wdiff draft-ietf-tsvwg-byte-pkt-congest-01.txt draft-ietf-tsvwg-byte-pkt-congest-02.txt

Transport Area Working Group B. Briscoe
Internet-Draft BT
Updates: 2309 (if approved) October 23, 2009 J. Manner
Intended status: Informational Aalto University
Expires: April 26, January 13, 2011 July 12, 2010

Byte and Packet Congestion Notification
draft-ietf-tsvwg-byte-pkt-congest-01
draft-ietf-tsvwg-byte-pkt-congest-02

Abstract

This memo concerns dropping or marking packets using active queue
management (AQM) such as random early detection (RED) or pre-
congestion notification (PCN). We give two strong recommendations:
(1) packet size should not be taken into account when transports read
congestion indications, not when network equipment writes them, and
(2) byte-mode packet drop variant of AQM algorithms, such as RED,
should not be used to drop fewer small packets.

Status of this This Memo

This Internet-Draft is submitted to IETF in full conformance with the
provisions of BCP 78 and BCP 79.

Internet-Drafts are working documents of the Internet Engineering
Task Force (IETF), its areas, and its working groups. (IETF). Note that other groups may also distribute
working documents as Internet-Drafts. The list of current Internet-
Drafts.
Drafts is at http://datatracker.ietf.org/drafts/current/.

Internet-Drafts are draft documents valid for a maximum of six months
and may be updated, replaced, or obsoleted by other documents at any
time. It is inappropriate to use Internet-Drafts as reference
material or to cite them other than as "work in progress."

The list of current Internet-Drafts can be accessed at
http://www.ietf.org/ietf/1id-abstracts.txt.

The list of Internet-Draft Shadow Directories can be accessed at
http://www.ietf.org/shadow.html.

This Internet-Draft will expire on April 26, 2010. January 13, 2011.

This document is subject to BCP 78 and the IETF Trust's Legal
Provisions Relating to IETF Documents
(http://trustee.ietf.org/license-info) in effect on the date of
publication of this document (http://trustee.ietf.org/license-info). document. Please review these documents
carefully, as they describe your rights and restrictions with respect
to this document.

Abstract

This memo concerns dropping or marking packets using active queue
management (AQM) such Code Components extracted from this document must
include Simplified BSD License text as random early detection (RED) or pre-
congestion notification (PCN). The primary conclusion is that packet
size should be taken into account when transports read congestion
indications, not when network equipment writes them. Reducing drop described in Section 4.e of small packets has some tempting advantages: i) it drops less
control packets, which tend to be small
the Trust Legal Provisions and ii) it makes TCP's bit-
rate less dependent on packet size. However, there are ways of
addressing these issues at the transport layer, rather than reverse
engineering network forwarding to fix specific transport problems.
Network layer algorithms like provided without warranty as
described in the byte-mode packet drop variant of
RED should not be used to drop fewer small packets, because that
creates a perverse incentive for transports to use tiny segments,
consequently also opening up a DoS vulnerability.

Table Simplified BSD License.

Table of Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 6 4
1.1. Requirements Notation Terminology and Scoping . . . . . . . . . . . . . . . . . 6
1.2. Why now? . . 9 . . . . . . . . . . . . . . . . . . . . . . . 7
2. Motivating Arguments . . . . . . . . . . . . . . . . . . . . . 9 8
2.1. Scaling Congestion Control with Packet Size . . . . . . . 9 8
2.2. Avoiding Perverse Incentives to (ab)use Smaller Packets . 10
2.3. Small != Control . . . . . . . . . . . . . . . . . . . . . 12 11
2.4. Implementation Efficiency . . . . . . . . . . . . . . . . 12 11
3. Working Definition The State of Congestion Notification . . . the Art . . . . . 12
4. Congestion Measurement . . . . . . . . . . . . . . . . 11
3.1. Congestion Measurement: Status . . . . 13
4.1. Congestion Measurement by Queue Length . . . . . . . . . . 13
4.1.1. 12
3.1.1. Fixed Size Packet Buffers . . . . . . . . . . . . . . 13
4.2.
3.1.2. Congestion Measurement without a Queue . . . . . . . . 14
3.2. Congestion Coding: Status . . . 14
5. Idealised Wire Protocol Coding . . . . . . . . . . . . . 14
3.2.1. Network Bias when Encoding . . . 15
6. The State of the Art . . . . . . . . . . . 14
3.2.2. Transport Bias when Decoding . . . . . . . . . . 17
6.1. Congestion Measurement: Status . . . 16
3.2.3. Making Transports Robust against Control Packet
Losses . . . . . . . . . . . 17
6.2. Congestion Coding: Status . . . . . . . . . . . . . 17
3.2.4. Congestion Coding: Summary of Status . . . 18
6.2.1. Network Bias when Encoding . . . . . . 18
4. Outstanding Issues and Next Steps . . . . . . . . 18
6.2.2. Transport Bias when Decoding . . . . . . 20
4.1. Bit-congestible World . . . . . . . 20
6.2.3. Making Transports Robust against Control Packet
Losses . . . . . . . . . . . 20
4.2. Bit- & Packet-congestible World . . . . . . . . . . . . . 21
6.2.4. Congestion Coding: Summary of Status
5. Recommendation and Conclusions . . . . . . . . . 22
7. Outstanding Issues and Next Steps . . . . . . . 22
5.1. Recommendation on Queue Measurement . . . . . . . . 24
7.1. Bit-congestible World . . . 22
5.2. Recommendation on Notifying Congestion . . . . . . . . . . 23
5.3. Recommendation on Responding to Congestion . . . . . . . . 24
7.2. Bit- & Packet-congestible World
5.4. Recommended Future Research . . . . . . . . . . . . . . . 24
8.
6. Security Considerations . . . . . . . . . . . . . . . . . . . 25
9. Conclusions 24
7. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 25
8. Comments Solicited . . 26
10. Acknowledgements . . . . . . . . . . . . . . . . . . . . 25
9. References . . . . 28
11. Comments Solicited . . . . . . . . . . . . . . . . . . . . . . 28
12. 25
9.1. Normative References . . . . . . . . . . . . . . . . . . . 25
9.2. Informative References . . . . . . . . . . 28
12.1. Normative References . . . . . . . . 26
Appendix A. Congestion Notification Definition: Further
Justification . . . . . . . . . . . 28
12.2. Informative References . . . . . . . . . 30
Appendix B. Idealised Wire Protocol . . . . . . . . . . . 29
Editorial Comments . . . . 30
B.1. Protocol Coding . . . . . . . . . . . . . . . . . . . .
Appendix A. . 30
B.2. Example Scenarios . . . . . . . . . . . . . . . . . . 32
A.1. Notation . . 32
B.2.1. Notation . . . . . . . . . . . . . . . . . . . . . . . 32
A.2.
B.2.2. Bit-congestible resource, equal bit rates (Ai) . . . . . . 32
A.3.
B.2.3. Bit-congestible resource, equal packet rates (Bi) . . . . 33
A.4.
B.2.4. Pkt-congestible resource, equal bit rates (Aii) . . . . . 34
A.5.
B.2.5. Pkt-congestible resource, equal packet rates (Bii) . . . . 35

Appendix B. Congestion Notification Definition: Further
Justification . . . . . . . . . . . . . . . . . . . . 35
Appendix C. Byte-mode Drop Complicates Policing C. Byte-mode Drop Complicates Policing Congestion
Response . . . . . . . . . . . . . . . . . . . . . . 36
Author's Address . . . . . . . . . . . . . . 35
Appendix D. Changes from Previous Versions . . . . . . . . . . . 37

Changes from Previous Versions

To be removed by 36

1. Introduction

When notifying congestion, the RFC Editor on publication.

Full incremental diffs between each version are available at
<http://www.cs.ucl.ac.uk/staff/B.Briscoe/pubs.html#byte-pkt-congest>
or
<http://tools.ietf.org/wg/tsvwg/draft-ietf-tsvwg-byte-pkt-congest/>
(courtesy problem of the rfcdiff tool):

From -00 how (and whether) to -01 (this version):

* Minor clarifications throughout take
packet sizes into account has exercised the minds of researchers and updated references

From briscoe-byte-pkt-mark-02
practitioners for as long as active queue management (AQM) has been
discussed. Indeed, one reason AQM was originally introduced was to ietf-byte-pkt-congest-00:

* Added note
reduce the lock-out effects that small packets can have on relationship large
packets in drop-tail queues. This memo aims to existing RFCs

* Posed state the question of whether packet-congestion could become
common principles
we should be using and deferred it to the IRTF ICCRG. Added ref come to conclusions on what these
principles will mean for future protocol design, taking into account
the
dual-resource queue (DRQ) proposal.

* Changed PCN references from deployments we have already.

The byte vs. packet dilemma arises at three stages in the PCN charter & architecture to congestion
notification process:

Measuring congestion: When the PCN marking behaviour draft most likely congested resource decides locally to imminently
become
measure how congested it is. (Should the standards track WG item.

From -01 to -02:

* Abstract reorganised to align with clearer separation of issue queue measure its length
in bytes or packets?);

Coding congestion notification into the memo.

* Introduction reorganised with motivating arguments removed wire protocol: When the
congested resource decides whether to
new Section 2.

* Clarified avoiding lock-out of large packets is not notify the main or
only motivation for RED.

* Mentioned choice level of drop
congestion on each particular packet. (When a queue considers
whether to notify congestion by dropping or marking explicitly throughout,
rather than trying to coin a word to mean either.

* Generalised particular
packet, should its decision depend on the discussion throughout to any byte-size of the
particular packet forwarding
function on any network equipment, not just routers.

* Clarified being dropped or marked?);

Decoding congestion notification from the last point about why this is a good time to sort
out this issue: because it will be hard / impossible wire protocol: When the
transport interprets the notification in order to design
new transports unless we decide whether the network or how much
to respond to congestion. (Should the transport is allowing for packet size.

* Added statement explaining take into account
the horizon byte-size of each missing or marked packet?).

Consensus has emerged over the years concerning the first stage:
whether queues are measured in bytes or packets, termed byte-mode
queue measurement or packet-mode queue measurement. This memo is long
term, but with short term expediency
records this consensus in mind.

* Added material the RFC Series. In summary the choice
solely depends on scaling congestion control whether the resource is congested by bytes or
packets.

The controversy is mainly around the last two stages to do with packet size
(Section 2.1).

* Separated out issue of normalising TCP's bit rate from issue of
preference
encoding congestion notification into packets: whether to control packets (Section 2.3).

* Divided up Congestion Measurement section allow for clarity,
including new material on fixed
the size of the specific packet buffers and buffer
carving (Section 4.1.1 & Section 6.2.1) and on notifying congestion
measurement in wireless link technologies without queues
(Section 4.2).

* Added section i) when the
network encodes or ii) when the transport decodes the congestion
notification.

Currently, the RFC series is silent on 'Making Transports Robust against Control
Packet Losses' (Section 6.2.3) with existing & new material
included.

* Added tabulated results this matter other than a paper
trail of vendor survey on advice referenced from [RFC2309], which conditionally
recommends byte-mode (packet-size dependent) drop
variant [pktByteEmail].
The primary purpose of RED (Table 2).

From -00 this memo is to -01:

* Clarified applicability build a definitive consensus
against such deliberate preferential treatment for small packets in
AQM algorithms and to drop as well as ECN.

* Highlighted DoS vulnerability.

* Emphasised that drop-tail suffers from similar problems record this advice within the RFC series.
Fortunately all the implementers who responded to
byte-mode drop, so only byte-mode drop should be turned off, our survey
(Section 3.2.4) have not RED itself.

