draft-ietf-tcpm-rack-03.txt   draft-ietf-tcpm-rack-04.txt 
TCP Maintenance Working Group Y. Cheng TCP Maintenance Working Group Y. Cheng
Internet-Draft N. Cardwell Internet-Draft N. Cardwell
Intended status: Experimental N. Dukkipati Intended status: Experimental N. Dukkipati
Expires: September 6, 2018 P. Jha Expires: January 3, 2019 P. Jha
Google, Inc Google, Inc
March 5, 2018 July 2, 2018
RACK: a time-based fast loss detection algorithm for TCP RACK: a time-based fast loss detection algorithm for TCP
draft-ietf-tcpm-rack-03 draft-ietf-tcpm-rack-04
Abstract Abstract
This document presents a new TCP loss detection algorithm called RACK This document presents a new TCP loss detection algorithm called RACK
("Recent ACKnowledgment"). RACK uses the notion of time, instead of ("Recent ACKnowledgment"). RACK uses the notion of time, instead of
packet or sequence counts, to detect losses, for modern TCP packet or sequence counts, to detect losses, for modern TCP
implementations that can support per-packet timestamps and the implementations that can support per-packet timestamps and the
selective acknowledgment (SACK) option. It is intended to replace selective acknowledgment (SACK) option. It is intended to replace
the conventional DUPACK threshold approach and its variants, as well the conventional DUPACK threshold approach and its variants, as well
as other nonstandard approaches. as other nonstandard approaches.
skipping to change at page 1, line 38 skipping to change at page 1, line 38
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 September 6, 2018. This Internet-Draft will expire on January 3, 2019.
Copyright Notice Copyright Notice
Copyright (c) 2018 IETF Trust and the persons identified as the Copyright (c) 2018 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/license-info) in effect on the date of (https://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
skipping to change at page 2, line 38 skipping to change at page 2, line 38
limited, so it cannot send new data to probe losses and has to limited, so it cannot send new data to probe losses and has to
rely on retransmission timeouts (RTOs). rely on retransmission timeouts (RTOs).
3. Reordering. Link layer protocols (e.g., 802.11 block ACK) or 3. Reordering. Link layer protocols (e.g., 802.11 block ACK) or
routers' internal load-balancing can deliver TCP packets out of routers' internal load-balancing can deliver TCP packets out of
order. The degree of such reordering is usually within the order order. The degree of such reordering is usually within the order
of the path round trip time. of the path round trip time.
Despite TCP stacks (e.g. Linux) that implement many of the standard Despite TCP stacks (e.g. Linux) that implement many of the standard
and proposed loss detection algorithms and proposed loss detection algorithms
[RFC3517][RFC4653][RFC5827][RFC5681][RFC6675][RFC7765][FACK][THIN- [RFC4653][RFC5827][RFC5681][RFC6675][RFC7765][FACK][THIN-STREAM],
STREAM][TLP], we've found that together they do not perform well. we've found that together they do not perform well. The main reason
The main reason is that many of them are based on the classic rule of is that many of them are based on the classic rule of counting
counting duplicate acknowledgments [RFC5681]. They can either detect duplicate acknowledgments [RFC5681]. They can either detect loss
loss quickly or accurately, but not both, especially when the sender quickly or accurately, but not both, especially when the sender is
is application-limited or under reordering that is unpredictable. application-limited or under reordering that is unpredictable. And
And under these conditions none of them can detect lost under these conditions none of them can detect lost retransmissions
retransmissions well. well.
Also, these algorithms, including RFCs, rarely address the Also, these algorithms, including RFCs, rarely address the
interactions with other algorithms. For example, FACK may consider a interactions with other algorithms. For example, FACK may consider a
packet is lost while RFC3517 may not. Implementing N algorithms packet is lost while RFC6675 may not. Implementing N algorithms
while dealing with N^2 interactions is a daunting task and error- while dealing with N^2 interactions is a daunting task and error-
prone. prone.
The goal of RACK is to solve all the problems above by replacing many The goal of RACK is to solve all the problems above by replacing many
of the loss detection algorithms above with one simpler, and also of the loss detection algorithms above with one more effective
more effective, algorithm. algorithm to handle loss and reordering.
2. Overview 2. Overview
The main idea behind RACK is that if a packet has been delivered out The main idea behind RACK is that if a packet has been delivered out
of order, then the packets sent chronologically before that were of order, then the packets sent chronologically before that were
either lost or reordered. This concept is not fundamentally either lost or reordered. This concept is not fundamentally
different from [RFC5681][RFC3517][FACK]. But the key innovation in different from [RFC5681][RFC6675][FACK]. But the key innovation in
RACK is to use a per-packet transmission timestamp and widely RACK is to use a per-packet transmission timestamp and widely
deployed SACK options to conduct time-based inferences instead of deployed SACK options to conduct time-based inferences instead of
inferring losses with packet or sequence counting approaches. inferring losses with packet or sequence counting approaches.
Using a threshold for counting duplicate acknowledgments (i.e., Using a threshold for counting duplicate acknowledgments (i.e.,
DupThresh) is no longer reliable because of today's prevalent DupThresh) alone is no longer reliable because of today's prevalent
reordering patterns. A common type of reordering is that the last reordering patterns. A common type of reordering is that the last
"runt" packet of a window's worth of packet bursts gets delivered "runt" packet of a window's worth of packet bursts gets delivered
first, then the rest arrive shortly after in order. To handle this first, then the rest arrive shortly after in order. To handle this
effectively, a sender would need to constantly adjust the DupThresh effectively, a sender would need to constantly adjust the DupThresh
to the burst size; but this would risk increasing the frequency of to the burst size; but this would risk increasing the frequency of
RTOs on real losses. RTOs on real losses.
Today's prevalent lost retransmissions also cause problems with Today's prevalent lost retransmissions also cause problems with
packet-counting approaches [RFC5681][RFC3517][FACK], since those packet-counting approaches [RFC5681][RFC6675][FACK], since those
approaches depend on reasoning in sequence number space. approaches depend on reasoning in sequence number space.
Retransmissions break the direct correspondence between ordering in Retransmissions break the direct correspondence between ordering in
sequence space and ordering in time. So when retransmissions are sequence space and ordering in time. So when retransmissions are
lost, sequence-based approaches are often unable to infer and quickly lost, sequence-based approaches are often unable to infer and quickly
repair losses that can be deduced with time-based approaches. repair losses that can be deduced with time-based approaches.
Instead of counting packets, RACK uses the most recently delivered Instead of counting packets, RACK uses the most recently delivered
packet's transmission time to judge if some packets sent previous to packet's transmission time to judge if some packets sent previous to
that time have "expired" by passing a certain reordering settling that time have "expired" by passing a certain reordering settling
window. On each ACK, RACK marks any already-expired packets lost, window. On each ACK, RACK marks any already-expired packets lost,
and for any packets that have not yet expired it waits until the and for any packets that have not yet expired it waits until the
reordering window passes and then marks those lost as well. In reordering window passes and then marks those lost as well. In
either case, RACK can repair the loss without waiting for a (long) either case, RACK can repair the loss without waiting for a (long)
RTO. RACK can be applied to both fast recovery and timeout recovery, RTO. RACK can be applied to both fast recovery and timeout recovery,
and can detect losses on both originally transmitted and and can detect losses on both originally transmitted and
retransmitted packets, making it a great all-weather loss detection retransmitted packets, making it a great all-weather loss detection
mechanism. mechanism.
3. Requirements 3. Design Rationale for Reordering Tolerance
The reordering behavior of networks can evolve (over years) in
response to the behavior of transport protocols and applications, as
well as the needs of network designers and operators. From a network
or link designer's viewpoint, parallelization (eg. link bonding) is
the easiest way to get a network to go faster. Therefore their main
constraint on speed is reordering, and there is pressure to relax
that constraint. If RACK becomes widely deployed, the underlying
networks may introduce more reordering for higher throughput. But
this may result in excessive reordering that hurts end to end
performance:
1. End host packet processing: extreme reordering on high-speed
networks would incur high CPU cost by greatly reducing the
effectiveness of aggregation mechanisms, such as large receive
offload (LRO) and generic receive offload (GRO), and
significantly increasing the number of ACKs.
2. Congestion control: TCP congestion control implicitly assumes the
feedback from ACKs are from the same bottleneck. Therefore it
cannot handle well scenarios where packets are traversing largely
disjoint paths.
3. Loss recovery: Having an excessively large reordering window to
accommodate widely different latencies from different paths would
increase the latency of loss recovery.
An end-to-end transport protocol cannot tell immediately whether a
hole is reordering or loss. It can only distinguish between the two
in hindsight if the hole in the sequence space gets filled later
without a retransmission. How long the sender waits for such
potential reordering events to settle is determined by the current
reordering window.
Given these considerations, a core design philosophy of RACK is to
adapt to the measured duration of reordering events, within
reasonable and specific bounds. To accomplish this RACK places the
following mandates on the reordering window:
1. The initial RACK reordering window SHOULD be set to a small
fraction of the round-trip time.
2. If no reordering has been observed, then RACK SHOULD honor the
classic 3-DUPACK rule for initiating fast recovery. One simple
way to implement this is to temporarily override the reorder
window to 0.
3. The RACK reordering window SHOULD leverage Duplicate Selective
Acknowledgement (DSACK) information [RFC3708] to adaptively
estimate the duration of reordering events.
4. The RACK reordering window MUST be bounded and this bound SHOULD
be one round trip.
