congestion is observed anytime during slow start or CSS, enter If loss or ECN-marking is observed anytime during standard slow start
congestion avoidance. or CSS, enter congestion avoidance.
ssthresh = cwnd * ssthresh = cwnd
4.3. Tuning constants 4.3. Tuning constants
It is RECOMMENDED that a HyStart++ implementation use the following It is RECOMMENDED that a HyStart++ implementation use the following
constants: constants:
LOW_CWND = 16 * LOW_CWND = 16
MIN_RTT_THRESH = 4 msec * MIN_RTT_THRESH = 4 msec
MAX_RTT_THRESH = 16 msec * MAX_RTT_THRESH = 16 msec
N_RTT_SAMPLE = 8 * N_RTT_SAMPLE = 8
CSS_GROWTH_DIVISOR = 4 * CSS_GROWTH_DIVISOR = 4
CSS_ROUNDS = 5 * CSS_ROUNDS = 5
These constants have been determined with lab measurements and real These constants have been determined with lab measurements and real
world deployments. An implementation MAY tune them for different world deployments. An implementation MAY tune them for different
network characteristics. network characteristics.
Using smaller values of LOW_CWND will cause the algorithm to kick in Using smaller values of LOW_CWND will cause the algorithm to kick in
before the last round RTT can be measured, particularly if the before the last round RTT can be measured, particularly if the
implementation uses an initial cwnd of 10 MSS. Higher values will implementation uses an initial cwnd of 10 MSS. Higher values will
delay the detection of delay increase and reduce the ability of delay the detection of delay increase and reduce the ability of
HyStart++ to prevent overshoot problems. HyStart++ to prevent overshoot problems.
skipping to change at page 6, line 43 skipping to change at page 6, line 45
MAX_RTT_THRESH. Smaller values of MIN_RTT_THRESH may cause spurious MAX_RTT_THRESH. Smaller values of MIN_RTT_THRESH may cause spurious
exits from slow start. Larger values of MAX_RTT_THRESH may result in exits from slow start. Larger values of MAX_RTT_THRESH may result in
slow start not exiting until loss is encountered for connections on slow start not exiting until loss is encountered for connections on
large RTT paths. large RTT paths.
A TCP implementation is required to take at least one RTT sample each A TCP implementation is required to take at least one RTT sample each
round. Using lower values of N_RTT_SAMPLE will lower the accuracy of round. Using lower values of N_RTT_SAMPLE will lower the accuracy of
the measured RTT for the round; higher values will improve accuracy the measured RTT for the round; higher values will improve accuracy
at the cost of more processing. at the cost of more processing.
The minimum value of CSS_GROWTH_DIVISOR SHOULD be at least 2. The minimum value of CSS_GROWTH_DIVISOR MUST be at least 2. A value
Otherwise the cwnd growth could again become too aggressive and cause of 1 results in the same aggressive behavior as regular slow start.
ideal send rate overshoot. Values larger than 4 will cause the Values larger than 4 will cause the algorithm to be less aggressive
algorithm to be less aggressive and maybe less performant. and maybe less performant.
Smaller values of CSS_ROUNDS may miss detecting jitter and larger Smaller values of CSS_ROUNDS may miss detecting jitter and larger
values may limit performance. values may limit performance.
An implementation SHOULD use HyStart++ only for the initial slow An implementation SHOULD use HyStart++ only for the initial slow
start (when ssthresh is at its initial value of arbitrarily high per start (when ssthresh is at its initial value of arbitrarily high per
[RFC5681]) and fall back to using traditional slow start for the [RFC5681]) and fall back to using traditional slow start for the
remainder of the connection lifetime. This is acceptable because remainder of the connection lifetime. This is acceptable because
subsequent slow starts will use the discovered ssthresh value to exit subsequent slow starts will use the discovered ssthresh value to exit
slow start and avoid the overshoot problem. An implementation MAY slow start and avoid the overshoot problem. An implementation MAY
use HyStart++ to grow the restart window ([RFC5681]) after a long use HyStart++ to grow the restart window ([RFC5681]) after a long
idle period. idle period.