* Clarified followed the original apparent motivations for recommending
byte-mode drop included protecting SYNs and pure ACKs more than
equalising earlier advice, so the bit rates of TCPs with different segment sizes.
Removed some conjectured motivations.

* Added support
consensus this memo argues for updates seems to TCP already exist in progress (ackcc & ecn-syn-
ack).

* Updated survey results with newly arrived data.

* Pulled all recommendations together into the conclusions.

* Moved some detailed points into two additional appendices and a
note.

* Considerable clarifications throughout.

* Updated references

1. Introduction

When notifying congestion, the problem
implementations.

The primary conclusion of how (and whether) to take this memo is that packet sizes size should be
taken into account has exercised the minds when transports read congestion indications, not
when network equipment writes them. Reducing drop of researchers and
practitioners for as long as active queue management (AQM) has been
discussed. Indeed, one reason AQM was originally introduced was to
reduce the lock-out effects that small packets can have on large
packets in drop-tail queues. This memo aims
has some tempting advantages: i) it drops less control packets, which
tend to state the principles
we should be using small and to come to conclusions ii) it makes TCP's bit-rate less dependent on what
packet size. However, there are ways of addressing these
principles will mean for future protocol design, taking into account issues at
the deployments we have already.

Note transport layer, rather than reverse engineering network
forwarding to fix specific transport problems.

The second conclusion is that network layer algorithms like the byte vs. byte-
mode packet dilemma concerns congestion
notification irrespective of whether it is signalled implicitly by drop or using explicit congestion notification (ECN [RFC3168] or PCN
[I-D.ietf-pcn-marking-behaviour]). Throughout this document, unless
clear from the context, the term marking will be used to mean
notifying congestion explicitly, while congestion notification will variant of RED should not be used to mean notifying congestion either implicitly by drop or
explicitly by marking.

If the load on fewer
small packets, because that creates a resource depends on the rate at which packets
arrive, it is called packet-congestible. If the load depends on the
rate at which bits arrive it perverse incentive for
transports to use tiny segments, consequently also opening up a DoS
vulnerability.

This memo is called bit-congestible.

Examples of packet-congestible resources are route look-up engines
and firewalls, because load depends on initially concerned with how many we should correctly scale
congestion control functions with packet headers they
have size for the long term. But
it also recognises that expediency may be necessary to process. Examples deal with
existing widely deployed protocols that don't live up to the long
term goal. It turns out that the 'correct' variant of bit-congestible resources are
transmission links, radio power and most buffer memory, because RED to deploy
seems to be the
load depends on how many bits they have one everyone has deployed, and no-one who responded
to transmit or store. Some
machine architectures use fixed size packet buffers, so buffer memory
in these cases our survey has implemented the other variant. However, at the
transport layer, TCP congestion control is packet-congestible (see Section 4.1.1).

Note a widely deployed protocol
that information is generally processed or transmitted we argue doesn't scale correctly with packet size. To date this
hasn't been a
minimum granularity greater significant problem because most TCPs have been used
with similar packet sizes. But, as we design new congestion
controls, we should build in scaling with packet size rather than a bit (e.g. octets). The
appropriate granularity for
assuming we should follow TCP's example.

This memo continues as follows. Terminology and scoping are
discussed next, and the resource in question SHOULD be used,
but for the sake of brevity we will talk in terms of bytes reasons to make the recommendations presented
in this
memo.

Resources may be congestible at higher levels of granularity than
packets, memo now are given in Section 1.2. Motivating arguments for instance stateful firewalls
our advice are flow-congestible given in Section 2. We then survey the advice given
previously in the RFC series, the research literature and
call-servers are session-congestible. This memo focuses on
congestion of connectionless resources, but the same principles may
be applicable for congestion notification
deployed legacy (Section 3) before listing outstanding issues
(Section 4) that will need resolution both to inform future protocols controlling per-
flow
designs and per-session processing or state. to handle legacy. We then give concrete recommendations
for the way forward in (Section 5). We finally give security
considerations in Section 6. The interested reader can also find
further discussions about the theme of byte vs. packet dilemma arises at three stages in the congestion
notification process:

Measuring congestion When
appendices.

This memo intentionally includes a non-negligible amount of material
on the congested resource decides locally how subject. A busy reader can jump right into Section 5 to measure how congested it is. (Should read
a summary of the queue recommendations for the Internet community.

1.1. Terminology and Scoping

The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
"SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
document are to be measured interpreted as described in
bytes or packets?);

Coding congestion notification into the wire protocol: When the
congested resource decides how [RFC2119].

Congestion Notification: Rather than aim to notify the level of achieve what many have
tried and failed, this memo will not try to define congestion.
(Should the level It
will give a working definition of what congestion notification depend on the byte-size of each
particular packet carrying the notification?);

Decoding congestion notification from the wire protocol: When the
transport interprets the notification. (Should the byte-size of a
missing or marked packet
should be taken into account?).

In RED, whether to use packets or bytes when measuring queues is
called packet-mode or byte-mode queue measurement. This choice is
now fairly well understood but mean for this document. Congestion
notification is included in Section 4 a changing signal that aims to document
it in communicate the RFC series.

The controversy
ratio E/L. E is mainly around the other two stages: whether instantaneous excess load offered to
allow for packet size when the network codes or when the transport
decodes congestion notification. In RED, the variant that reduces
drop probability for packets based on their size in bytes is called
byte-mode drop, while the variant a
resource that doesn't it is called packet mode
drop. Whether queues are measured in bytes either incapable of serving or packets unwilling to
serve. L is an
orthogonal choice, termed byte-mode queue measurement or packet-mode
queue measurement.

Currently, the RFC series instantaneous offered load.

The phrase `unwilling to serve' is silent on this matter other than added, because AQM systems
(e.g. RED, PCN [RFC5670]) set a paper
trail of advice referenced from [RFC2309], which conditionally
recommends byte-mode (packet-size dependent) drop [pktByteEmail].
However, all virtual limit smaller than the implementers who responded
actual limit to our survey
(Section 6.2.4) have not followed this advice. The primary purpose
of the resource, then notify when this memo virtual limit
is to build a definitive consensus against deliberate
preferential treatment for small packets exceeded in AQM algorithms and order to
record this advice within avoid congestion of the RFC series.

Now is a good time to discuss whether fairness between different
sized packets would best be implemented actual capacity.

Note that the denominator is offered load, not capacity.
Therefore congestion notification is a real number bounded by the
range [0,1]. This ties in with the network layer, or at most well-understood measure
of congestion notification: drop fraction (often loosely called
loss rate). It also means that congestion has a natural
interpretation as a probability; the transport, for probability of offered
traffic not being served (or being marked as at risk of not being
served). Appendix A describes a number further incidental benefit that
arises from using load as the denominator of reasons:

1. congestion
notification.

Explicit and Implicit Notification: The packet vs. byte issue requires speedy resolution because the
IETF pre-congestion vs. packet dilemma
concerns congestion notification (PCN) working group is about to
standardise the external behaviour irrespective of a whether it is
signalled implicitly by drop or using explicit congestion
notification (ECN [RFC3168] or PCN [RFC5670]). Throughout this
document, unless clear from the context, the term marking will be
used to mean notifying congestion explicitly, while congestion
notification (AQM) algorithm [I-D.ietf-pcn-marking-behaviour];

2. [RFC2309] says RED may will be used to mean notifying congestion either take account of packet size
implicitly by drop or not
when dropping, but gives no recommendation between explicitly by marking.

Bit-congestible vs. Packet-congestible: If the two,
referring instead to advice load on a resource
depends on the performance implications in an
email [pktByteEmail], rate at which recommends byte-mode drop. Further,
just before RFC2309 was issued, an addendum was added to packets arrive, it is called packet-
congestible. If the
archived email that revisited load depends on the issue of packet vs. byte-mode
drop in its last para, making the recommendation less clear-cut;

3. Without the present memo, the only advice in the RFC series rate at which bits arrive
it is called bit-congestible.

Examples of packet-congestible resources are route look-up engines
and firewalls, because load depends on how many packet size bias in AQM algorithms would be a reference headers
they have to an
archived email in [RFC2309] (including an addendum at the end process. Examples of bit-congestible resources are
transmission links, radio power and most buffer memory, because
the email to correct the original).

4. The IRTF Internet Congestion Control Research Group (ICCRG)
recently took on the challenge of building consensus load depends on what
common congestion control support should be required from network
forwarding functions in future
[I-D.irtf-iccrg-welzl-congestion-control-open-research]. The
wider Internet community needs how many bits they have to discuss whether the complexity
of adjusting for packet size should be in the network transmit or store.
Some machine architectures use fixed size packet buffers, so
buffer memory in
transports;

5. Given there are many good reasons why larger path max
transmission units (PMTUs) would help solve these cases is packet-congestible (see
Section 3.1.1).

Currently a number design goal of scaling
issues, we don't want network processing equipment such as
routers and firewalls is to create any bias against large packets keep packet processing uncongested
even under worst case bit rates with minimum packet sizes.
Therefore, packet-congestion is currently rare, but there is no
guarantee that it will not become common with future technology
trends.

Note that information is generally processed or transmitted with a
minimum granularity greater than their true cost;

6. a bit (e.g. octets). The IETF has started to consider
appropriate granularity for the question of fairness between
flows that use different packet sizes (e.g. resource in question should be
used, but for the small-packet
variant sake of TCP-friendly rate control, TFRC-SP [RFC4828]). Given
transports with different packet sizes, if brevity we don't decide
whether the network or the transport should allow for packet
size, it will talk in terms of bytes
in this memo.

Coarser granularity: Resources may be hard if not impossible to design any transport
protocol so that its bit-rate relative to other transports meets
design guidelines [RFC5033] (Note however that, if the concern
were fairness between users, rather than between flows
[Rate_fair_Dis], relative rates between flows would have to come
under run-time control rather congestible at higher levels
of granularity than being embedded in protocol
designs). packets, for instance stateful firewalls are
flow-congestible and call-servers are session-congestible. This
memo is initially concerned with how we should correctly scale focuses on congestion control functions with packet size for of connectionless resources, but the long term. But
it also recognises that expediency
same principles may be necessary to deal with
existing widely deployed applicable for congestion notification
protocols that don't live up controlling per-flow and per-session processing or
state.

RED Terminology: In RED, whether to use packets or bytes when
measuring queues is respectively called packet-mode or byte-mode
queue measurement. And if the long
term goal. It turns out that probability of dropping a packet
depends on its byte-size it is called byte-mode drop, whereas if
the 'correct' variant drop probability is independent of RED to deploy
seems a packet's byte-size it is
called packet-mode drop.

1.2. Why now?

Now is a good time to be the one everyone has deployed, and no-one who responded
to our survey has discuss whether fairness between different
sized packets would best be implemented in the other variant. However, network layer, or at
the
transport layer, TCP congestion control is transport, for a widely deployed protocol
that we argue doesn't scale correctly with number of reasons:

1. The packet size. To date this
hasn't been a significant problem vs. byte issue requires speedy resolution because most TCPs have been used
with similar packet sizes. But, as we design new the
IETF pre-congestion notification (PCN) working group is
standardising the external behaviour of a PCN congestion
controls, we should build in scaling with
notification (AQM) algorithm [RFC5670];

2. [RFC2309] says RED may either take account of packet size rather than
assuming we should follow TCP's example.