As a flow starts, either condition 1 or condition 2 or both would
trigger RACK to start the recovery process quickly. The low initial
reordering window and use of the 3-DUPACK rule are key to achieving
low-latency loss recovery for short flows by risking spurious
retransmissions to recover losses quickly. This rationale is that
spurious retransmissions for short flows are not expected to produce
excessive network traffic.
For long flows the design tolerates reordering within a round trip.
This handles reordering caused by path divergence in small time
scales (reordering within the round-trip time of the shortest path),
which should tolerate much of the reordering from link bonding,
multipath routing, or link-layer out-of-order delivery. It also
relaxes ordering constraints to allow sending flights of TCP packets
on different paths dynamically for better load-balancing (e.g.
flowlets).
However, the fact that the initial RACK reordering window is low, and
the RACK reordering window's adaptive growth is bounded, means that
there will continue to be a cost to reordering and a limit to RACK's
adaptation to reordering. This maintains a disincentive for network
designers and operators to introduce needless or excessive
reordering, particularly because they have to allow for low round
trip time paths. This means RACK will not encourage networks to
perform inconsiderate fine-grained packet-spraying over highly
disjoint paths with very different characteristics. There are good
alternative solutions, such as MPTCP, for such networks.
To conclude, the RACK algorithm aims to adapt to small degrees of
reordering, quickly recover most losses within one to two round
trips, and avoid costly retransmission timeouts (RTOs). In the
presence of reordering, the adaptation algorithm can impose
sometimes-needless delays when it waits to disambiguate loss from
reordering, but the penalty for waiting is bounded to one round trip
and such delays are confined to longer-running flows.
This document provides a concrete and detailed reordering window
adaptation algorithm for implementors. We note that the specifics of
the algorithm are likely to evolve over time. But that is a separate
engineering optimization that's out of scope for this document.
4. Requirements
The reader is expected to be familiar with the definitions given in The reader is expected to be familiar with the definitions given in
the TCP congestion control [RFC5681] and selective acknowledgment the TCP congestion control [RFC5681] and selective acknowledgment
[RFC2018] RFCs. Familiarity with the conservative SACK-based [RFC2018] RFCs. Familiarity with the conservative SACK-based
recovery for TCP [RFC6675] is not expected but helps. recovery for TCP [RFC6675] is not expected but helps.
RACK has three requirements: RACK has three requirements:
1. The connection MUST use selective acknowledgment (SACK) options 1. The connection MUST use selective acknowledgment (SACK) options
[RFC2018]. [RFC2018].
2. For each packet sent, the sender MUST store its most recent 2. For each packet sent, the sender MUST store its most recent
transmission time with (at least) millisecond granularity. For transmission time with (at least) millisecond granularity. For
skipping to change at page 4, line 31 skipping to change at page 6, line 35
We assume that requirement 1 implies the sender keeps a SACK We assume that requirement 1 implies the sender keeps a SACK
scoreboard, which is a data structure to store selective scoreboard, which is a data structure to store selective
acknowledgment information on a per-connection basis ([RFC6675] acknowledgment information on a per-connection basis ([RFC6675]
section 3). For the ease of explaining the algorithm, we use a section 3). For the ease of explaining the algorithm, we use a
pseudo-scoreboard that manages the data in sequence number ranges. pseudo-scoreboard that manages the data in sequence number ranges.
But the specifics of the data structure are left to the implementor. But the specifics of the data structure are left to the implementor.
RACK does not need any change on the receiver. RACK does not need any change on the receiver.
4. Definitions of variables 5. Definitions of variables
A sender needs to store these new RACK variables: A sender needs to store these new RACK variables:
"Packet.xmit_ts" is the time of the last transmission of a data "Packet.xmit_ts" is the time of the last transmission of a data
packet, including retransmissions, if any. The sender needs to packet, including retransmissions, if any. The sender needs to
record the transmission time for each packet sent and not yet record the transmission time for each packet sent and not yet
acknowledged. The time MUST be stored at millisecond granularity or acknowledged. The time MUST be stored at millisecond granularity or
finer. finer.
"RACK.packet". Among all the packets that have been either "RACK.packet". Among all the packets that have been either
selectively or cumulatively acknowledged, RACK.packet is the one that selectively or cumulatively acknowledged, RACK.packet is the one that
was sent most recently (including retransmissions). was sent most recently (including retransmissions).
"RACK.xmit_ts" is the latest transmission timestamp of RACK.packet. "RACK.xmit_ts" is the latest transmission timestamp of RACK.packet.
"RACK.end_seq" is the ending TCP sequence number of RACK.packet. "RACK.end_seq" is the ending TCP sequence number of RACK.packet.
"RACK.RTT" is the associated RTT measured when RACK.xmit_ts, above, "RACK.rtt" is the RTT of the most recently transmitted packet that
was changed. It is the RTT of the most recently transmitted packet has been delivered (either cumulatively acknowledged or selectively
that has been delivered (either cumulatively acknowledged or acknowledged) on the connection.
selectively acknowledged) on the connection.
"RACK.reo_wnd" is a reordering window for the connection, computed in "RACK.rtt_seq" is the SND.NXT when RACK.rtt is updated.
the unit of time used for recording packet transmission times. It is
used to defer the moment at which RACK marks a packet lost. "RACK.reo_wnd" is a reordering window computed in the unit of time
used for recording packet transmission times. It is used to defer
the moment at which RACK marks a packet lost.
"RACK.min_RTT" is the estimated minimum round-trip time (RTT) of the "RACK.min_RTT" is the estimated minimum round-trip time (RTT) of the
connection. connection.
"RACK.ack_ts" is the time when all the sequences in RACK.packet were "RACK.ack_ts" is the time when all the sequences in RACK.packet were
selectively or cumulatively acknowledged. selectively or cumulatively acknowledged.
"RACK.reo_wnd_incr" is the multiplier applied to adjust RACK.reo_wnd "RACK.reo_wnd_incr" is the multiplier applied to adjust RACK.reo_wnd
"RACK.reo_wnd_persist" is the number of loss recoveries before "RACK.reo_wnd_persist" is the number of loss recoveries before
resetting RACK.reo_wnd "RACK.dsack" indicates if RACK.reo_wnd has resetting RACK.reo_wnd "RACK.dsack" indicates if a DSACK option has
been adjusted upon receiving a DSACK option been received since last RACK.reo_wnd change "RACK.pkts_sacked"
returns the total number of packets selectively acknowledged in the
SACK scoreboard.
"RACK.reord" indicates the connection has detected packet reordering
event(s)
"RACK.fack" is the highest selectively or cumulatively acknowledged
sequence
Note that the Packet.xmit_ts variable is per packet in flight. The Note that the Packet.xmit_ts variable is per packet in flight. The
RACK.xmit_ts, RACK.end_seq, RACK.RTT, RACK.reo_wnd, and RACK.min_RTT RACK.xmit_ts, RACK.end_seq, RACK.rtt, RACK.reo_wnd, and RACK.min_RTT
variables are kept in the per-connection TCP control block. variables are kept in the per-connection TCP control block.
RACK.packet and RACK.ack_ts are used as local variables in the RACK.packet and RACK.ack_ts are used as local variables in the
algorithm. algorithm.
5. Algorithm Details 6. Algorithm Details
5.1. Transmitting a data packet 6.1. Transmitting a data packet
Upon transmitting a new packet or retransmitting an old packet, Upon transmitting a new packet or retransmitting an old packet,
record the time in Packet.xmit_ts. RACK does not care if the record the time in Packet.xmit_ts. RACK does not care if the
retransmission is triggered by an ACK, new application data, an RTO, retransmission is triggered by an ACK, new application data, an RTO,
or any other means. or any other means.
5.2. Upon receiving an ACK 6.2. Upon receiving an ACK
Step 1: Update RACK.min_RTT. Step 1: Update RACK.min_RTT.
Use the RTT measurements obtained via [RFC6298] or [RFC7323] to Use the RTT measurements obtained via [RFC6298] or [RFC7323] to
update the estimated minimum RTT in RACK.min_RTT. The sender can update the estimated minimum RTT in RACK.min_RTT. The sender can
track a simple global minimum of all RTT measurements from the track a simple global minimum of all RTT measurements from the
connection, or a windowed min-filtered value of recent RTT connection, or a windowed min-filtered value of recent RTT
measurements. This document does not specify an exact approach. measurements. This document does not specify an exact approach.
Step 2: Update RACK stats Step 2: Update RACK stats
skipping to change at page 6, line 10 skipping to change at page 8, line 28
ACKed or SACKed is marked as delivered in the scoreboard. Among all ACKed or SACKed is marked as delivered in the scoreboard. Among all
the packets newly ACKed or SACKed in the connection, record the most the packets newly ACKed or SACKed in the connection, record the most
recent Packet.xmit_ts in RACK.xmit_ts if it is ahead of RACK.xmit_ts. recent Packet.xmit_ts in RACK.xmit_ts if it is ahead of RACK.xmit_ts.
Sometimes the timestamps of RACK.Packet and Packet could carry the Sometimes the timestamps of RACK.Packet and Packet could carry the
same transmit timestamps due to clock granularity or segmentation same transmit timestamps due to clock granularity or segmentation
offloading (i.e. the two packets were sent as a jumbo frame into the offloading (i.e. the two packets were sent as a jumbo frame into the
NIC). In that case the sequence numbers of RACK.end_seq and NIC). In that case the sequence numbers of RACK.end_seq and
Packet.end_seq are compared to break the tie. Packet.end_seq are compared to break the tie.