5. Deployments and Performance Evaluations 5. Deployments and Performance Evaluations
As of the time of writing, HyStart++ has been default enabled for all As of the time of writing, HyStart++ draft 01 was default enabled for
TCP connections in Windows for two years. The original Hystart has all TCP connections in Windows for two years. The original Hystart
been default-enabled for all TCP connections in Linux TCP for a has been default-enabled for all TCP connections using the default
decade. congestion control module CUBIC ([RFC8312]) for a decade.
In lab measurements with Windows TCP, HyStart++ shows both goodput In lab measurements with Windows TCP, HyStart++ shows both goodput
improvements as well as reductions in packet loss and improvements as well as reductions in packet loss and
retransmissions. For example across a variety of tests on a 100 Mbps retransmissions. For example across a variety of tests on a 100 Mbps
link with a bottleneck buffer size of bandwidth-delay product, link with a bottleneck buffer size of bandwidth-delay product,
HyStart++ reduces bytes retransmitted by 50% and retransmission HyStart++ reduces bytes retransmitted by 50% and retransmission
timeouts by 36%. timeouts by 36%.
In an A/B test across a large Windows device population, out of 52 In an A/B test for HyStart++ draft 01 across a large Windows device
billion TCP connections, 0.7% of connections move from 1 RTO to 0 population, out of 52 billion TCP connections, 0.7% of connections
RTOs and another 0.7% connections move from 2 RTOs to 1 RTO with move from 1 RTO to 0 RTOs and another 0.7% connections move from 2
HyStart++. This test did not focus on send heavy connections and the RTOs to 1 RTO with HyStart++. This test did not focus on send heavy
impact on send heavy connections is likely much higher. We plan to connections and the impact on send heavy connections is likely much
conduct more such production experiments to gather more data in the higher. We plan to conduct more such production experiments to
future. gather more data in the future.
6. Security Considerations 6. Security Considerations
HyStart++ enhances slow start and inherits the general security HyStart++ enhances slow start and inherits the general security
considerations discussed in [RFC5681]. considerations discussed in [RFC5681].
7. IANA Considerations 7. IANA Considerations
This document has no actions for IANA. This document has no actions for IANA.
skipping to change at page 8, line 22 skipping to change at page 8, line 22
8.2. Informative References 8.2. Informative References
[HyStart] Ha, S. and I. Ree, "Hybrid Slow Start for High-Bandwidth [HyStart] Ha, S. and I. Ree, "Hybrid Slow Start for High-Bandwidth
and Long-Distance Networks", and Long-Distance Networks",
DOI 10.1145/1851182.1851192, International Workshop on DOI 10.1145/1851182.1851192, International Workshop on
Protocols for Fast Long-Distance Networks, 2008, Protocols for Fast Long-Distance Networks, 2008,
<https://pdfs.semanticscholar.org/25e9/ <https://pdfs.semanticscholar.org/25e9/
ef3f03315782c7f1cbcd31b587857adae7d1.pdf>. ef3f03315782c7f1cbcd31b587857adae7d1.pdf>.
[RFC8312] Rhee, I., Xu, L., Ha, S., Zimmermann, A., Eggert, L., and
R. Scheffenegger, "CUBIC for Fast Long-Distance Networks",
RFC 8312, DOI 10.17487/RFC8312, February 2018,
<https://www.rfc-editor.org/info/rfc8312>.
Authors' Addresses Authors' Addresses
Praveen Balasubramanian Praveen Balasubramanian
Microsoft Microsoft
One Microsoft Way One Microsoft Way
Redmond, WA 98052 Redmond, WA 98052
USA United States of America
Phone: +1 425 538 2782 Phone: +1 425 538 2782
Email: pravb@microsoft.com Email: pravb@microsoft.com
Yi Huang Yi Huang
Microsoft Microsoft
Phone: +1 425 703 0447 Phone: +1 425 703 0447
Email: huanyi@microsoft.com Email: huanyi@microsoft.com
 End of changes. 44 change blocks. 
94 lines changed or deleted 97 lines changed or added

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