Motivating arguments for our or not
when dropping, but gives no recommendation between the two,
referring instead to advice are given next on the performance implications in Section 2.
Then an
email [pktByteEmail], which recommends byte-mode drop. Further,
just before RFC2309 was issued, an addendum was added to the body of
archived email that revisited the memo starts from first principles, defining
congestion notification issue of packet vs. byte-mode
drop in Section 3 then determining its last paragraph, making the correct way
to measure congestion (Section 4) and to design an idealised
congestion notification protocol (Section 5). It then surveys recommendation less clear-
cut;

3. Without the present memo, the only advice given previously in the RFC series, series on
packet size bias in AQM algorithms would be a reference to an
archived email in [RFC2309] (including an addendum at the research literature
and end of
the deployed legacy (Section 6) before listing outstanding issues
(Section 7) that will need resolution both email to achieve correct the ideal
protocol and original).

4. The IRTF Internet Congestion Control Research Group (ICCRG)
recently took on the challenge of building consensus on what
common congestion control support should be required from network
forwarding functions in future [I-D.irtf-iccrg-welzl]. The wider
Internet community needs to handle legacy. After discussing security
considerations (Section 8) strong recommendations for discuss whether the way forward
are given complexity of
adjusting for packet size should be in the conclusions (Section 9).

1.1. Requirements Notation

The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
"SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" network or in this
document
transports;

5. Given there are many good reasons why larger path max
transmission units (PMTUs) would help solve a number of scaling
issues, we don't want to create any bias against large packets
that is greater than their true cost;

6. The IETF has started to consider the question of fairness between
flows that use different packet sizes (e.g. in the small-packet
variant of TCP-friendly rate control, TFRC-SP [RFC4828]). Given
transports with different packet sizes, if we don't decide
whether the network or the transport should allow for packet
size, it will be interpreted as described hard if not impossible to design any transport
protocol so that its bit-rate relative to other transports meets
design guidelines [RFC5033] (Note however that, if the concern
were fairness between users, rather than between flows
[Rate_fair_Dis], relative rates between flows would have to come
under run-time control rather than being embedded in [RFC2119]. protocol
designs).

2. Motivating Arguments

2.1. Scaling Congestion Control with Packet Size

There are two ways of interpreting a dropped or marked packet. It
can either be considered as a single loss event or as loss/marking of
the bytes in the packet. Here we try to design a test to see which
approach scales with packet size.

Given bit-congestible is the more common case, case (see Section 1.1),
consider a bit-
congestible bit-congestible link shared by many flows, so that each
busy period tends to cause packets to be lost from different flows.
The test compares two identical scenarios with the same applications,
the same numbers of sources and the same load. But the sources break
the load into large packets in one scenario and small packets in the
other. Of course, because the load is the same, there will be
proportionately more packets in the small packet case.

The test of whether a congestion control scales with packet size is
that it should respond in the same way to the same congestion
excursion, irrespective of the size of the packets that the bytes
causing congestion happen to be broken down into.

A bit-congestible queue suffering a congestion excursion has to drop
or mark the same excess bytes whether they are in a few large packets
or many small packets. So for the same congestion excursion, the
same amount of bytes have to be shed to get the load back to its
operating point. But, of course, for smaller packets more packets
will have to be discarded to shed the same bytes.

If all the transports interpret each drop/mark as a single loss event
irrespective of the size of the packet dropped, those with smaller
packets will respond more to the same congestion excursion, failing
our test. On the other hand, if they respond proportionately less
when smaller packets are dropped/marked, overall they will be able to
respond the same to the same congestion excursion.

Therefore, for a congestion control to scale with packet size it
should respond to dropped or marked bytes (as TFRC-SP [RFC4828]
effectively does), not just to dropped or marked packets irrespective
of packet size (as TCP does).

The email [pktByteEmail] referred to by RFC2309 says the question of
whether a packet's own size should affect its drop probability
"depends on the dominant end-to-end congestion control mechanisms".
But we argue the network layer should not be optimised for whatever
transport is predominant.

TCP congestion control ensures that flows competing for the same
resource each maintain the same number of segments in flight,
irrespective of segment size. So under similar conditions, flows
with different segment sizes will get different bit rates. But even
though reducing the drop probability of small packets helps ensure
TCPs with different packet sizes will achieve similar bit rates, we
argue this correction should be achieved in made to TCP itself, not in to the network.
network in order to fix one transport, no matter how prominent it is.

Effectively, favouring small packets is reverse engineering of the
network layer around TCP, contrary to the excellent advice in
[RFC3426], which asks designers to question "Why are you proposing a
solution at this layer of the protocol stack, rather than at another
layer?"

2.2. Avoiding Perverse Incentives to (ab)use Smaller Packets

Increasingly, it is being recognised that a protocol design must take
care not to cause unintended consequences by giving the parties in
the protocol exchange perverse incentives [Evol_cc][RFC3426]. Again,
imagine a scenario where the same bit rate of packets will contribute
the same to congestion bit-congestion of a link irrespective of whether it is
sent as fewer larger packets or more smaller packets. A protocol
design that caused larger packets to be more likely to be dropped
than smaller ones would be dangerous in this case:

Normal transports: Even if a transport is not actually malicious, if
it finds small packets go faster, over time it will tend to act in
its own interest and use them. Queues that give advantage to
small packets create an evolutionary pressure for transports to
send at the same bit-rate but break their data stream down into
tiny segments to reduce their drop rate. Encouraging a high
volume of tiny packets might in turn unnecessarily overload a
completely unrelated part of the system, perhaps more limited by
header-processing than bandwidth.

Malicious transports: A queue that gives an advantage to small
packets can be used to amplify the force of a flooding attack. By
sending a flood of small packets, the attacker can get the queue
to discard more traffic in large packets, allowing more attack
traffic to get through to cause further damage. Such a queue
allows attack traffic to have a disproportionately large effect on
regular traffic without the attacker having to do much work.

Note that, although the byte-mode drop variant of RED amplifies
small packet attacks, drop-tail queues amplify small packet
attacks even more (see Security Considerations in Section 8). 6).
Wherever possible neither should be used.

Normal transports: Even if a transport is not malicious, if it finds
small packets go faster, it will tend to act in its own interest
and use them. Queues that give advantage to small packets create
an evolutionary pressure for transports to send at the same bit-
rate but break their data stream down into tiny segments to reduce
their drop rate. Encouraging a high volume of tiny packets might
in turn unnecessarily overload a completely unrelated part of the
system, perhaps more limited by header-processing than bandwidth.

Imagine two unresponsive flows arrive at a bit-congestible
transmission link each with the same bit rate, say 1Mbps, but one
consists of 1500B and the other 60B packets, which are 25x smaller.
Consider a scenario where gentle RED [gentle_RED] is used, along with
the variant of RED we advise against, i.e. where the RED algorithm is
configured to adjust the drop probability of packets in proportion to
each packet's size (byte mode packet drop). In this case, if RED
drops 25% of the larger packets, it will aim to drop 1% of the
smaller packets (but in practice it may drop more as congestion
increases [RFC4828](S.B.4)[Note_Variation]). [RFC4828](S.B.4)). Even though both flows arrive with the
same bit rate, the bit rate the RED queue aims to pass to the line
will be 750k for the flow of larger packet but 990k for the smaller
packets (but because of rate variation it will be less than this
target).

It can be seen that this behaviour reopens the same denial of service
vulnerability that drop tail queues offer to floods of small packet,
though not necessarily as strongly (see Section 8). 6).

2.3. Small != Control

It is tempting to drop small packets with lower probability to
improve performance, because many control packets are small (TCP SYNs
& ACKs, DNS queries & responses, SIP messages, HTTP GETs, etc) and
dropping fewer control packets considerably improves performance.
However, we must not give control packets preference purely by virtue
of their smallness, otherwise it is too easy for any data source to
get the same preferential treatment simply by sending data in smaller
packets. Again we should not create perverse incentives to favour
small packets rather than to favour control packets, which is what we
intend.

Just because many control packets are small does not mean all small
packets are control packets.

So again, rather than fix these problems in the network layer, we
argue that the transport should be made more robust against losses of
control packets (see 'Making Transports Robust against Control Packet
Losses' in Section 6.2.3). 3.2.3).

2.4. Implementation Efficiency

Allowing for packet size at the transport rather than in the network
ensures that neither the network nor the transport needs to do a
multiply operation--multiplication by packet size is effectively
achieved as a repeated add when the transport adds to its count of
marked bytes as each congestion event is fed to it. This isn't a
principled reason in itself, but it is a happy consequence of the
other principled reasons.

3. Working Definition of Congestion Notification

Rather than aim to achieve what many have tried and failed, this memo
will not try to define congestion. It will give a working definition The State of what congestion notification should be taken to mean the Art

The original 1993 paper on RED [RED93] proposed two options for this
document. Congestion notification is a changing signal that aims to
communicate the ratio E/L, where E is the instantaneous excess load
offered to a resource that it cannot (or would not) serve
RED active queue management algorithm: packet mode and L is
the instantaneous offered load.

The phrase `would not serve' is added, because AQM systems (e.g.
RED, PCN [I-D.ietf-pcn-marking-behaviour]) use a virtual capacity
smaller than actual capacity, then notify congestion of this virtual
capacity in order to avoid congestion of the actual capacity.

Note that the denominator is offered load, not capacity. Therefore
congestion notification is a real number bounded by byte mode.
Packet mode measured the range [0,1].
This ties queue length in packets and dropped (or
marked) individual packets with the most well-understood measure of congestion
notification: drop fraction (often loosely called loss rate). It
also means that congestion has a natural interpretation as a
probability; the probability independent of offered traffic not being served (or
being marked as at risk of not being served). Appendix B describes a
further incidental benefit that arises from using load as their
size. Byte mode measured the
denominator of congestion notification.

4. Congestion Measurement

4.1. Congestion Measurement by Queue Length

Queue queue length is usually the most correct in bytes and simplest way to measure
congestion of a resource. To avoid the pathological effects of drop
tail, marked an AQM function can then be used
individual packet with probability in proportion to its size
(relative to transform queue length into the probability of dropping or marking a maximum packet (e.g. RED's
piecewise linear function between thresholds). If the resource is
bit-congestible, size). In the length paper's outline of
further work, it was stated that no recommendation had been made on
whether the queue SHOULD be size should be measured in bytes.
If the resource is packet-congestible, the length of bytes or packets, but
noted that the queue SHOULD difference could be measured significant.

When RED was recommended for general deployment in packets. No other choice makes sense, because 1998 [RFC2309],
the
number two modes were mentioned implying the choice between them was a
question of packets waiting performance, referring to a 1997 email [pktByteEmail] for
advice on tuning. This email clarified that there were in the fact two
orthogonal choices: whether to measure queue isn't relevant if the resource
gets congested by length in bytes or
packets (Section 3.1 below) and vice versa. We discuss whether the implications
on RED's byte mode and drop probability of an
individual packet mode for measuring should depend on its own size (Section 3.2 below).

3.1. Congestion Measurement: Status

The choice of which metric to use to measure queue length was left
open in
Section 6.

4.1.1. Fixed Size Packet Buffers

Some, mostly older, queuing hardware sets aside fixed sized buffers RFC2309. It is now well understood that queues for bit-
congestible resources should be measured in which to store each packet bytes, and queues for
packet-congestible resources should be measured in the queue. Also, with some
hardware, any fixed sized packets.

Where buffers not completely filled by a packet are padded when transmitted not configured or legacy buffers cannot be
configured to the wire. If above guideline, we imagine do not have to make allowances
for such legacy in future protocol design. If a theoretical
forwarding system with both queuing and transmission bit-congestible
buffer is measured in fixed, MTU-
sized units, it should clearly be treated as packet-congestible,
because packets, the queue length in packets would be a good model of
congestion of operator will have set the lower layer link.