Since an ACK can also acknowledge retransmitted data packets, Since an ACK can also acknowledge retransmitted data packets,
RACK.RTT can be vastly underestimated if the retransmission was RACK.rtt can be vastly underestimated if the retransmission was
spurious. To avoid that, ignore a packet if any of its TCP sequences spurious. To avoid that, ignore a packet if any of its TCP sequences
have been retransmitted before and either of two conditions is true: have been retransmitted before and either of two conditions is true:
1. The Timestamp Echo Reply field (TSecr) of the ACK's timestamp 1. The Timestamp Echo Reply field (TSecr) of the ACK's timestamp
option [RFC7323], if available, indicates the ACK was not option [RFC7323], if available, indicates the ACK was not
acknowledging the last retransmission of the packet. acknowledging the last retransmission of the packet.
2. The packet was last retransmitted less than RACK.min_rtt ago. 2. The packet was last retransmitted less than RACK.min_rtt ago.
While it is still possible the packet is spuriously retransmitted
because of a recent RTT decrease, we believe that our experience
suggests this is a reasonable heuristic.
If the ACK is not ignored as invalid, update the RACK.RTT to be the If the ACK is not ignored as invalid, update the RACK.rtt to be the
RTT sample calculated using this ACK, and continue. If this ACK or RTT sample calculated using this ACK, and continue. If this ACK or
SACK was for the most recently sent packet, then record the SACK was for the most recently sent packet, then record the
RACK.xmit_ts timestamp and RACK.end_seq sequence implied by this ACK. RACK.xmit_ts timestamp and RACK.end_seq sequence implied by this ACK.
Otherwise exit here and omit the following steps. Otherwise exit here and omit the following steps.
Notice that the second condition above is a heuristic. This
heuristic would fail to update RACK stats if the packet is spuriously
retransmitted because of a recent minimum RTT decrease (e.g. path
change). Consequentially RACK may not detect losses from ACK events
and the recovery resorts to the (slower) TLP or RTO timer-based
events. However such events should be rare and the connection would
pick up the new minimum RTT when the recovery ends to avoid repeated
similar failures.
Step 2 may be summarized in pseudocode as: Step 2 may be summarized in pseudocode as:
RACK_sent_after(t1, seq1, t2, seq2): RACK_sent_after(t1, seq1, t2, seq2):
If t1 > t2: If t1 > t2:
Return true Return true
Else if t1 == t2 AND seq1 > seq2: Else if t1 == t2 AND seq1 > seq2:
Return true Return true
Else: Else:
Return false Return false
RACK_update(): RACK_update():
For each Packet newly acknowledged cumulatively or selectively: For each Packet newly acknowledged cumulatively or selectively:
rtt = Now() - RACK.xmit_ts rtt = Now() - Packet.xmit_ts
If Packet has been retransmitted: If Packet.retransmitted is TRUE:
If ACK.ts_option.echo_reply < Packet.xmit_ts: If ACK.ts_option.echo_reply < Packet.xmit_ts:
Return Return
If rtt < RACK.min_rtt: If rtt < RACK.min_rtt:
Return Return
RACK.RTT = rtt RACK.rtt = rtt
If RACK_sent_after(Packet.xmit_ts, Packet.end_seq If RACK_sent_after(Packet.xmit_ts, Packet.end_seq
RACK.xmit_ts, RACK.end_seq): RACK.xmit_ts, RACK.end_seq):
RACK.xmit_ts = Packet.xmit_ts RACK.xmit_ts = Packet.xmit_ts
RACK.end_seq = Packet.end_seq
Step 3: Update RACK reordering window Step 3: Detect packet reordering
To handle the prevalent small degree of reordering, RACK.reo_wnd To detect reordering, the sender looks for original data packets
serves as an allowance for settling time before marking a packet being delivered out of order in sequence space. The sender tracks
lost. Use a conservative window of min_RTT / 4 if the connection is the highest sequence selectively or cumulatively acknowledged in the
not currently in loss recovery. When in loss recovery, use a RACK.fack variable. The name fack stands for the most forward ACK
RACK.reo_wnd of zero in order to retransmit quickly. originated from the [FACK] draft. If the ACK selectively or
cumulatively acknowledges an unacknowledged and also never
retransmitted sequence below RACK.fack, then the corresponding packet
has been reordered and RACK.reord is set to 1.
Extension 1: Optionally size the window based on DSACK Further, the The heuristic above only detects reordering if the re-ordered packet
sender MAY leverage DSACK [RFC3708] to adapt the reordering window to has not yet been retransmitted. This is a major drawback because if
higher degrees of reordering. Receiving an ACK with a DSACK RACK has a low reordering window and the network is reordering
indicates a spurious retransmission, which in turn suggests that the packets, RACK may falsely retransmit frequently. Consequently RACK
RACK reordering window, RACK.reo_wnd, is likely too small. The may fail to detect reordering to increase the reordering window,
sender MAY increase the RACK.reo_wnd window linearly for every round because the reordered packets were already (falsely) retransmitted.
trip in which the sender receives a DSACK, so that after N distinct
round trips in which a DSACK is received, the RACK.reo_wnd is N *
min_RTT / 4. The inflated RACK.reo_wnd would persist for 16 loss
recoveries and then reset to its starting value, min_RTT / 4.
Extension 2: Optionally size the window if reordering has been DSACK [RFC3708] can help mitigate this issue. The false
observed retransmission would solicit DSACK option in the ACK. Therefore if
the ACK has a DSACK option covering some sequence that were both
acknowledged and retransmitted, this implies the original packet was
reordered but RACK retransmitted the packet too quickly and should
set RACK.reord to 1.
If the reordering window is too small or the connection does not RACK_detect_reordering():
support DSACK, then RACK can trigger spurious loss recoveries and For each Packet newly acknowledged cumulatively or selectively:
reduce the congestion window unnecessarily. If the implementation If Packet.end_seq > RACK.fack:
supports reordering detection such as [REORDER-DETECT], then the RACK.fack = Packet.end_seq
sender MAY use the dynamically-sized reordering window based on Else if Packet.end_seq < RACK.fack AND
min_RTT during loss recovery instead of a zero reordering window to Packet.retransmitted is FALSE:
compensate. Extension 3: Optionally size the window with the classic RACK.reord = TRUE
DUPACK threshold heuristic The DUPACK threshold approach in the For each Packet covered by the DSACK option:
current standards [RFC5681][RFC6675] is simple, and for decades has If Packet.retransmitted is TRUE:
been effective in quickly detecting losses, despite the drawbacks RACK.reord = TRUE
discussed earlier. RACK can easily maintain the DUPACK threshold's
advantages of quick detection by resetting the reordering window to
zero (using RACK.reo_wnd = 0) when the DUPACK threshold is met (i.e.
when at least three packets have been selectively acknowledged). The
subtle differences are discussed in the section "RACK and TLP
discussions".
The following algorithm includes the basic and all the extensions Step 4: Update RACK reordering window
mentioned above. Note that individual extensions that require
additional TCP features (e.g. DSACK) would work if the feature
functions simply return false.
RACK_update_reo_wnd: To handle the prevalent small degree of reordering, RACK.reo_wnd
RACK.min_RTT = TCP_min_RTT() serves as an allowance for settling time before marking a packet
If RACK_ext_TCP_ACK_has_DSACK_option(): lost. This section documents a detailed algorithm following the
RACK.dsack = true design rationale section. RACK starts initially with a conservative
window of min_RTT/4. If no reordering has been observed, RACK uses
RACK.reo_wnd of 0 during loss recovery, in order to retransmit
quickly, or when the number of DUPACKs exceeds the classic DUPACK
threshold. The subtle difference of this approach and conventional
one [RFC5681][RFC6675] is dicussed later in the section of "RACK and
TLP discussions".
If SND.UNA < RACK.roundtrip_seq: Further, RACK MAY use DSACK [RFC3708] to adapt the reordering window,
RACK.dsack = false /* React to DSACK once within a round trip */ to higher degrees of reordering, if DSACK is supported. Receiving an
ACK with a DSACK indicates a spurious retransmission, which in turn
suggests that the RACK reordering window, RACK.reo_wnd, is likely too
small. The sender MAY increase the RACK.reo_wnd window linearly for
every round trip in which the sender receives a DSACK, so that after
N distinct round trips in which a DSACK is received, the RACK.reo_wnd
becomes (N+1) * min_RTT / 4, with an upper-bound of SRTT. The
inflated RACK.reo_wnd would persist for 16 loss recoveries and after
which it resets to its starting value, min_RTT / 4.
If RACK.dsack: The following pseudocode implements above algorithm. Note that
RACK.reo_wnd_incr += 1 extensions that require additional TCP features (e.g. DSACK) would
RACK.dsack = false work if the feature functions simply return false.
RACK.roundtrip_seq = SND.NXT
RACK.reo_wnd_persist = 16 /* Keep window for 16 loss recoveries */
Else if exiting loss recovery:
RACK.reo_wnd_persist -= 1
If RACK.reo_wnd_persist <= 0:
RACK.reo_wnd_incr = 1
If in loss recovery and not RACK_ext_TCP_seen_reordering(): RACK_update_reo_wnd():
RACK.reo_wnd = 0 RACK.min_RTT = TCP_min_RTT()
Else if RACK_ext_TCP_dupack_threshold_hit(): /* DUPTHRESH emulation mode */ If DSACK option is present:
RACK.reo_wnd = 0 RACK.dsack = true
Else:
RACK.reo_wnd = RACK.min_RTT / 4 * RACK.reo_wnd_incr
RACK.reo_wnd = min(RACK.reo_wnd, SRTT)
Step 4: Detect losses. If SND.UNA < RACK.rtt_seq:
RACK.dsack = false /* React to DSACK once per round trip */
If RACK.dsack:
RACK.reo_wnd_incr += 1
RACK.dsack = false
RACK.rtt_seq = SND.NXT
RACK.reo_wnd_persist = 16 /* Keep window for 16 recoveries */
Else if exiting loss recovery:
RACK.reo_wnd_persist -= 1
If RACK.reo_wnd_persist <= 0:
RACK.reo_wnd_incr = 1
If RACK.reordering_seen is FALSE:
If in loss recovery: /* If in fast or timeout recovery */
RACK.reo_wnd = 0
Return
Else if RACK.pkts_sacked >= RACK.dupthresh:
RACK.reo_wnd = 0
return
RACK.reo_wnd = RACK.min_RTT / 4 * RACK.reo_wnd_incr
RACK.reo_wnd = min(RACK.reo_wnd, SRTT)
Step 5: Detect losses.
For each packet that has not been SACKed, if RACK.xmit_ts is after For each packet that has not been SACKed, if RACK.xmit_ts is after
Packet.xmit_ts + RACK.reo_wnd, then mark the packet (or its Packet.xmit_ts + RACK.reo_wnd, then mark the packet (or its
corresponding sequence range) lost in the scoreboard. The rationale corresponding sequence range) lost in the scoreboard. The rationale
is that if another packet that was sent later has been delivered, and is that if another packet that was sent later has been delivered, and
the reordering window or "reordering settling time" has already the reordering window or "reordering settling time" has already
passed, then the packet was likely lost. passed, then the packet was likely lost.
If another packet that was sent later has been delivered, but the If another packet that was sent later has been delivered, but the
reordering window has not passed, then it is not yet safe to deem the reordering window has not passed, then it is not yet safe to deem the
skipping to change at page 9, line 38 skipping to change at page 12, line 20
lost if the reordering window for a packet has elapsed through the lost if the reordering window for a packet has elapsed through the
sum of: sum of:
1. delta in transmit time between a packet and the RACK.packet 1. delta in transmit time between a packet and the RACK.packet
2. delta in time between RACK.ack_ts and now 2. delta in time between RACK.ack_ts and now
So we mark a packet as lost if: So we mark a packet as lost if:
RACK.xmit_ts >= Packet.xmit_ts RACK.xmit_ts >= Packet.xmit_ts
AND AND
(RACK.xmit_ts - Packet.xmit_ts) + (now - RACK.ack_ts) >= RACK.reo_wnd (RACK.xmit_ts - Packet.xmit_ts) + (now - RACK.ack_ts) >= RACK.reo_wnd
If we solve this second condition for "now", the moment at which we If we solve this second condition for "now", the moment at which we
can declare a packet lost, then we get: can declare a packet lost, then we get:
now >= Packet.xmit_ts + RACK.reo_wnd + (RACK.ack_ts - RACK.xmit_ts) now >= Packet.xmit_ts + RACK.reo_wnd + (RACK.ack_ts - RACK.xmit_ts)
Then (RACK.ack_ts - RACK.xmit_ts) is just the RTT of the packet we Then (RACK.ack_ts - RACK.xmit_ts) is just the RTT of the packet we
used to set RACK.xmit_ts, so this reduces to: used to set RACK.xmit_ts, so this reduces to:
Packet.xmit_ts + RACK.RTT + RACK.reo_wnd - now <= 0 Packet.xmit_ts + RACK.rtt + RACK.reo_wnd - now <= 0
The following pseudocode implements the algorithm above. When an ACK The following pseudocode implements the algorithm above. When an ACK
is received or the RACK timer expires, call RACK_detect_loss(). The is received or the RACK timer expires, call RACK_detect_loss(). The
algorithm includes an additional optimization to break timestamp ties algorithm includes an additional optimization to break timestamp ties
by using the TCP sequence space. The optimization is particularly by using the TCP sequence space. The optimization is particularly
useful to detect losses in a timely manner with TCP Segmentation useful to detect losses in a timely manner with TCP Segmentation
Offload, where multiple packets in one TSO blob have identical Offload, where multiple packets in one TSO blob have identical
timestamps. It is also useful when the timestamp clock granularity timestamps. It is also useful when the timestamp clock granularity
is close to or longer than the actual round trip time. is close to or longer than the actual round trip time.
RACK_detect_loss(): RACK_detect_loss():
timeout = 0 timeout = 0
For each packet, Packet, in the scoreboard: For each packet, Packet, not acknowledged yet:
If Packet is already SACKed If Packet.lost is TRUE or Packet.retransmitted is FALSE:
or marked lost and not yet retransmitted: Continue /* Lost packet not retransmitted yet */
Continue
If RACK_sent_after(RACK.xmit_ts, RACK.end_seq, If RACK_sent_after(RACK.xmit_ts, RACK.end_seq,
Packet.xmit_ts, Packet.end_seq): Packet.xmit_ts, Packet.end_seq):
remaining = Packet.xmit_ts + RACK.RTT + RACK.reo_wnd - Now() remaining = Packet.xmit_ts + RACK.rtt + RACK.reo_wnd - Now()
If remaining <= 0: If remaining <= 0:
Mark Packet lost Packet.lost = TRUE
Else: Else:
timeout = max(remaining, timeout) timeout = max(remaining, timeout)
If timeout != 0 If timeout != 0
Arm a timer to call RACK_detect_loss() after timeout Arm a timer to call RACK_detect_loss() after timeout
Implementation optimization: looping through packets in the SACK Implementation optimization: looping through packets in the SACK
scoreboard above could be very costly on large BDP networks since the scoreboard above could be very costly on large-BDP networks since the
inflight could be very large. If the implementation can organize the inflight could be very large. If the implementation can organize the
scoreboard data structures to have packets sorted by the last scoreboard data structures to have packets sorted by the last
(re)transmission time, then the loop can start on the least recently (re)transmission time, then the loop can start on the least recently
sent packet and aborts on the first packet sent after RACK.time_ts. sent packet and abort on the first packet sent after RACK.time_ts.
This can be implemented by using a seperate list sorted in time This can be implemented by using a seperate list sorted in time
order. The implementation inserts the packet to the tail of the list order. The implementation inserts the packet at the tail of the list
when it is (re)transmitted, and removes a packet from the list when when it is (re)transmitted, and removes a packet from the list when
it is delivered or marked lost. We RECOMMEND such an optimization it is delivered or marked lost. We RECOMMEND such an optimization
for implementations for support high BDP networks. The optimization because it enables implementations to support high-BDP networks.
is implemented in Linux and sees orders of magnitude improvement on This optimization is implemented in Linux and sees orders of
CPU usage on high speed WAN networks. magnitude improvement in CPU usage on high-speed WAN networks.
Tail Loss Probe: fast recovery on tail losses 6.3. Tail Loss Probe: fast recovery for tail losses
This section describes a supplemental algorithm, Tail Loss Probe This section describes a supplemental algorithm, Tail Loss Probe
(TLP), which leverages RACK to further reduce RTO recoveries. TLP (TLP), which leverages RACK to further reduce RTO recoveries. TLP
triggers fast recovery to quickly repair tail losses that can triggers fast recovery to quickly repair tail losses that can
otherwise be recovered by RTOs only. After an original data otherwise be recovered by RTOs only. After an original data
transmission, TLP sends a probe data segment within one to two RTTs. transmission, TLP sends a probe data segment within one to two RTTs.
The probe data segment can either be new, previously unsent data, or The probe data segment can either be new, previously unsent data, or
a retransmission of previously sent data just below SND.NXT. In a retransmission of previously sent data just below SND.NXT. In
either case the goal is to elicit more feedback from the receiver, in either case the goal is to elicit more feedback from the receiver, in
the form of an ACK (potentially with SACK blocks), to allow RACK to the form of an ACK (potentially with SACK blocks), to allow RACK to
trigger fast recovery instead of an RTO. trigger fast recovery instead of an RTO.
An RTO occurs when the first unacknowledged sequence number is not An RTO occurs when the first unacknowledged sequence number is not
acknowledged after a conservative period of time has elapsed acknowledged after a conservative period of time has elapsed
[RFC6298]. Common causes of RTOs include: [RFC6298]. Common causes of RTOs include:
1. The entire flight is lost 1. The entire flight of data is lost.
2. Tail losses at the end of an application transaction 2. Tail losses of data segments at the end of an application
transaction.
3. Lost retransmits, which can halt fast recovery based on [RFC6675] 3. Tail losses of ACKs at the end of an application transaction.
4. Lost retransmits, which can halt fast recovery based on [RFC6675]
if the ACK stream completely dries up. For example, consider a if the ACK stream completely dries up. For example, consider a
window of three data packets (P1, P2, P3) that are sent; P1 and window of three data packets (P1, P2, P3) that are sent; P1 and
P2 are dropped. On receipt of a SACK for P3, RACK marks P1 and P2 are dropped. On receipt of a SACK for P3, RACK marks P1 and
P2 as lost and retransmits them as R1 and R2. Suppose R1 and R2 P2 as lost and retransmits them as R1 and R2. Suppose R1 and R2
are lost as well, so there are no more returning ACKs to detect are lost as well, so there are no more returning ACKs to detect
R1 and R2 as lost. Recovery stalls. R1 and R2 as lost. Recovery stalls.