If we now imagine
thresholds mindful of a hybrid forwarding system with transmission delay
largely dependent on the byte-size typical mix of packets but buffers of one MTU
per packet, it should strictly require a more complex sizes. Any AQM
algorithm to
determine the probability of congestion. It should on such a buffer will be treated as two
resources in sequence, where the sum oversensitive to high proportions
of the byte-sizes small packets, e.g. a DoS attack, and undersensitive to high
proportions of the packets
within each packet large packets. But an operator can safely keep such a
legacy buffer models because any undersensitivity during unusual traffic
mixes cannot lead to congestion of the line while the
length of collapse given the buffer will
eventually revert to tail drop, discarding proportionately more large
packets.

Some modern queue implementations give a choice for setting RED's
thresholds in packets models congestion of the queue. Then
the probability of congesting the forwarding buffer would byte-mode or packet-mode. This may merely be a
conditional probability--conditional on an
administrator-interface preference, not altering how the previously calculated
probability of congesting queue itself
is measured but on some hardware it does actually change the line.

However, in systems that use fixed size buffers, way it
measures its queue. Whether a resource is unusual for
all bit-congestible or packet-
congestible is a property of the buffers used by resource, so an interface to admin should not
ever need to, or be able to, configure the same size. Typically
pools way a queue measures
itself.

We believe the question of different sized buffers are provided (Cisco uses the term
'buffer carving' for the process of dividing up memory into whether to measure queues in bytes or
packets is fairly well understood these
pools [IOSArch]). Usually, if days. The only outstanding
issues concern how to measure congestion when the pool of small buffers queue is
exhausted, arriving small packets can borrow space in the pool of
large buffers, bit
congestible but not the resource is packet congestible or vice versa. However, it

There is easier no controversy over what should be done. It's just you have
to be an expert in probability to work out what should be done
(summarised in the following section) and, even if we temporarily set you have, it's not
always easy to find a practical algorithm to implement it.

3.1.1. Fixed Size Packet Buffers

Some, mostly older, queuing hardware sets aside fixed sized buffers
in which to store each packet in the possibility of
such borrowing. Then, queue. Also, with some
hardware, any fixed pools of buffers for different sized packets and no borrowing, buffers not completely filled by a packet
are padded when transmitted to the size of each pool wire. If we imagine a theoretical
forwarding system with both queuing and transmission in fixed, MTU-
sized units, it should clearly be treated as packet-congestible,
because the current queue length in each pool packets would both be measured in packets. So an
AQM algorithm would have to maintain the queue length for each pool,
and judge whether to drop/mark a packet good model of
congestion of the lower layer link.

If we now imagine a particular size by
looking at hybrid forwarding system with transmission delay
largely dependent on the pool for byte-size of packets but buffers of that size one MTU
per packet, it should strictly require a more complex algorithm to
determine the probability of congestion. It should be treated as two
resources in sequence, where the sum of the byte-sizes of the packets
within each packet buffer models congestion of the line while the
length of the queue in packets models congestion of the queue. Then
the probability of congesting the forwarding buffer would be a
conditional probability--conditional on the previously calculated
probability of congesting the line.

In systems that use fixed size buffers, it is unusual for all the
buffers used by an interface to be the same size. Typically pools of
different sized buffers are provided (Cisco uses the term 'buffer
carving' for the process of dividing up memory into these pools
[IOSArch]). Usually, if the pool of small buffers is exhausted,
arriving small packets can borrow space in the pool of large buffers,
but not vice versa. However, it is easier to work out what should be
done if we temporarily set aside the possibility of such borrowing.
Then, with fixed pools of buffers for different sized packets and no
borrowing, the size of each pool and the current queue length in each
pool would both be measured in packets. So an AQM algorithm would
have to maintain the queue length for each pool, and judge whether to
drop/mark a packet of a particular size by looking at the pool for
packets of that size and using the length (in packets) of its queue.

We now return to the issue we temporarily set aside: small packets
borrowing space in larger buffers. In this case, the only difference
is that the pools for smaller packets have a maximum queue size that
includes all the pools for larger packets. And every time a packet
takes a larger buffer, the current queue size has to be incremented
for all queues in the pools of buffers less than or equal to the
buffer size used.

We will return to borrowing of fixed sized buffers when we discuss
biasing the drop/marking probability of a specific packet because of
its size in Section 6.2.1. 3.2.1. But here we can give a simple summary of
the present discussion on how to measure the length of queues of
fixed buffers: no matter how complicated the scheme is, ultimately
any fixed buffer system will need to measure its queue length in
packets not bytes.

4.2.

3.1.2. Congestion Measurement without a Queue

AQM algorithms are nearly always described assuming there is a queue
for a congested resource and the algorithm can use the queue length
to determine the probability that it will drop or mark each packet.
But not all congested resources lead to queues. For instance,
wireless spectrum is bit-congestible (for a given coding scheme),
because interference increases with the rate at which bits are
transmitted. But wireless link protocols do not always maintain a
queue that depends on spectrum interference. Similarly, power
limited resources are also usually bit-congestible if energy is
primarily required for transmission rather than header processing,
but it is rare for a link protocol to build a queue as it approaches
maximum power.

However,

Nonetheless, AQM algorithms don't do not require a queue in order to work.
For instance spectrum congestion can be modelled by signal quality
using target bit-energy-to-noise-density ratio. And, to model radio
power exhaustion, transmission power levels can be measured and
compared to the maximum power available. [ECNFixedWireless] proposes
a practical and theoretically sound way to combine congestion
notification for different bit-congestible resources at different
layers along an end to end path, whether wireless or wired, and
whether with or without queues.

5. Idealised Wire Protocol Coding

We will start by inventing an idealised congestion notification
protocol before discussing how to make it practical. The idealised
protocol is shown to be correct using examples in Appendix A.

3.2. Congestion notification involves the congested resource coding a
congestion notification signal into the packet stream and the
transports decoding it. Coding: Status

3.2.1. Network Bias when Encoding

The idealised protocol uses two different
(imaginary) fields in each datagram previously mentioned email [pktByteEmail] referred to signal congestion: one for
byte congestion and one for packet congestion.

We are not saying two ECN fields will be needed (and by
[RFC2309] gave advice we are not
saying now disagree with. It said that somehow a resource should be able to drop a packet in one
of two different ways so that
probability should depend on the transport can distinguish which
sort size of drop it was!). These two congestion notification channels
are just a conceptual device. They allow us to defer having to
decide whether to distinguish between byte and the packet congestion when being considered
for drop if the network resource codes the signal or when the transport decodes
it.

However, although this idealised mechanism isn't intended for
implementation, we do want to emphasise that we may need to find a
way to implement it, because is bit-congestible, but not if it could become necessary to somehow
distinguish between bit and packet congestion [RFC3714]. Currently a
design goal of network processing equipment such as routers and
firewalls is to keep packet processing uncongested even under worst
case bit rates with minimum packet sizes. Therefore, packet-
congestion is currently rare,
congestible, but there is no guarantee advised that it will
not become common with future technology trends. most scarce resources in the Internet
were currently bit-congestible. The idealised wire protocol is given below. It accounts for argument continued that if
packet
sizes at the transport layer, not in drops were inflated by packet size (byte-mode dropping), "a
flow's fraction of the network, and packet drops is then only in
the case a good indication of bit-congestible resources. that
flow's fraction of the link bandwidth in bits per second". This avoids was
consistent with a referenced policing mechanism being worked on at
the perverse
incentive to send smaller packets time for detecting unusually high bandwidth flows, eventually
published in 1999 [pBox]. [The problem could and should have been
solved by making the DoS vulnerability that
would otherwise result if policing mechanism count the network were volume of bytes
randomly dropped, not the number of packets.]

A few months before RFC2309 was published, an addendum was added to bias towards them (see
the motivating argument about avoiding perverse incentives above archived email referenced from the RFC, in
Section 2.2):

1. A packet-congestible resource trying which the final
paragraph seemed to code congestion level p_p
into partially retract what had previously been said.
It clarified that the question of whether the probability of
dropping/marking a packet stream should mark depend on its size was not related
to whether the idealised `packet
congestion' field resource itself was bit congestible, but a completely
orthogonal question. However the only example given had the queue
measured in each packets but packet with probability p_p
irrespective drop depended on the byte-size of the packet's size. The transport should then
take a
packet with in question. No example was given the packet congestion field marked to mean
just one mark, irrespective other way round.

In 2000, Cnodder et al [REDbyte] pointed out that there was an error
in the part of the packet size.

2. A bit-congestible resource trying original 1993 RED algorithm that aimed to code time-varying byte-
congestion level p_b
distribute drops uniformly, because it didn't correctly take into a packet stream should mark
account the `byte
congestion' field in each adjustment for packet with probability p_b, again
irrespective of the packet's size. Unlike before, the transport
should take They recommended an
algorithm called RED_4 to fix this. But they also recommended a
further change, RED_5, to adjust drop rate dependent on the square of
relative packet size. This was indeed consistent with the one implied
motivation behind RED's byte congestion field marked mode drop--that we should reverse
engineer the network to
count as a mark on each byte in improve the packet.

The worked examples in Appendix A show that transports can extract
sufficient and correct performance of dominant end-to-
end congestion notification from these protocols control mechanisms.

By 2003, a further change had been made to the adjustment for cases when two flows with different packet sizes have matching
bit rates or matching
size, this time in the RED algorithm of the ns2 simulator. Instead
of taking each packet's size relative to a `maximum packet rates. Examples are also given that mix
these two flows into one size' it
was taken relative to show that a flow with mixed `mean packet sizes
would still size', intended to be a static
value representative of the `typical' packet size on the link. We
have not been able to extract sufficient and correct information.

Sufficient and correct congestion information means that there is
sufficient information find a justification for the two different types of transport
requirements:

Ratio-based: Established transport congestion controls like TCP's
[RFC5681] aim to achieve equal segment rates per RTT through the
same bottleneck--TCP friendliness [RFC3448]. They work with the
ratio of dropped to delivered segments (or marked to unmarked
segments this change in the case of ECN). The example scenarios show
literature, however Eddy and Allman conducted experiments [REDbias]
that
these ratio-based transports are effectively the same whether
counting in bytes or packets, because the units cancel out.
(Incidentally, this is why TCP's bit rate is still proportional assessed how sensitive RED was to
packet size even when byte-counting is used, as recommended for
TCP in [RFC5681], mainly for orthogonal security reasons.)

Absolute-target-based: Other congestion controls proposed in the
research community aim this parameter, amongst other
things. No-one seems to limit the volume of congestion caused have pointed out that this changed algorithm
can often lead to
a constant weight parameter. [MulTCP][WindowPropFair] are
examples drop probabilities of weighted proportionally fair transports designed for
cost-fair environments [Rate_fair_Dis]. In this case, the
transport requires a count (not greater than 1 [which should
ring alarm bells hinting that there's a ratio) of dropped/marked bytes mistake in the bit-congestible case and theory
somewhere]. On 10-Nov-2004, this variant of dropped/marked packets byte-mode packet drop
was made the default in the
packet congestible case.

6. ns2 simulator.