4. Tail losses of ACKs.
5. An unexpectedly long round-trip time (RTT). This can cause ACKs 5. An unexpectedly long round-trip time (RTT). This can cause ACKs
to arrive after the RTO timer expires. The F-RTO algorithm to arrive after the RTO timer expires. The F-RTO algorithm
[RFC5682] is designed to detect such spurious retransmission [RFC5682] is designed to detect such spurious retransmission
timeouts and at least partially undo the consequences of such timeouts and at least partially undo the consequences of such
events, but F-RTO cannot be used in many situations. events, but F-RTO cannot be used in many situations.
5.3. Tail Loss Probe: An Example 6.4. Tail Loss Probe: An Example
Following is an example of TLP. All events listed are at a TCP Following is an example of TLP. All events listed are at a TCP
sender. sender.
1. Sender transmits segments 1-10: 1, 2, 3, ..., 8, 9, 10. There is 1. Sender transmits segments 1-10: 1, 2, 3, ..., 8, 9, 10. There is
no more new data to transmit. A PTO is scheduled to fire in 2 no more new data to transmit. A PTO is scheduled to fire in 2
RTTs, after the transmission of the 10th segment. RTTs, after the transmission of the 10th segment.
2. Sender receives acknowledgements (ACKs) for segments 1-5; 2. Sender receives acknowledgements (ACKs) for segments 1-5;
segments 6-10 are lost and no ACKs are received. The sender segments 6-10 are lost and no ACKs are received. The sender
skipping to change at page 12, line 12 skipping to change at page 15, line 5
3. When PTO fires, sender retransmits segment 10. 3. When PTO fires, sender retransmits segment 10.
4. After an RTT, a SACK for packet 10 arrives. The ACK also carries 4. After an RTT, a SACK for packet 10 arrives. The ACK also carries
SACK holes for segments 6, 7, 8 and 9. This triggers RACK-based SACK holes for segments 6, 7, 8 and 9. This triggers RACK-based
loss recovery. loss recovery.
5. The connection enters fast recovery and retransmits the remaining 5. The connection enters fast recovery and retransmits the remaining
lost segments. lost segments.
5.4. Tail Loss Probe Algorithm Details 6.5. Tail Loss Probe Algorithm Details
We define the terminology used in specifying the TLP algorithm: We define the terminology used in specifying the TLP algorithm:
FlightSize: amount of outstanding data in the network, as defined in FlightSize: amount of outstanding data in the network, as defined in
[RFC5681]. [RFC5681].
RTO: The transport's retransmission timeout (RTO) is based on RTO: The transport's retransmission timeout (RTO) is based on
measured round-trip times (RTT) between the sender and receiver, as measured round-trip times (RTT) between the sender and receiver, as
specified in [RFC6298] for TCP. PTO: Probe timeout (PTO) is a timer specified in [RFC6298] for TCP. PTO: Probe timeout (PTO) is a timer
event indicating that an ACK is overdue. Its value is constrained to event indicating that an ACK is overdue. Its value is constrained to
be smaller than or equal to an RTO. be smaller than or equal to an RTO.
SRTT: smoothed round-trip time, computed as specified in [RFC6298]. SRTT: smoothed round-trip time, computed as specified in [RFC6298].
Open state: the sender's loss recovery state machine is in its
normal, default state: there are no SACKed sequence ranges in the
SACK scoreboard, and neither fast recovery, timeout-based recovery,
nor ECN-based cwnd reduction are underway.
The TLP algorithm has three phases, which we discuss in turn. The TLP algorithm has three phases, which we discuss in turn.
5.4.1. Phase 1: Scheduling a loss probe 6.5.1. Phase 1: Scheduling a loss probe
Step 1: Check conditions for scheduling a PTO. Step 1: Check conditions for scheduling a PTO.
A sender should check to see if it should schedule a PTO in two A sender should check to see if it should schedule a PTO in the
situations: following situations:
1. After transmitting new data 1. After transmitting new data that was not itself a TLP probe
2. Upon receiving an ACK that cumulatively acknowledges data. 2. Upon receiving an ACK that cumulatively acknowledges data
3. Upon receiving a SACK that selectively acknowledges data that was
last sent before the segment with SEG.SEQ=SND.UNA was last
(re)transmitted
A sender should schedule a PTO only if all of the following A sender should schedule a PTO only if all of the following
conditions are met: conditions are met:
1. The connection supports SACK [RFC2018] 1. The connection supports SACK [RFC2018]
2. The connection is not in loss recovery 2. The connection has no SACKed sequences in the SACK scoreboard
3. The connection is either limited by congestion window (the data
in flight matches or exceeds the cwnd) or application-limited
(there is no unsent data that the receiver window allows to be
sent).
4. The most recently transmitted data was not itself a TLP probe 3. The connection is not in loss recovery
(i.e. a sender MUST NOT send consecutive or back-to-back TLP
probes). If a PTO can be scheduled according to these conditions, the sender
should schedule a PTO. If there was a previously scheduled PTO or
RTO pending, then that pending PTO or RTO should first be cancelled,
and then the new PTO should be scheduled.
If a PTO cannot be scheduled according to these conditions, then the If a PTO cannot be scheduled according to these conditions, then the
sender MUST arm the RTO timer if there is unacknowledged data in sender MUST arm the RTO timer if there is unacknowledged data in
flight. flight.
Step 2: Select the duration of the PTO. Step 2: Select the duration of the PTO.
A sender SHOULD use the following logic to select the duration of a A sender SHOULD use the following logic to select the duration of a
PTO: PTO:
skipping to change at page 13, line 43 skipping to change at page 16, line 35
PTO = TCP_RTO_expire() - Now() PTO = TCP_RTO_expire() - Now()
Aiming for a PTO value of 2*SRTT allows a sender to wait long enough Aiming for a PTO value of 2*SRTT allows a sender to wait long enough
to know that an ACK is overdue. Under normal circumstances, i.e. no to know that an ACK is overdue. Under normal circumstances, i.e. no
losses, an ACK typically arrives in one SRTT. But choosing PTO to be losses, an ACK typically arrives in one SRTT. But choosing PTO to be
exactly an SRTT is likely to generate spurious probes given that exactly an SRTT is likely to generate spurious probes given that
network delay variance and even end-system timings can easily push an network delay variance and even end-system timings can easily push an
ACK to be above an SRTT. We chose PTO to be the next integral ACK to be above an SRTT. We chose PTO to be the next integral
multiple of SRTT. multiple of SRTT.
Similarly, current end-system processing latencies and timer Similarly, network delay variations and end-system processing
granularities can easily delay ACKs, so senders SHOULD add at least latencies and timer granularities can easily delay ACKs beyond
2ms to a computed PTO value (and MAY add more if the sending host OS 2*SRTT, so senders SHOULD add at least 2ms to a computed PTO value
timer granularity is more coarse than 1ms). (and MAY add more if the sending host OS timer granularity is more
coarse than 1ms).
WCDelAckT stands for worst case delayed ACK timer. When FlightSize WCDelAckT stands for worst case delayed ACK timer. When FlightSize
is 1, PTO is inflated by WCDelAckT time to compensate for a potential is 1, PTO is inflated by WCDelAckT time to compensate for a potential
long delayed ACK timer at the receiver. The RECOMMENDED value for long delayed ACK timer at the receiver. The RECOMMENDED value for
WCDelAckT is 200ms. WCDelAckT is 200ms.
Finally, if the time at which an RTO would fire (here denoted Finally, if the time at which an RTO would fire (here denoted
"TCP_RTO_expire") is sooner than the computed time for the PTO, then "TCP_RTO_expire") is sooner than the computed time for the PTO, then
a probe is scheduled to be sent at that earlier time.. a probe is scheduled to be sent at that earlier time.
5.4.2. Phase 2: Sending a loss probe 6.5.2. Phase 2: Sending a loss probe
When the PTO fires, transmit a probe data segment: When the PTO fires, transmit a probe data segment:
TLP_send_probe(): TLP_send_probe():
If a previously unsent segment exists AND If an unsent segment exists AND
the receive window allows new data to be sent: the receive window allows new data to be sent:
Transmit that new segment Transmit the lowest-sequence unsent segment of up to SMSS
FlightSize += SMSS Increment FlightSize by the size of the newly-sent segment
Else: Else:
Retransmit the last segment Retransmit the highest-sequence segment sent so far
The cwnd remains unchanged The cwnd remains unchanged
5.4.3. Phase 3: ACK processing When the loss probe is a retransmission, the sender uses the highest-
sequence segment sent so far. This is in order to deal with the
retransmission ambiguity problem in TCP. Suppose a sender sends N
segments, and then retransmits the last segment (segment N) as a loss
probe, and then the sender receives a SACK for segment N. As long as
the sender waits for any required RACK reordering settling timer to
then expire, it doesn't matter if that SACK was for the original
transmission of segment N or the TLP retransmission; in either case
the arrival of the SACK for segment N provides evidence that the N-1
segments preceding segment N were likely lost. In the case where
there is only one original outstanding segment of data (N=1), the
same logic (trivially) applies: an ACK for a single outstanding
segment tells the sender the N-1=0 segments preceding that segment
were lost. Furthermore, whether there are N>1 or N=1 outstanding
segments, there is a question about whether the original last segment
or its TLP retransmission were lost; the sender estimates this using
TLP recovery detection (see below).
On each incoming ACK, the sender should cancel any existing loss Note that after transmitting a TLP, the sender MUST arm an RTO timer,
probe timer. The sender should then reschedule the loss probe timer and not the PTO timer. This ensures that the sender does not send
if the conditions in Step 1 of Phase 1 allow. repeated, back-to-back TLP probes.
5.5. TLP recovery detection 6.5.3. Phase 3: ACK processing
On each incoming ACK, the sender should check the conditions in Step
1 of Phase 1 to see if it should schedule (or reschedule) the loss
probe timer.