The State byte-mode drop variant of the Art

The original 1993 paper on RED [RED93] proposed two options for is, of course, not the
RED active queue management algorithm: packet mode and byte mode.
Packet mode measured the queue length in only
possible bias towards small packets and dropped (or
marked) individual in queueing algorithms. We have
already mentioned that tail-drop queues naturally tend to lock-out
large packets once they are full. But also queues with a probability independent of their
size. Byte mode measured fixed sized
buffers reduce the queue length in bytes and marked an
individual packet with probability in proportion to its size
(relative that small packets will be dropped if
(and only if) they allow small packets to borrow buffers from the maximum packet size). In the paper's outline of
further work, it
pools for larger packets. As was stated that no recommendation had been made explained in Section 3.1.1 on
whether fixed
size buffer carving, borrowing effectively makes the maximum queue
size should be measured in bytes or packets, but
noted for small packets greater than that the difference could be significant.

When RED was recommended for general deployment in 1998 [RFC2309],
the two modes were mentioned implying large packets, because
more buffers can be used by small packets while less will fit large
packets.

In itself, the choice between them was a
question of performance, referring to a 1997 email [pktByteEmail] bias towards small packets caused by buffer borrowing
is perfectly correct. Lower drop probability for
advice on tuning. This email clarified that there were in fact two
orthogonal choices: whether to measure queue length small packets is
legitimate in bytes or buffer borrowing schemes, because small packets (Section 6.1 below) and whether
genuinely congest the drop probability of an
individual packet should depend on its own size (Section 6.2 below).

6.1. Congestion Measurement: Status

The choice of which metric to use to measure queue length was left
open in RFC2309. It is now well understood that queues for bit-
congestible resources should be measured in bytes, and queues for
packet-congestible resources should be measured machine's buffer memory less than large
packets, given they can fit in more spaces. The bias towards small
packets (see
Section 4).

Where buffers are is not configured or legacy buffers cannot be
configured to the above guideline, we don't have to make allowances
for such legacy in future protocol design. If a bit-congestible
buffer artificially added (as it is measured in packets, the operator will have set RED's byte-mode drop
algorithm), it merely reflects the
thresholds mindful of a typical mix reality of the way fixed buffer
memory gets congested. Incidentally, the bias towards small packets sizes. Any AQM
algorithm on such a
from buffer will be oversensitive to high proportions borrowing is nothing like as large as that of small packets, e.g. a DoS attack, and undersensitive RED's byte-
mode drop.

Nonetheless, fixed-buffer memory with tail drop is still prone to high
proportions of
lock-out large packets. But an operator can safely keep such packets, purely because of the tail-drop aspect. So a
legacy
good AQM algorithm like RED with packet-mode drop should be used with
fixed buffer because any undersensitivity during unusual traffic
mixes cannot lead memories where possible. If RED is too complicated to congestion collapse given the
implement with multiple fixed buffer will
eventually revert pools, the minimum necessary to tail drop, discarding proportionately more
prevent large
packets.

Some modern queue implementations give a choice for setting RED's
thresholds in byte-mode or packet-mode. This may merely be an
administrator-interface preference, not altering how the queue itself packet lock-out is measured but on some hardware it does actually change to ensure smaller packets never use
the way it
measures its queue. Whether a resource is bit-congestible or packet-
congestible is a property last available buffer in any of the resource, so an admin SHOULD NOT
ever need to, or be able to, configure pools for larger packets.

3.2.2. Transport Bias when Decoding

The above proposals to alter the way network equipment to bias towards
smaller packets have largely carried on outside the IETF process
(unless one counts a queue measures
itself.

We believe reference in an informational RFC to an archived
email!). Whereas, within the question of whether IETF, there are many different
proposals to measure queues in bytes or
packets is fairly well understood these days. The only outstanding
issues concern how alter transport protocols to measure congestion when achieve the queue is bit
congestible but same goals,
i.e. either to make the resource is flow bit-rate take account of packet congestible size, or vice versa (see
Section 4). But there is no controversy over what should be done.
It's just you have
to be an expert in probability protect control packets from loss. This memo argues that altering
transport protocols is the more principled approach.

A recently approved experimental RFC adapts its transport layer
protocol to work out what
should be done and, even if you have, it's not always easy take account of packet sizes relative to find typical TCP
packet sizes. This proposes a
practical algorithm to implement it.

6.2. Congestion Coding: Status

6.2.1. Network Bias when Encoding

The previously mentioned email [pktByteEmail] referred to by
[RFC2309] said new small-packet variant of TCP-
friendly rate control [RFC3448] called TFRC-SP [RFC4828].
Essentially, it proposes a rate equation that inflates the choice over whether flow rate
by the ratio of a packet's own typical TCP segment size
should affect its drop probability "depends on (1500B including TCP
header) over the dominant end-to-
end congestion control mechanisms". [Section 2 argues against actual segment size [PktSizeEquCC]. (There are also
other important differences of detail relative to TFRC, such as using
virtual packets [CCvarPktSize] to avoid responding to multiple losses
per round trip and using a minimum inter-packet interval.)

Section 4.5.1 of this
approach, citing TFRC-SP spec discusses the excellent advice implications of
operating in RFC3246.] The referenced
email went on an environment where queues have been configured to argue that drop
smaller packets with proportionately lower probability should depend on the
size of the packet being considered for drop if the resource is bit-
congestible, but not if than larger
ones. But it is packet-congestible, but advised that
most scarce resources only discusses TCP operating in such an environment,
only mentioning TFRC-SP briefly when discussing how to define
fairness with TCP. And it only discusses the Internet were currently bit-congestible.
The argument continued that if packet drops were inflated by packet
size (byte-mode dropping), "a flow's fraction byte-mode dropping
version of RED as it was before Cnodder et al pointed out it didn't
sufficiently bias towards small packets to make TCP independent of the
packet drops is
then a good indication size.

So the TFRC-SP spec doesn't address the issue of that flow's fraction which of the link bandwidth
in bits per second". This was consistent with a referenced policing
mechanism being worked on at the time for detecting unusually high
bandwidth flows, eventually published in 1999 [pBox]. [The problem
could have been solved by making the policing mechanism count network
or the
volume of bytes randomly dropped, not transport _should_ handle fairness between different packet
sizes. In its Appendix B.4 it discusses the number possibility of packets.]

A few months before RFC2309 was published, an addendum was added both
TFRC-SP and some network buffers duplicating each other's attempts to
deliberately bias towards small packets. But the above archived email referenced from discussion is not
conclusive, instead reporting simulations of many of the RFC,
possibilities in which the final
paragraph seemed order to partially retract what had previously been said.
It clarified assess performance but not recommending any
particular course of action.

The paper originally proposing TFRC with virtual packets (VP-TFRC)
[CCvarPktSize] proposed that there should perhaps be two variants to
cater for the question different variants of whether RED. However, as the probability of
dropping/marking TFRC-SP
authors point out, there is no way for a packet should depend on its size was not related transport to know whether the resource itself was bit congestible, but a completely
orthogonal question. However the only example given had the queue
measured in packets but
some queues on its path have deployed RED with byte-mode packet drop depended on the byte-size of
(except if an exhaustive survey found that no-one has deployed it!--
see Section 3.2.4). Incidentally, VP-TFRC also proposed that byte-
mode RED dropping should really square the packet in question. No example was given the other way round.

In 2000, Cnodder et al [REDbyte] pointed out size compensation
factor (like that there was an error
in the part of the original 1993 RED algorithm that aimed RED_5, but apparently unaware of it).

Pre-congestion notification [I-D.ietf-pcn] is a proposal to
distribute drops uniformly, because it didn't correctly take into
account the adjustment use a
virtual queue for packet size. They recommended an
algorithm called RED_4 AQM marking for packets within one Diffserv class
in order to fix this. But they also recommended a
further change, RED_5, give early warning prior to adjust drop rate dependent on the square any real queuing. The
proposed PCN marking algorithms have been designed not to take
account of
relative packet size. This was indeed consistent with one stated
motivation behind RED's byte mode drop--that we should reverse
engineer the network to improve size when forwarding through queues. Instead the performance of dominant end-to-
end congestion control mechanisms.

By 2003, a further change had
general principle has been made to the adjustment for packet
size, this time in the RED algorithm take account of the ns2 simulator. Instead sizes of taking each packet's size relative to a `maximum packet size' it
was taken relative to a `mean packet size', intended to be a static
value representative marked
packets when monitoring the fraction of marking at the `typical' packet size on edge of the link. We
network.

3.2.3. Making Transports Robust against Control Packet Losses

Recently, two RFCs have not been able defined changes to find a justification for this change in the
literature, however Eddy and Allman conducted experiments [REDbias] TCP that assessed how sensitive RED was to this parameter, amongst other
things. No-one seems to have pointed out that this changed algorithm
can often lead to drop probabilities of greater than 1 [which should
ring alarm bells hinting that there's a mistake in the theory
somewhere]. On 10-Nov-2004, this variant of byte-mode packet drop
was made the default in the ns2 simulator.

The byte-mode drop variant of RED is, of course, not the only
possible bias towards make it more
robust against losing small control packets in queueing algorithms. We have
already mentioned that tail-drop queues naturally tend to lock-out
large packets once [RFC5562] [RFC5690]. In
both cases they are full. But also queues with fixed sized
buffers reduce the probability note that small packets will the case for these TCP changes would be dropped
weaker if
(and only if) they allow RED were biased against dropping small packets to borrow buffers from the
pools for larger packets. As was explained in Section 4.1.1 on fixed
size buffer carving, borrowing effectively makes the maximum queue
size for small packets greater than We argue
here that for large packets, because these two proposals are a safer and more buffers can principled way to
achieve TCP performance improvements than reverse engineering RED to
benefit TCP.

Although no proposals exist as far as we know, it would also be used by small packets while less will fit large
packets.

However, in itself, the bias towards small
possible and perfectly valid to make control packets caused robust against
drop by buffer
borrowing is perfectly correct. Lower explicitly requesting a lower drop probability for small
packets is legitimate in buffer borrowing schemes, because small
packets genuinely congest the machine's buffer memory less than large
packets, given they can fit in more spaces. The bias towards small
packets is not artificially added (as it is in RED's byte-mode drop
algorithm), it merely reflects the reality of the way fixed buffer
memory gets congested. Incidentally, the bias towards small packets
from buffer borrowing is nothing like as large as that of RED's byte-
mode drop.

Nonetheless, fixed-buffer memory with tail drop is still prone using their
Diffserv code point [RFC2474] to
lock-out large packets, purely because of the tail-drop aspect. So request a
good AQM algorithm like RED with packet-mode drop should be used scheduling class with
fixed buffer memories where possible. If RED
lower drop.

The re-ECN protocol proposal [I-D.briscoe-tsvwg-re-ecn-tcp] is too complicated to
implement with multiple fixed buffer pools, the minimum necessary
designed so that transports can be made more robust against losing
control packets. It gives queues an incentive to
prevent large packet lock-out is optionally give
preference against drop to ensure smaller packets never use with the last available buffer 'feedback not
established' codepoint in any of the pools for larger packets.

6.2.2. Transport Bias when Decoding

The above proposals proposed 'extended ECN' field. Senders
have incentives to alter use this codepoint sparingly, but they can use it
on control packets to reduce their chance of being dropped. For
instance, the network layer proposed modification to give a bias towards
smaller packets have largely carried TCP for re-ECN uses this
codepoint on outside the IETF process
(unless one counts a reference in an informational RFC SYN and SYN-ACK.

Although not brought to an archived
email!). Whereas, within the IETF, there are many different
proposals to alter transport protocols to achieve the same goals,
i.e. either to make the flow bit-rate take account of packet size, or
to protect control packets a simple proposal from loss. This memo argues Wischik
[DupTCP] suggests that altering
transport protocols is the more principled approach.

A recently approved experimental RFC adapts its transport layer
protocol to take account first three packets of packet sizes relative to typical every TCP
packet sizes. This proposes flow
should be routinely duplicated after a new small-packet variant short delay. It shows that
this would greatly improve the chances of TCP-
friendly rate control [RFC3448] called TFRC-SP [RFC4828].
Essentially, short flows completing
quickly, but it proposes a rate equation that inflates would hardly increase traffic levels on the flow rate
by Internet,
because Internet bytes have always been concentrated in the ratio large
flows. It further shows that the performance of a many typical TCP segment size (1500B including TCP
header) over the actual segment size [PktSizeEquCC]. (There are also
other important differences
applications depends on completion of detail relative to TFRC, such as using
virtual packets [CCvarPktSize] to avoid responding to multiple losses
per round trip and using a minimum inter-packet interval.)

Section 4.5.1 long serial chains of short
messages. It argues that, given most of the value people get from
the Internet is concentrated within short flows, this TFRC-SP spec discusses simple
expedient would greatly increase the implications value of
operating in an environment where queues have been configured to drop
smaller packets with proportionately lower probability than larger
ones. But it only discusses TCP operating in such an environment,
only mentioning the best efforts
Internet at minimal cost.

3.2.4. Congestion Coding: Summary of Status

Table 1: Dependence of flow bit-rate per RTT on packet size s and
drop rate p when discussing how network and/or transport bias towards small packets
to define
fairness with TCP. And it only discusses varying degrees

Table 1 aims to summarise the byte-mode dropping
version positions we may now be in. Each
column shows a different possible AQM behaviour in different queues
in the network, using the terminology of RED as it was before Cnodder et al pointed out it didn't
sufficiently bias towards small packets to make outlined
earlier (RED_1 is basic RED with packet-mode drop). Each row shows a
different transport behaviour: TCP independent of
packet size.

So [RFC5681] and TFRC [RFC3448] on
the top row with TFRC-SP spec doesn't address [RFC4828] below. Suppressing all
inessential details the issue of which of table shows that independence from packet
size should either be achievable by not altering the network TCP transport in
a RED_5 network, or using the transport _should_ handle fairness between different small packet
sizes. In its Appendix B.4 it discusses the possibility of both TFRC-SP and some transport in a
network buffers duplicating each other's attempts to
deliberately bias towards small packets. But the discussion is not
conclusive, instead reporting simulations of many of without any byte-mode dropping RED (top right and bottom
left). Top left is the
possibilities `do nothing' scenario, while bottom right is
the `do-both' scenario in order to assess performance but not recommending which bit-rate would become far too biased
towards small packets. Of course, if any
particular course form of action.

The paper originally proposing TFRC with virtual packets (VP-TFRC)
[CCvarPktSize] proposed that there should perhaps be two variants to
cater for the different variants byte-mode dropping
RED has been deployed on a selection of RED. However, as the TFRC-SP
authors point out, there is no way for congested queues, each path
will present a transport different hybrid scenario to know whether
some queues on its path have deployed RED with transport.

Whatever, we can see that the linear byte-mode packet drop
(except if an exhaustive survey found that no-one has deployed it!--
see Section 6.2.4). Incidentally, VP-TFRC also proposed column in the
middle considerably complicates the Internet. It's a half-way house
that byte-
mode RED dropping doesn't bias enough towards small packets even if one believes
the network should really square be doing the packet size compensation
factor (like biasing. We argue below that of RED_5, but apparently unaware of it).

Pre-congestion notification [I-D.ietf-pcn-marking-behaviour] is a
proposal to use a virtual queue for AQM marking for _all_
network layer bias towards small packets within
one Diffserv class in order to give early warning prior to should be turned off--if
indeed any real
queuing. The proposed PCN marking algorithms equipment vendors have been designed not
to take account of implemented it--leaving packet size when forwarding through queues.
Instead
bias solely as the general principle preserve of the transport layer (solely the
leftmost, packet-mode drop column).

A survey has been to take account conducted of 84 vendors to assess how widely drop
probability based on packet size has been implemented in RED. Prior
to the sizes
of marked packets when monitoring survey, an individual approach to Cisco received confirmation
that, having checked the fraction code-base for each of marking at the edge product ranges,
Cisco has not implemented any discrimination based on packet size in
any AQM algorithm in any of the network.

6.2.3. Making Transports Robust against Control Packet Losses

Recently, two drafts have proposed changes its products. Also an individual
approach to TCP Alcatel-Lucent drew a confirmation that make it more
robust against losing small control packets [I-D.ietf-tcpm-ecnsyn]
[I-D.floyd-tcpm-ackcc]. In both cases they note was very
likely that the case for
these TCP changes would be weaker if none of their products contained RED were biased against dropping
small packets. We argue here code that these two proposals are a safer
and more principled way to achieve TCP performance improvements than
reverse engineering RED
implemented any packet-size bias.

Turning to benefit TCP. our more formal survey (Table 2), about 19% of those
surveyed have replied so far, giving a sample size of 16. Although no proposals exist as far as
we know, it would also be
possible and perfectly valid to make control packets robust against
drop by explicitly requesting a lower drop probability using their
Diffserv code point [RFC2474] do not have permission to request a scheduling class with
lower drop.

The re-ECN protocol proposal [I-D.briscoe-tsvwg-re-ecn-tcp] is
designed so that transports identify the respondents, we can be made more robust against losing
control packets. It gives queues an incentive to optionally give
preference against drop to packets with the 'feedback not
established' codepoint in the proposed 'extended ECN' field. Senders say
that those that have incentives to use this codepoint sparingly, but they can use it
on control packets to reduce their chance responded include most of being dropped. For
instance, the proposed modification to TCP for re-ECN uses this
codepoint on the SYN and SYN-ACK.

Although not brought to the IETF, a simple proposal from Wischik
[DupTCP] suggests that the first three packets of every TCP flow
should be routinely duplicated after larger vendors,
covering a short delay. It shows that
this would greatly improve the chances large fraction of short flows completing
quickly, but it would hardly increase traffic levels on the Internet,
because Internet bytes have always been concentrated in market. They range across the large
flows. It further shows
network equipment vendors at L3 & L2, firewall vendors, wireless
equipment vendors, as well as large software businesses with a small
selection of networking products. So far, all those who have
responded have confirmed that they have not implemented the performance variant
of many typical
applications depends RED with drop dependent on completion of long serial chains of short
messages. It argues that, given most packet size (2 were fairly sure they
had not but needed to check more thoroughly).

+-------------------------------+----------------+-----------------+
| Response | No. of the value people get from
the Internet is concentrated within short flows, this simple
expedient would greatly increase the value vendors | %age of the best efforts
Internet at minimal cost.

6.2.4. Congestion Coding: Summary of Status

Table 1: Dependence of flow bit-rate per RTT 2: Vendor Survey on packet size s and byte-mode drop rate p when network and/or transport bias towards variant of RED (lower drop
probability for small packets
to varying degrees

Table 1 aims to summarise packets)

Where reasons have been given, the positions we may now be in. Each
column shows extra complexity of packet bias
code has been most prevalent, though one vendor had a different possible AQM behaviour in different queues
in the network, using more principled
reason for avoiding it--similar to the terminology argument of Cnodder et al outlined
earlier (RED_1 is basic this document. We
have established that Linux does not implement RED with packet-mode drop). Each row shows a
different transport behaviour: TCP [RFC5681] and TFRC [RFC3448] on
the top row with TFRC-SP [RFC4828] below. Suppressing all
inessential details the table shows that independence from packet size should either be achievable by
drop bias, although we have not altering the TCP transport in
a RED_5 network, or using the small packet TFRC-SP transport in investigated a
network without any byte-mode dropping RED (top right and bottom
left). Top left wider range of open
source code.

Finally, we repeat that RED's byte mode drop is not the `do nothing' scenario, while bottom right is
the `do-both' scenario in which bit-rate would become far too biased only way to
bias towards small packets. Of course, if any form packets--tail-drop tends to lock-out large packets
very effectively. Our survey was of byte-mode dropping
RED vendor implementations, so we
cannot be certain about operator deployment. But we believe many
queues in the Internet are still tail-drop. The company of one of
the co-authors (BT) has been widely deployed on a selection of congested RED, but there are bound to
be many tail-drop queues, each path
will present particularly in access network equipment
and on middleboxes like firewalls, where RED is not always available.

Routers using a different hybrid scenario to its transport.

Whatever, we can see that the linear byte-mode drop column memory architecture based on fixed size buffers with
borrowing may also still be prevalent in the
middle considerably complicates the Internet. It's As explained
in Section 3.2.1, these also provide a half-way house
that doesn't marginal (but legitimate) bias enough
towards small packets packets. So even if one believes
the network should be doing the biasing. We argue below that _all_
network layer though RED byte-mode drop is not
prevalent, it is likely there is still some bias towards small
packets in the Internet due to tail drop and fixed buffer borrowing.

4. Outstanding Issues and Next Steps

4.1. Bit-congestible World

For a connectionless network with nearly all resources being bit-
congestible we believe the recommended position is now unarguably
clear--that the network should be turned off--if
indeed any equipment vendors have implemented it--leaving not make allowance for packet size
bias solely as the preserve of sizes
and the transport layer (solely the
leftmost, packet-mode drop column).

A survey has been conducted of 84 vendors should. This leaves two outstanding issues:

o How to assess how widely handle any legacy of AQM with byte-mode drop
probability based on packet size has been implemented in RED. Prior already
deployed;

o The need to the survey, an individual approach start a programme to Cisco received confirmation
that, having checked the code-base for each update transport congestion
control protocol standards to take account of the product ranges,
Cisco has not implemented any discrimination based on packet size in
any AQM algorithm in any size.

The sample of its products. Also an individual
approach to Alcatel-Lucent drew a confirmation returns from our vendor survey Section 3.2.4 suggest
that byte-mode packet drop seems not to be implemented at all let
alone deployed, or if it was very is, it is likely that none of their products contained RED code that
implemented any packet-size bias.

Turning to our more formal survey (Table 2), about 19% of those
surveyed have replied so far, giving a sample size of 16. Although be very sparse.
Therefore, we do not have permission really need a migration strategy from all but
nothing to identify the respondents, we can say
that those that have responded include most nothing.

A programme of the larger vendors,
covering a large fraction standards updates to take account of packet size in
transport congestion control protocols has started with TFRC-SP
[RFC4828], while weighted TCPs implemented in the market. They range across research community
[WindowPropFair] could form the large
network equipment vendors at L3 & L2, firewall vendors, wireless
equipment vendors, as well as large software businesses basis of a future change to TCP
congestion control [RFC5681] itself.

4.2. Bit- & Packet-congestible World

Nonetheless, a connectionless network with both bit-congestible and
packet-congestible resources is a small
selection different matter. If we believe we
should allow for this possibility in the future, this space contains
a truly open research issue.

We develop the concept of networking products. So far, all those who have
responded have confirmed an idealised congestion notification
protocol that they have not implemented the variant supports both bit-congestible and packet-congestible
resources in Appendix B. The congestion notification requires at
least two flags for congestion of RED bit-congestible and packet-
congestible resources. This hides a fundamental problem--much more
fundamental than whether we can magically create header space for yet
another ECN flag in IPv4, or whether it would work while being
deployed incrementally. A congestion notification protocol must
survive a transition from low levels of congestion to high. Marking
two states is feasible with explicit marking, but much harder if
packets are dropped. Also, it will not always be cost-effective to
implement AQM at every low level resource, so drop dependent on will often have to
suffice. Distinguishing drop from delivery naturally provides just
one congestion flag--it is hard to drop a packet size (2 in two ways that are fairly sure they
haven't but need
distinguishable remotely. This is a similar problem to check more thoroughly).