6.6. TLP recovery detection
If the only loss in an outstanding window of data was the last If the only loss in an outstanding window of data was the last
segment, then a TLP loss probe retransmission of that data segment segment, then a TLP loss probe retransmission of that data segment
might repair the loss. TLP recovery detection examines ACKs to might repair the loss. TLP recovery detection examines ACKs to
detect when the probe might have repaired a loss, and thus allows detect when the probe might have repaired a loss, and thus allows
congestion control to properly reduce the congestion window (cwnd) congestion control to properly reduce the congestion window (cwnd)
[RFC5681]. [RFC5681].
Consider a TLP retransmission episode where a sender retransmits a Consider a TLP retransmission episode where a sender retransmits a
tail packet in a flight. The TLP retransmission episode ends when tail packet in a flight. The TLP retransmission episode ends when
skipping to change at page 15, line 15 skipping to change at page 18, line 33
SACK support [RFC2018]. SACK support [RFC2018].
Definitions of variables Definitions of variables
TLPRxtOut: a boolean indicating whether there is an unacknowledged TLPRxtOut: a boolean indicating whether there is an unacknowledged
TLP retransmission. TLP retransmission.
TLPHighRxt: the value of SND.NXT at the time of sending a TLP TLPHighRxt: the value of SND.NXT at the time of sending a TLP
retransmission. retransmission.
5.5.1. Initializing and resetting state 6.6.1. Initializing and resetting state
When a connection is created, or suffers a retransmission timeout, or When a connection is created, or suffers a retransmission timeout, or
enters fast recovery, it executes the following: enters fast recovery, it executes the following:
TLPRxtOut = false TLPRxtOut = false
5.5.2. Recording loss probe states 6.6.2. Recording loss probe states
Senders must only send a TLP loss probe retransmission if TLPRxtOut Senders MUST only send a TLP loss probe retransmission if TLPRxtOut
is false. This ensures that at any given time a connection has at is false. This ensures that at any given time a connection has at
most one outstanding TLP retransmission. This allows the sender to most one outstanding TLP retransmission. This allows the sender to
use the algorithm described in this section to estimate whether any use the algorithm described in this section to estimate whether any
data segments were lost. data segments were lost.
Note that this condition only restricts TLP loss probes that are Note that this condition only restricts TLP loss probes that are
retransmissions. There may be an arbitrary number of outstanding retransmissions. There may be an arbitrary number of outstanding
unacknowledged TLP loss probes that consist of new, previously-unsent unacknowledged TLP loss probes that consist of new, previously-unsent
data, since the retransmission timeout and fast recovery algorithms data, since the retransmission timeout and fast recovery algorithms
are sufficient to detect losses of such probe segments. are sufficient to detect losses of such probe segments.
skipping to change at page 17, line 5 skipping to change at page 20, line 21
a congestion control response equivalent to fast recovery. a congestion control response equivalent to fast recovery.
More precisely, on each ACK the sender executes the following: More precisely, on each ACK the sender executes the following:
if (TLPRxtOut and SEG.ACK >= TLPHighRxt) { if (TLPRxtOut and SEG.ACK >= TLPHighRxt) {
TLPRxtOut = false TLPRxtOut = false
EnterRecovery() EnterRecovery()
ExitRecovery() ExitRecovery()
} }
6. RACK and TLP discussions 7. RACK and TLP discussions
6.1. Advantages 7.1. Advantages
The biggest advantage of RACK is that every data packet, whether it The biggest advantage of RACK is that every data packet, whether it
is an original data transmission or a retransmission, can be used to is an original data transmission or a retransmission, can be used to
detect losses of the packets sent chronologically prior to it. detect losses of the packets sent chronologically prior to it.
Example: TAIL DROP. Consider a sender that transmits a window of Example: TAIL DROP. Consider a sender that transmits a window of
three data packets (P1, P2, P3), and P1 and P3 are lost. Suppose the three data packets (P1, P2, P3), and P1 and P3 are lost. Suppose the
transmission of each packet is at least RACK.reo_wnd (1 millisecond transmission of each packet is at least RACK.reo_wnd (1 millisecond
by default) after the transmission of the previous packet. RACK will by default) after the transmission of the previous packet. RACK will
mark P1 as lost when the SACK of P2 is received, and this will mark P1 as lost when the SACK of P2 is received, and this will
skipping to change at page 17, line 45 skipping to change at page 21, line 13
[RFC6675], nor the Forward Acknowledgment [FACK] algorithm can detect [RFC6675], nor the Forward Acknowledgment [FACK] algorithm can detect
such losses. And such a lost retransmission is very common when TCP such losses. And such a lost retransmission is very common when TCP
is being rate-limited, particularly by token bucket policers with is being rate-limited, particularly by token bucket policers with
large bucket depth and low rate limit. Retransmissions are often large bucket depth and low rate limit. Retransmissions are often
lost repeatedly because standard congestion control requires multiple lost repeatedly because standard congestion control requires multiple
round trips to reduce the rate below the policed rate. round trips to reduce the rate below the policed rate.
Example: SMALL DEGREE OF REORDERING. Consider a common reordering Example: SMALL DEGREE OF REORDERING. Consider a common reordering
event: a window of packets are sent as (P1, P2, P3). P1 and P2 carry event: a window of packets are sent as (P1, P2, P3). P1 and P2 carry
a full payload of MSS octets, but P3 has only a 1-octet payload. a full payload of MSS octets, but P3 has only a 1-octet payload.
Suppose the sender has detected reordering previously (e.g., by Suppose the sender has detected reordering previously and thus
implementing the algorithm in [REORDER-DETECT]) and thus RACK.reo_wnd RACK.reo_wnd is min_RTT/4. Now P3 is reordered and delivered first,
is min_RTT/4. Now P3 is reordered and delivered first, before P1 and before P1 and P2. As long as P1 and P2 are delivered within
P2. As long as P1 and P2 are delivered within min_RTT/4, RACK will min_RTT/4, RACK will not consider P1 and P2 lost. But if P1 and P2
not consider P1 and P2 lost. But if P1 and P2 are delivered outside are delivered outside the reordering window, then RACK will still
the reordering window, then RACK will still falsely mark P1 and P2 falsely mark P1 and P2 lost. We discuss how to reduce false
lost. We discuss how to reduce false positives in the end of this positives in the end of this section.
section.
The examples above show that RACK is particularly useful when the The examples above show that RACK is particularly useful when the
sender is limited by the application, which is common for sender is limited by the application, which is common for
interactive, request/response traffic. Similarly, RACK still works interactive, request/response traffic. Similarly, RACK still works
when the sender is limited by the receive window, which is common for when the sender is limited by the receive window, which is common for
applications that use the receive window to throttle the sender. applications that use the receive window to throttle the sender.
For some implementations (e.g., Linux), RACK works quite efficiently For some implementations (e.g., Linux), RACK works quite efficiently
with TCP Segmentation Offload (TSO). RACK always marks the entire with TCP Segmentation Offload (TSO). RACK always marks the entire
TSO blob lost because the packets in the same TSO blob have the same TSO blob lost because the packets in the same TSO blob have the same
transmission timestamp. By contrast, the counting based algorithms transmission timestamp. By contrast, the algorithms based on
(e.g., [RFC3517][RFC5681]) may mark only a subset of packets in the sequence counting (e.g., [RFC6675][RFC5681]) may mark only a subset
TSO blob lost, forcing the stack to perform expensive fragmentation of packets in the TSO blob lost, forcing the stack to perform
of the TSO blob, or to selectively tag individual packets lost in the expensive fragmentation of the TSO blob, or to selectively tag
scoreboard. individual packets lost in the scoreboard.
6.2. Disadvantages 7.2. Disadvantages
RACK requires the sender to record the transmission time of each RACK requires the sender to record the transmission time of each
packet sent at a clock granularity of one millisecond or finer. TCP packet sent at a clock granularity of one millisecond or finer. TCP
implementations that record this already for RTT estimation do not implementations that record this already for RTT estimation do not
require any new per-packet state. But implementations that are not require any new per-packet state. But implementations that are not
yet recording packet transmission times will need to add per-packet yet recording packet transmission times will need to add per-packet
internal state (commonly either 4 or 8 octets per packet or TSO blob) internal state (commonly either 4 or 8 octets per packet or TSO blob)
to track transmission times. In contrast, the conventional [RFC6675] to track transmission times. In contrast, the conventional [RFC6675]
loss detection approach does not require any per-packet state beyond loss detection approach does not require any per-packet state beyond
the SACK scoreboard. This is particularly useful on ultra-low RTT the SACK scoreboard. This is particularly useful on ultra-low RTT
skipping to change at page 18, line 45 skipping to change at page 22, line 13
[RFC6675][RFC5681] by resetting the reordering window to zero when [RFC6675][RFC5681] by resetting the reordering window to zero when
the threshold is met. Note that this approach differs slightly from the threshold is met. Note that this approach differs slightly from
[RFC6675] which considers a packet lost when at least #DupThresh [RFC6675] which considers a packet lost when at least #DupThresh
higher-sequenc packets are SACKed. RACK's approach considers a higher-sequenc packets are SACKed. RACK's approach considers a
packet lost when at least one higher sequence packet is SACKed and packet lost when at least one higher sequence packet is SACKed and
the total number of SACKed packets is at least DupThresh. For the total number of SACKed packets is at least DupThresh. For
example, suppose a connection sends 10 packets, and packets 3, 5, 7 example, suppose a connection sends 10 packets, and packets 3, 5, 7
are SACKed. [RFC6675] considers packets 1 and 2 lost. RACK are SACKed. [RFC6675] considers packets 1 and 2 lost. RACK
considers packets 1, 2, 4, 6 lost. considers packets 1, 2, 4, 6 lost.