+-------------------------------+----------------+-----------------+
| Response | No. that of vendors | %age
distinguishing wireless transmission losses from congestive losses.

We should also note that, strictly, packet-congestible resources are
actually cycle-congestible because load also depends on the
complexity of vendors |
+-------------------------------+----------------+-----------------+
| Not implemented | 14 | 17% |
| Not implemented (probably) | 2 | 2% |
| Implemented | 0 | 0% |
| No response | 68 | 81% |
| Total companies/orgs surveyed | 84 | 100% |
+-------------------------------+----------------+-----------------+

Table 2: Vendor Survey on byte-mode drop variant of RED (lower drop
probability for small packets)

Where reasons have been given, each look-up and whether the extra complexity pattern of packet bias
code has been most prevalent, though one vendor had a more principled
reason for avoiding it--similar arrivals is
amenable to the argument of caching or not. Further, this document. We
have established reminds us that Linux does not implement RED with packet size
drop bias, although we have any
solution must not investigated require a wider range of open
source code.

Finally, we repeat that RED's byte mode drop is not the only way to
bias towards small packets--tail-drop tends forwarding engine to lock-out large packets
very effectively. Our survey was of vendor implementations, so we
cannot be certain about operator deployment. But we believe many
queues use excessive
processor cycles in order to decide how to say it has no spare
processor cycles.

Recently, the Internet are still tail-drop. My own company (BT) dual resource queue (DRQ) proposal [DRQ] has
widely deployed RED, but there are bound to be many tail-drop queues,
particularly been made
on the premise that, as network processors become more cost
effective, per packet operations will become more complex
(irrespective of whether more function in access the network equipment and on middleboxes like
firewalls, where RED layer is not always available. Routers using a memory
architecture based on fixed size buffers with borrowing may also
still be prevalent in
desirable). Consequently the Internet. As explained in Section 6.2.1,
these also provide premise is that CPU congestion will
become more common. DRQ is a marginal (but legitimate) bias towards small
packets. So even though proposed modification to the RED byte-mode drop
algorithm that folds both bit congestion and packet congestion into
one signal (either loss or ECN).

The problem of signalling packet processing congestion is not prevalent, it is
likely there is still some bias towards small packets in the
pressing, as most Internet
due to tail drop and fixed buffer borrowing.

7. Outstanding Issues and Next Steps

7.1. Bit-congestible World

For a connectionless network with nearly all resources being are designed to be bit-
congestible we believe the recommended position is now unarguably
clear--that the network should not make allowance for before packet sizes
and the transport should. This leaves two outstanding issues:

o How to handle any legacy of AQM with byte-mode drop already
deployed;

o The need to start a programme processing starts to update transport congest (see
Section 1.1). However, the IRTF Internet congestion control protocol standards to take account of packet size.

The sample research
group (ICCRG) has set itself the task of returns from our vendor survey Section 6.2.4 suggest reaching consensus on
generic forwarding mechanisms that byte-mode packet drop seems not are necessary and sufficient to be implemented
support the Internet's future congestion control requirements (the
first challenge in [I-D.irtf-iccrg-welzl]). Therefore, rather than
not giving this problem any thought at all let
alone deployed, or if it is, all, just because it is likely to be very sparse.
Therefore, hard
and currently hypothetical, we do not really need a migration strategy from all but
nothing to nothing.

A programme of standards updates to take account defer the question of whether packet size in
transport
congestion control protocols has started with TFRC-SP
[RFC4828], while weighted TCPs implemented in the research community
[WindowPropFair] could form the basis of a future change might become common and what to TCP
congestion control [RFC5681] itself.

7.2. Bit- & Packet-congestible World

Nonetheless, a connectionless network with both bit-congestible do if it does to the IRTF
(the 'Small Packets' challenge in [I-D.irtf-iccrg-welzl]).

5. Recommendation and
packet-congestible resources Conclusions

5.1. Recommendation on Queue Measurement

Queue length is a different matter. If we believe we
should allow for this possibility in usually the future, this space contains
a truly open research issue.

The idealised wire protocol coding described in Section 5 requires at
least two flags for most correct and simplest way to measure
congestion of bit-congestible and packet-
congestible resources. This hides a fundamental problem--much more
fundamental than whether we resource. To avoid the pathological effects of drop
tail, an AQM function can magically create header space for yet
another ECN flag in IPv4, then be used to transform queue length into
the probability of dropping or whether it would work while being
deployed incrementally. A congestion notification protocol must
survive marking a transition from low levels of congestion to high. Marking
two states packet (e.g. RED's
piecewise linear function between thresholds).

If the resource is feasible with explicit marking, but much harder if
packets are dropped. Also, it will not always bit-congestible, the length of the queue SHOULD be cost-effective to
implement AQM at every low level resource, so drop will often have to
suffice. Distinguishing drop from delivery naturally provides just
one congestion flag--it is hard to drop a packet
measured in two ways that are
distinguishable remotely. This bytes. If the resource is a similar problem to that of
distinguishing wireless transmission losses from congestive losses.

We should also note that, strictly, packet-congestible resources are
actually cycle-congestible because load also depends on packet-congestible, the
complexity length
of each look-up and whether the pattern queue SHOULD be measured in packets. No other choice makes
sense, because the number of arrivals is
amenable to caching or not. Further, this reminds us that any
solution must not require a forwarding engine to use excessive
processor cycles packets waiting in order to decide how to say it has no spare
processor cycles.

Recently, the dual resource queue (DRQ) proposal [DRQ] has been made
on isn't
relevant if the premise that, as network processors become more cost
effective, per resource gets congested by bytes and vice versa. We
discuss the implications on RED's byte mode and packet operations will become more complex
(irrespective of whether more function mode for
measuring queue length in the network layer is
desirable). Consequently the premise is Section 3.

NOTE WELL that CPU congestion will
become more common. DRQ RED's byte-mode queue measurement is a proposed modification fine, being
completely orthogonal to the byte-mode drop. If a RED
algorithm that folds both bit congestion and packet congestion into
one signal (either loss or ECN).

The problem of signalling packet processing congestion is implementation has
a byte-mode but does not
pressing, as specify what sort of byte-mode, it is most Internet resources are designed to be bit-
congestible before packet processing starts to congest.
probably byte-mode queue measurement, which is fine. However, if in
doubt, the
IRTF Internet congestion control research group (ICCRG) has set
itself the task of reaching consensus vendor should be consulted.

5.2. Recommendation on generic forwarding
mechanisms Notifying Congestion

The strong recommendation is that are necessary and sufficient to support AQM algorithms such as RED SHOULD
NOT use byte-mode drop. More generally, the Internet's future congestion control requirements (the first
challenge in
[I-D.irtf-iccrg-welzl-congestion-control-open-research]). Therefore,
rather than not giving this problem any thought at all, just because
it is hard and currently hypothetical, we defer the question
notification protocols (drop, ECN & PCN) SHOULD take account of
whether
packet congestion might become common and what to do if size when the notification is read by the transport layer, NOT
when it
does to is written by the IRTF (the 'Small Packets' challenge in
[I-D.irtf-iccrg-welzl-congestion-control-open-research]).

8. Security Considerations network layer. This draft recommends approach offers
sufficient and correct congestion information for all known and
future transport protocols and also ensures no perverse incentives
are created that queues do not bias would encourage transports to use inappropriately
small packet sizes.

The alternative of deflating RED's drop probability
towards small packets as this for smaller
packet sizes (byte-mode drop) has no enduring advantages. It is more
complex, it creates a the perverse incentive for
transports to break down their flows fragment segments into
tiny segments. One of pieces and it reopens the
benefits of implementing AQM was meant to be vulnerability to remove this perverse
incentive floods of small-
packets that drop-tail queues gave suffered from and AQM was designed to small packets. Of course, if
transports really want
remove.

Byte-mode drop is a change to make the greatest gains, they don't have to
respond to congestion anyway. But we don't want applications network layer that
are trying makes allowance
for an omission from the design of TCP, effectively reverse
engineering the network layer to behave contrive to discover that they can go faster make two TCPs with
different packet sizes run at equal bit rates (rather than packet
rates) under the same path conditions.

It also improves TCP performance by using
smaller packets.

In practice, transports cannot all reducing the chance that a SYN or
a pure ACK will be trusted dropped, because they are small. But we SHOULD
NOT hack the network layer to respond improve or fix certain transport
protocols. No matter how predominant a transport protocol is (even
if it's TCP), trying to
congestion. So another reason correct for recommending that queues do not
bias drop probability its failings by biasing towards
small packets is to avoid in the
vulnerability network layer creates a perverse incentive to small packet DDoS attacks that would otherwise
result. One
break down all flows from all transports into tiny segments.

So far, our survey of 84 vendors across the benefits industry has drawn
responses from about 19%, none of implementing AQM was meant to be to
remove drop-tail's DoS vulnerability to small packets, so we
shouldn't add it back again.

If most queues whom have implemented AQM with byte-mode drop, the resulting
network would amplify the potency of a small byte mode
packet DDoS attack. At
the first queue the stream drop variant of packets would push aside a greater
proportion of large packets, so more of the small packets would
survive RED. Given there appears to attack the next queue. Thus a flood of small packets
would continue on towards the destination, pushing regular traffic
with large packets out be little, if
any, installed base it seems we can recommend removal of the way in one queue after the next, but
suffering much less byte-mode
drop itself.

Appendix C explains why the ability of networks to police the
response of _any_ transport to congestion depends on bit-congestible
network resources only doing packet-mode not from RED with little, if any, incremental deployment impact.

If a vendor has implemented byte-mode drop. In
summary, drop, and an operator has
turned it says that making drop probability depend on the size of
the packets on, it is strongly RECOMMENDED that bits happen to be divided into simply encourages the
bits to it SHOULD be divided into smaller packets. Byte-mode drop would
therefore irreversibly complicate any attempt to fix the Internet's
incentive structures.

9. Conclusions

The strong conclusion is turned
off. Note that AQM algorithms such as RED as a whole SHOULD NOT
use be turned off, as without
it, a drop tail queue also biases against large packets. But note
also that turning off byte-mode drop. More generally, may alter the Internet's congestion
notification protocols (drop, ECN & PCN) SHOULD take account relative performance of
applications using different packet size when sizes, so it would be advisable
to establish the implications before turning it off.

5.3. Recommendation on Responding to Congestion

Instead of network equipment biasing its congestion notification is read by for
small packets, the IETF transport layer, NOT
when it is written by the network layer. This approach offers
sufficient and correct area should continue its programme
of updating congestion information for all known and
future transport control protocols to take account of packet
size and also ensures no perverse incentives
are created that would encourage to make transports less sensitive to use inappropriately
small packet sizes. losing control packets
like SYNs and pure ACKS.

5.4. Recommended Future Research

The alternative of deflating RED's drop probability above conclusions cater for smaller
packet sizes (byte-mode drop) has no enduring advantages. It the Internet as it is today with
most, if not all, resources being primarily bit-congestible. A
secondary conclusion of this memo is that we may see more
complex, packet-
congestible resources in the future, so research may be needed to
extend the Internet's congestion notification (drop or ECN) so that
it can handle a mix of bit-congestible and packet-congestible
resources.