6.3. Adjusting the reordering window 7.3. Adjusting the reordering window
When the sender detects packet reordering, RACK uses a reordering When the sender detects packet reordering, RACK uses a reordering
window of min_rtt / 4. It uses the minimum RTT to accommodate window of min_rtt / 4. It uses the minimum RTT to accommodate
reordering introduced by packets traversing slightly different paths reordering introduced by packets traversing slightly different paths
(e.g., router-based parallelism schemes) or out-of-order deliveries (e.g., router-based parallelism schemes) or out-of-order deliveries
in the lower link layer (e.g., wireless links using link-layer in the lower link layer (e.g., wireless links using link-layer
retransmission). RACK uses a quarter of minimum RTT because Linux retransmission). RACK uses a quarter of minimum RTT because Linux
TCP used the same factor in its implementation to delay Early TCP used the same factor in its implementation to delay Early
Retransmit [RFC5827] to reduce spurious loss detections in the Retransmit [RFC5827] to reduce spurious loss detections in the
presence of reordering, and experience shows that this seems to work presence of reordering, and experience shows that this seems to work
reasonably well. We have evaluated using the smoothed RTT (SRTT from reasonably well. We have evaluated using the smoothed RTT (SRTT from
[RFC6298] RTT estimation) or the most recently measured RTT [RFC6298] RTT estimation) or the most recently measured RTT
(RACK.RTT) using an experiment similar to that in the Performance (RACK.rtt) using an experiment similar to that in the Performance
Evaluation section. They do not make any significant difference in Evaluation section. They do not make any significant difference in
terms of total recovery latency. terms of total recovery latency.
6.4. Relationships with other loss recovery algorithms 7.4. Relationships with other loss recovery algorithms
The primary motivation of RACK is to ultimately provide a simple and The primary motivation of RACK is to ultimately provide a simple and
general replacement for some of the standard loss recovery algorithms general replacement for some of the standard loss recovery algorithms
[RFC5681][RFC6675][RFC5827][RFC4653], as well as some nonstandard [RFC5681][RFC6675][RFC5827][RFC4653], as well as some nonstandard
ones [FACK][THIN-STREAM]. While RACK can be a supplemental loss ones [FACK][THIN-STREAM]. While RACK can be a supplemental loss
detection mechanism on top of these algorithms, this is not detection mechanism on top of these algorithms, this is not
necessary, because RACK implicitly subsumes most of them. necessary, because RACK implicitly subsumes most of them.
[RFC5827][RFC4653][THIN-STREAM] dynamically adjusts the duplicate ACK [RFC5827][RFC4653][THIN-STREAM] dynamically adjusts the duplicate ACK
threshold based on the current or previous flight sizes. RACK takes threshold based on the current or previous flight sizes. RACK takes
skipping to change at page 20, line 7 skipping to change at page 23, line 24
Furthermore, RACK naturally works well with Tail Loss Probe [TLP] Furthermore, RACK naturally works well with Tail Loss Probe [TLP]
because a tail loss probe solicits either an ACK or SACK, which can because a tail loss probe solicits either an ACK or SACK, which can
be used by RACK to detect more losses. RACK can be used to relax be used by RACK to detect more losses. RACK can be used to relax
TLP's requirement for using FACK and retransmitting the the highest- TLP's requirement for using FACK and retransmitting the the highest-
sequenced packet, because RACK is agnostic to packet sequence sequenced packet, because RACK is agnostic to packet sequence
numbers, and uses transmission time instead. Thus TLP could be numbers, and uses transmission time instead. Thus TLP could be
modified to retransmit the first unacknowledged packet, which could modified to retransmit the first unacknowledged packet, which could
improve application latency. improve application latency.
6.5. Interaction with congestion control 7.5. Interaction with congestion control
RACK intentionally decouples loss detection from congestion control. RACK intentionally decouples loss detection from congestion control.
RACK only detects losses; it does not modify the congestion control RACK only detects losses; it does not modify the congestion control
algorithm [RFC5681][RFC6937]. However, RACK may detect losses algorithm [RFC5681][RFC6937]. However, RACK may detect losses
earlier or later than the conventional duplicate ACK threshold earlier or later than the conventional duplicate ACK threshold
approach does. A packet marked lost by RACK SHOULD NOT be approach does. A packet marked lost by RACK SHOULD NOT be
retransmitted until congestion control deems this appropriate. retransmitted until congestion control deems this appropriate.
Specifically, Proportional Rate Reduction [RFC6937] SHOULD be used Specifically, Proportional Rate Reduction [RFC6937] SHOULD be used
when using RACK. when using RACK.
skipping to change at page 21, line 7 skipping to change at page 24, line 23
In both cases, the sender after the recovery would be in congestion In both cases, the sender after the recovery would be in congestion
avoidance. The difference in recovery latency (RTO + 4*RTT vs 6*RTT) avoidance. The difference in recovery latency (RTO + 4*RTT vs 6*RTT)
can be significant if the RTT is much smaller than the minimum RTO (1 can be significant if the RTT is much smaller than the minimum RTO (1
second in RFC6298) or if the RTT is large. The former case is common second in RFC6298) or if the RTT is large. The former case is common
in local area networks, data-center networks, or content distribution in local area networks, data-center networks, or content distribution
networks with deep deployments. The latter case is more common in networks with deep deployments. The latter case is more common in
developing regions with highly congested and/or high-latency developing regions with highly congested and/or high-latency
networks. The ending congestion window after recovery also impacts networks. The ending congestion window after recovery also impacts
subsequent data transfer. subsequent data transfer.
6.6. TLP recovery detection with delayed ACKs 7.6. TLP recovery detection with delayed ACKs
Delayed ACKs complicate the detection of repairs done by TLP, since Delayed ACKs complicate the detection of repairs done by TLP, since
with a delayed ACK the sender receives one fewer ACK than would with a delayed ACK the sender receives one fewer ACK than would
normally be expected. To mitigate this complication, before sending normally be expected. To mitigate this complication, before sending
a TLP loss probe retransmission, the sender should attempt to wait a TLP loss probe retransmission, the sender should attempt to wait
long enough that the receiver has sent any delayed ACKs that it is long enough that the receiver has sent any delayed ACKs that it is
withholding. The sender algorithm described above features such a withholding. The sender algorithm described above features such a
delay, in the form of WCDelAckT. Furthermore, if the receiver delay, in the form of WCDelAckT. Furthermore, if the receiver
supports duplicate selective acknowledgments (D-SACKs) [RFC2883] then supports duplicate selective acknowledgments (D-SACKs) [RFC2883] then
in the case of a delayed ACK the sender's TLP recovery detection in the case of a delayed ACK the sender's TLP recovery detection
algorithm (see above) can use the D-SACK information to infer that algorithm (see above) can use the D-SACK information to infer that
the original and TLP retransmission both arrived at the receiver. the original and TLP retransmission both arrived at the receiver.
If there is ACK loss or a delayed ACK without a D-SACK, then this If there is ACK loss or a delayed ACK without a D-SACK, then this
algorithm is conservative, because the sender will reduce cwnd when algorithm is conservative, because the sender will reduce cwnd when
in fact there was no packet loss. In practice this is acceptable, in fact there was no packet loss. In practice this is acceptable,
and potentially even desirable: if there is reverse path congestion and potentially even desirable: if there is reverse path congestion
then reducing cwnd can be prudent. then reducing cwnd can be prudent.
6.7. RACK for other transport protocols 7.7. RACK for other transport protocols
RACK can be implemented in other transport protocols. The algorithm RACK can be implemented in other transport protocols. The algorithm
can be simplified by skipping step 3 if the protocol can support a can be simplified by skipping step 3 if the protocol can support a
unique transmission or packet identifier (e.g. TCP echo options). unique transmission or packet identifier (e.g. TCP timestamp options
For example, the QUIC protocol implements RACK [QUIC-LR]. [RFC7323]). For example, the QUIC protocol implements RACK [QUIC-
LR].
7. Experiments and Performance Evaluations 8. Experiments and Performance Evaluations
RACK and TLP have been deployed at Google, for both connections to RACK and TLP have been deployed at Google, for both connections to
users in the Internet and internally. We conducted a performance users in the Internet and internally. We conducted a performance
evaluation experiment for RACK and TLP on a small set of Google Web evaluation experiment for RACK and TLP on a small set of Google Web
servers in Western Europe that serve mostly European and some African servers in Western Europe that serve mostly European and some African
countries. The experiment lasted three days in March 2017. The countries. The experiment lasted three days in March 2017. The
servers were divided evenly into four groups of roughly 5.3 million servers were divided evenly into four groups of roughly 5.3 million
flows each: flows each:
Group 1 (control): RACK off, TLP off, RFC 3517 on Group 1 (control): RACK off, TLP off, RFC 6675 on
Group 2: RACK on, TLP off, RFC 3517 on Group 2: RACK on, TLP off, RFC 6675 on
Group 3: RACK on, TLP on, RFC 3517 on Group 3: RACK on, TLP on, RFC 6675 on
Group 4: RACK on, TLP on, RFC 6675 off
Group 4: RACK on, TLP on, RFC 3517 off
All groups used Linux with CUBIC congestion control, an initial All groups used Linux with CUBIC congestion control, an initial
congestion window of 10 packets, and the fq/pacing qdisc. In terms congestion window of 10 packets, and the fq/pacing qdisc. In terms
of specific recovery features, all groups enabled RFC5682 (F-RTO) but of specific recovery features, all groups enabled RFC5682 (F-RTO) but
disabled FACK because it is not an IETF RFC. FACK was excluded disabled FACK because it is not an IETF RFC. FACK was excluded
because the goal of this setup is to compare RACK and TLP to RFC- because the goal of this setup is to compare RACK and TLP to RFC-
based loss recoveries. Since TLP depends on either FACK or RACK, we based loss recoveries. Since TLP depends on either FACK or RACK, we
could not run another group that enables TLP only (with both RACK and could not run another group that enables TLP only (with both RACK and
FACK disabled). Group 4 is to test whether RACK plus TLP can FACK disabled). Group 4 is to test whether RACK plus TLP can
completely replace the DupThresh-based [RFC3517]. completely replace the DupThresh-based [RFC6675].
The servers sit behind a load balancer that distributes the The servers sit behind a load balancer that distributes the
connections evenly across the four groups. connections evenly across the four groups.
Each group handles a similar number of connections and sends and Each group handles a similar number of connections and sends and
receives similar amounts of data. We compare total time spent in receives similar amounts of data. We compare total time spent in
loss recovery across groups. The recovery time is measured from when loss recovery across groups. The recovery time is measured from when
the recovery and retransmission starts, until the remote host has the recovery and retransmission starts, until the remote host has
acknowledged the highest sequence (SND.NXT) at the time the recovery acknowledged the highest sequence (SND.NXT) at the time the recovery
started. Therefore the recovery includes both fast recoveries and started. Therefore the recovery includes both fast recoveries and
timeout recoveries. timeout recoveries.
Our data shows that Group 2 recovery latency is only 0.3% lower than Our data shows that Group 2 recovery latency is only 0.3% lower than
the Group 1 recovery latency. But Group 3 recovery latency is 25% the Group 1 recovery latency. But Group 3 recovery latency is 25%
lower than Group 1 due to a 40% reduction in RTO-triggered lower than Group 1 due to a 40% reduction in RTO-triggered
recoveries! Therefore it is important to implement both TLP and RACK recoveries! Therefore it is important to implement both TLP and RACK
for performance. Group 4's total recovery latency is 0.02% lower for performance. Group 4's total recovery latency is 0.02% lower
than Group 3's, indicating that RACK plus TLP can successfully than Group 3's, indicating that RACK plus TLP can successfully
replace RFC3517 as a standalone recovery mechanism. replace RFC6675 as a standalone recovery mechanism.
We want to emphasize that the current experiment is limited in terms We want to emphasize that the current experiment is limited in terms
of network coverage. The connectivity in Western Europe is fairly of network coverage. The connectivity in Western Europe is fairly
good, therefore loss recovery is not a major performance bottleneck. good, therefore loss recovery is not a major performance bottleneck.
We plan to expand our experiments to regions with worse connectivity, We plan to expand our experiments to regions with worse connectivity,
in particular on networks with strong traffic policing. in particular on networks with strong traffic policing.
8. Security Considerations 9. Security Considerations
RACK does not change the risk profile for TCP. RACK does not change the risk profile for TCP.
An interesting scenario is ACK-splitting attacks [SCWA99]: for an An interesting scenario is ACK-splitting attacks [SCWA99]: for an
MSS-size packet sent, the receiver or the attacker might send MSS MSS-size packet sent, the receiver or the attacker might send MSS
ACKs that SACK or acknowledge one additional byte per ACK. This ACKs that SACK or acknowledge one additional byte per ACK. This
would not fool RACK. RACK.xmit_ts would not advance because all the would not fool RACK. RACK.xmit_ts would not advance because all the
sequences of the packet are transmitted at the same time (carry the sequences of the packet are transmitted at the same time (carry the
same transmission timestamp). In other words, SACKing only one byte same transmission timestamp). In other words, SACKing only one byte
of a packet or SACKing the packet in entirety have the same effect on of a packet or SACKing the packet in entirety have the same effect on
RACK. RACK.
9. IANA Considerations 10. IANA Considerations
This document makes no request of IANA. This document makes no request of IANA.
Note to RFC Editor: this section may be removed on publication as an Note to RFC Editor: this section may be removed on publication as an
RFC. RFC.
10. Acknowledgments 11. Acknowledgments
The authors thank Matt Mathis for his insights in FACK and Michael The authors thank Matt Mathis for his insights in FACK and Michael
Welzl for his per-packet timer idea that inspired this work. Eric Welzl for his per-packet timer idea that inspired this work. Eric
Dumazet, Randy Stewart, Van Jacobson, Ian Swett, Rick Jones, Jana Dumazet, Randy Stewart, Van Jacobson, Ian Swett, Rick Jones, Jana
Iyengar, and Hiren Panchasara contributed to the draft and the Iyengar, Hiren Panchasara, Praveen Balasubramanian, Yoshifumi
Nishida, and Bob Briscoe contributed to the draft and the
implementations in Linux, FreeBSD and QUIC. implementations in Linux, FreeBSD and QUIC.
11. References 12. References
11.1. Normative References 12.1. Normative References
[RFC2018] Mathis, M. and J. Mahdavi, "TCP Selective Acknowledgment [RFC2018] Mathis, M. and J. Mahdavi, "TCP Selective Acknowledgment
Options", RFC 2018, October 1996. Options", RFC 2018, October 1996.
[RFC2119] Bradner, S., "Key words for use in RFCs to Indicate [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate
Requirement Levels", RFC 2119, March 1997. Requirement Levels", RFC 2119, March 1997.
[RFC2883] Floyd, S., Mahdavi, J., Mathis, M., and M. Podolsky, "An [RFC2883] Floyd, S., Mahdavi, J., Mathis, M., and M. Podolsky, "An
Extension to the Selective Acknowledgement (SACK) Option Extension to the Selective Acknowledgement (SACK) Option
for TCP", RFC 2883, July 2000. for TCP", RFC 2883, July 2000.
skipping to change at page 24, line 20 skipping to change at page 27, line 40
[RFC6937] Mathis, M., Dukkipati, N., and Y. Cheng, "Proportional [RFC6937] Mathis, M., Dukkipati, N., and Y. Cheng, "Proportional
Rate Reduction for TCP", May 2013. Rate Reduction for TCP", May 2013.
[RFC7323] Borman, D., Braden, B., Jacobson, V., and R. [RFC7323] Borman, D., Braden, B., Jacobson, V., and R.
Scheffenegger, "TCP Extensions for High Performance", Scheffenegger, "TCP Extensions for High Performance",
September 2014. September 2014.
[RFC793] Postel, J., "Transmission Control Protocol", September [RFC793] Postel, J., "Transmission Control Protocol", September
1981. 1981.
11.2. Informative References 12.2. Informative References
[FACK] Mathis, M. and M. Jamshid, "Forward acknowledgement: [FACK] Mathis, M. and M. Jamshid, "Forward acknowledgement:
refining TCP congestion control", ACM SIGCOMM Computer refining TCP congestion control", ACM SIGCOMM Computer
Communication Review, Volume 26, Issue 4, Oct. 1996. , Communication Review, Volume 26, Issue 4, Oct. 1996. ,
1996. 1996.
[POLICER16] [POLICER16]
Flach, T., Papageorge, P., Terzis, A., Pedrosa, L., Cheng, Flach, T., Papageorge, P., Terzis, A., Pedrosa, L., Cheng,
Y., Karim, T., Katz-Bassett, E., and R. Govindan, "An Y., Karim, T., Katz-Bassett, E., and R. Govindan, "An
Analysis of Traffic Policing in the Web", ACM SIGCOMM , Analysis of Traffic Policing in the Web", ACM SIGCOMM ,
2016. 2016.
[QUIC-LR] Iyengar, J. and I. Swett, "QUIC Loss Recovery And [QUIC-LR] Iyengar, J. and I. Swett, "QUIC Loss Recovery And
Congestion Control", draft-tsvwg-quic-loss-recovery-01 Congestion Control", draft-tsvwg-quic-loss-recovery-01
(work in progress), June 2016. (work in progress), June 2016.
[REORDER-DETECT]
Zimmermann, A., Schulte, L., Wolff, C., and A. Hannemann,
"Detection and Quantification of Packet Reordering with
TCP", draft-zimmermann-tcpm-reordering-detection-02 (work
in progress), November 2014.
[RFC7765] Hurtig, P., Brunstrom, A., Petlund, A., and M. Welzl, "TCP [RFC7765] Hurtig, P., Brunstrom, A., Petlund, A., and M. Welzl, "TCP
and SCTP RTO Restart", February 2016. and SCTP RTO Restart", February 2016.
[SCWA99] Savage, S., Cardwell, N., Wetherall, D., and T. Anderson, [SCWA99] Savage, S., Cardwell, N., Wetherall, D., and T. Anderson,
"TCP Congestion Control With a Misbehaving Receiver", ACM "TCP Congestion Control With a Misbehaving Receiver", ACM
Computer Communication Review, 29(5) , 1999. Computer Communication Review, 29(5) , 1999.
[THIN-STREAM] [THIN-STREAM]
Petlund, A., Evensen, K., Griwodz, C., and P. Halvorsen, Petlund, A., Evensen, K., Griwodz, C., and P. Halvorsen,
"TCP enhancements for interactive thin-stream "TCP enhancements for interactive thin-stream
 End of changes. 100 change blocks. 
206 lines changed or deleted 354 lines changed or added

This html diff was produced by rfcdiff 1.47. The latest version is available from http://tools.ietf.org/tools/rfcdiff/