6. Security Considerations

This draft recommends that queues do not bias drop probability
towards small packets as this creates the a perverse incentive for
transports to fragment segments break down their flows into tiny pieces and it reopens segments. One of the vulnerability to floods
benefits of small-
packets that drop-tail queues suffered from and implementing AQM was designed meant to
remove. Byte-mode drop is a change be to the network layer remove this perverse
incentive that makes
allowance for an omission from the design of TCP, effectively reverse
engineering the network layer drop-tail queues gave to contrive small packets. Of course, if
transports really want to make two TCPs with
different packet sizes run at equal bit rates (rather than packet
rates) under the same path conditions. It also improves TCP
performance by reducing the chance that a SYN or a pure ACK will be
dropped, because greatest gains, they are small. don't have to
respond to congestion anyway. But we SHOULD NOT hack the network
layer to improve or fix certain transport protocols. No matter how
predominant a transport protocol is (even if it's TCP), don't want applications that
are trying to
correct for its failings behave to discover that they can go faster by biasing using
smaller packets.

In practice, transports cannot all be trusted to respond to
congestion. So another reason for recommending that queues do not
bias drop probability towards small packets in the
network layer creates a perverse incentive is to break down all flows
from all transports into tiny segments.

So far, our survey of 84 vendors across avoid the industry has drawn
responses from about 19%, none
vulnerability to small packet DDoS attacks that would otherwise
result. One of whom have implemented the byte mode
packet drop variant benefits of RED. Given there appears implementing AQM was meant to be little, if
any, installed base it seems to
remove drop-tail's DoS vulnerability to small packets, so we can recommend removal of byte-mode
drop from RED with little, if any, incremental deployment impact.
shouldn't add it back again.

If a vendor has most queues implemented AQM with byte-mode drop, and an operator has
turned it on, it is strongly RECOMMENDED that it SHOULD be turned
off. Note that RED as a whole SHOULD NOT be turned off, as without
it, the resulting
network would amplify the potency of a drop tail small packet DDoS attack. At
the first queue also biases against large packets. But note
also that turning off byte-mode may alter the relative performance stream of
applications using different packet sizes, packets would push aside a greater
proportion of large packets, so it more of the small packets would be advisable
survive to establish the implications before turning it off.

Instead, attack the IETF transport area should continue its programme next queue. Thus a flood of
updating congestion control protocols to take account small packets
would continue on towards the destination, pushing regular traffic
with large packets out of packet size
and to make transports the way in one queue after the next, but
suffering much less sensitive drop itself.

Appendix C explains why the ability of networks to losing control packets like
SYNs and pure ACKS.

NOTE WELL that RED's byte-mode queue measurement is fine, being
completely orthogonal police the
response of _any_ transport to congestion depends on bit-congestible
network resources only doing packet-mode not byte-mode drop. If a RED implementation has
a byte-mode but does not specify what sort of byte-mode, In
summary, it is most
probably byte-mode queue measurement, which is fine. However, if in
doubt, the vendor should be consulted.

The above conclusions cater for says that making drop probability depend on the Internet as it is today with
most, if not all, resources being primarily bit-congestible. A
secondary conclusion size of this memo is
the packets that we may see more packet-
congestible resources in bits happen to be divided into simply encourages the future, so research may
bits to be needed divided into smaller packets. Byte-mode drop would
therefore irreversibly complicate any attempt to
extend fix the Internet's congestion notification (drop or ECN) so that
it can handle a mix of bit-congestible and packet-congestible
resources.

10.
incentive structures.

7. Acknowledgements

Thank you to Sally Floyd, who gave extensive and useful review
comments. Also thanks for the reviews from Philip Eardley, Toby
Moncaster and Arnaud Jacquet as well as helpful explanations of
different hardware approaches from Larry Dunn and Fred Baker. I am
grateful to Bruce Davie and his colleagues for providing a timely and
efficient survey of RED implementation in Cisco's product range.
Also grateful thanks to Toby Moncaster, Will Dormann, John Regnault,
Simon Carter and Stefaan De Cnodder who further helped survey the
current status of RED implementation and deployment and, finally,
thanks to the anonymous individuals who responded.

Bob Briscoe is and Jukka Manner are partly funded by Trilogy, a research
project (ICT- 216372) supported by the European Community under its
Seventh Framework Programme. The views expressed here are those of
the
author authors only.

11.

8. Comments Solicited

Comments and questions are encouraged and very welcome. They can be
addressed to the IETF Transport Area working group mailing list
<tsvwg@ietf.org>, and/or to the authors.

12.

9. References

12.1.

9.1. Normative References

[RFC2119] Bradner, S., "Key words for use in
RFCs to Indicate Requirement Levels",
BCP 14, RFC 2119, March 1997.

[RFC2309] Braden, B., Clark, D., Crowcroft, J.,
Davie, B., Deering, S., Estrin, D.,
Floyd, S., Jacobson, V., Minshall,
G., Partridge, C., Peterson, L.,
Ramakrishnan, K., Shenker, S.,
Wroclawski, J., and L. Zhang,
"Recommendations on Queue Management
and Congestion Avoidance in the
Internet", RFC 2309, April 1998.

[RFC3168] Ramakrishnan, K., Floyd, S., and D.
Black, "The Addition of Explicit
Congestion Notification (ECN) to IP",
RFC 3168, September 2001.

[RFC3426] Floyd, S., "General Architectural and
Policy Considerations", RFC 3426,
November 2002.

[RFC5033] Floyd, S. and M. Allman, "Specifying
New Congestion Control Algorithms",
BCP 133, RFC 5033, August 2007.

12.2.

9.2. Informative References

[CCvarPktSize] Widmer, J., Boutremans, C., and J-Y.
Le Boudec, "Congestion Control for
Flows with Variable Packet Size", ACM
CCR 34(2) 137--151, 2004,
<http://doi.acm.org/10.1145/997150.997162>. <http://
doi.acm.org/10.1145/997150.997162>.

[DRQ] Shin, M., Chong, S., and I. Rhee,
"Dual-Resource TCP/AQM for
Processing-Constrained Networks",
IEEE/ACM Transactions on
Networking Vol 16, issue 2,
April 2008,
<http://dx.doi.org/10.1109/TNET.2007.900415>. <http://dx.doi.org/
10.1109/TNET.2007.900415>.

[DupTCP] Wischik, D., "Short messages", Royal
Society workshop on networks:
modelling and control ,
September 2007, <http://
www.cs.ucl.ac.uk/staff/ucacdjw/Research/shortmsg.html>.
www.cs.ucl.ac.uk/staff/ucacdjw/
Research/shortmsg.html>.

[ECNFixedWireless] Siris, V., "Resource Control for
Elastic Traffic in CDMA Networks",
Proc. ACM MOBICOM'02 ,
September 2002, <http://
www.ics.forth.gr/netlab/publications/
resource_control_elastic_cdma.html>.

[Evol_cc] Gibbens, R. and F. Kelly, "Resource
pricing and the evolution of
congestion control",
Automatica 35(12)1969--
1985, 35(12)1969--1985,
December 1999,
<http://www.statslab.cam.ac.uk/~frank/evol.html>. <http://
www.statslab.cam.ac.uk/~frank/
evol.html>.

[I-D.briscoe-tsvwg-re-ecn-tcp] Briscoe, B., Jacquet, A., Moncaster,
T., and A. Smith, "Re-ECN: Adding
Accountability for Causing Congestion
to TCP/IP", draft-briscoe-tsvwg-re-ecn-tcp-07 (work in
progress), March 2009.

[I-D.floyd-tcpm-ackcc]
Floyd, S., "Adding Acknowledgement Congestion Control to
TCP", draft-floyd-tcpm-ackcc-06
draft-briscoe-tsvwg-re-ecn-tcp-08
(work in progress),
July September 2009.

[I-D.ietf-pcn-marking-behaviour]

[I-D.ietf-pcn] Eardley, P., "Metering and marking
behaviour of PCN-
nodes", PCN-nodes",
draft-ietf-pcn-marking-behaviour-05
(work in progress), August 2009.

[I-D.ietf-tcpm-ecnsyn]
Floyd, S., "Adding Explicit Congestion Notification (ECN)
Capability to TCP's SYN/ACK Packets",
draft-ietf-tcpm-ecnsyn-10 (work in progress), May 2009.

[I-D.irtf-iccrg-welzl-congestion-control-open-research]

[I-D.irtf-iccrg-welzl] Welzl, M., Scharf, M., Briscoe, B.,
and D. Papadimitriou, "Open Research
Issues in Internet Congestion
Control",
draft-irtf-iccrg-welzl-congestion-control-open-research-05 draft-irtf-iccrg-welzl-
congestion-control-open-research-07
(work in progress), September 2009. June 2010.

[IOSArch] Bollapragada, V., White, R., and C.
Murphy, "Inside Cisco IOS Software
Architecture", Cisco Press: CCIE
Professional Development ISBN13: 978-1-57870-181-0, 978-
1-57870-181-0, July 2000.

[MulTCP] Crowcroft, J. and Ph. Oechslin,
"Differentiated End to End Internet
Services using a Weighted
Proportional Fair Sharing TCP",
CCR 28(3) 53--69, July 1998, <http://
www.cs.ucl.ac.uk/staff/J.Crowcroft/hipparch/pricing.html>.
www.cs.ucl.ac.uk/staff/J.Crowcroft/
hipparch/pricing.html>.

[PktSizeEquCC] Vasallo, P., "Variable Packet Size
Equation-Based Congestion Control",
ICSI Technical Report tr-00-008,
2000, <http://http.icsi.berkeley.edu/ftp/global/pub/
techreports/2000/tr-00-008.pdf>. <http://http.icsi.berkeley.edu/
ftp/global/pub/techreports/2000/
tr-00-008.pdf>.

[RED93] Floyd, S. and V. Jacobson, "Random
Early Detection (RED) gateways for
Congestion Avoidance", IEEE/ACM
Transactions on Networking 1(4) 397--413, 397--
413, August 1993,
<http://www.icir.org/floyd/papers/red/red.html>. <http://
www.icir.org/floyd/papers/red/
red.html>.

[REDbias] Eddy, W. and M. Allman, "A Comparison
of RED's Byte and Packet Modes",
Computer Networks 42(3) 261--280,
June 2003,
<http://www.ir.bbn.com/documents/articles/redbias.ps>. <http://www.ir.bbn.com/
documents/articles/redbias.ps>.

[REDbyte] De Cnodder, S., Elloumi, O., and K.
Pauwels, "RED behavior with different
packet sizes", Proc. 5th IEEE
Symposium on Computers and
Communications (ISCC) 793--799,
July 2000,
<http://www.icir.org/floyd/red/Elloumi99.pdf>. <http://www.icir.org/
floyd/red/Elloumi99.pdf>.

[RFC2474] Nichols, K., Blake, S., Baker, F.,
and D. Black, "Definition of the
Differentiated Services Field (DS
Field) in the IPv4 and IPv6 Headers",
RFC 2474, December 1998.

[RFC3448] Handley, M., Floyd, S., Padhye, J.,
and J. Widmer, "TCP Friendly Rate
Control (TFRC): Protocol
Specification", RFC 3448,
January 2003.

[RFC3714] Floyd, S. and J. Kempf, "IAB Concerns
Regarding Congestion Control for
Voice Traffic in the Internet",
RFC 3714, March 2004.

[RFC4782] Floyd, S., Allman, M., Jain, A., and
P. Sarolahti, "Quick-
Start "Quick-Start for TCP
and IP", RFC 4782, January 2007.

[RFC4828] Floyd, S. and E. Kohler, "TCP
Friendly Rate Control (TFRC): The
Small-Packet (SP) Variant", RFC 4828,
April 2007.

[RFC5562] Kuzmanovic, A., Mondal, A., Floyd,
S., and K. Ramakrishnan, "Adding
Explicit Congestion Notification
(ECN) Capability to TCP's SYN/ACK
Packets", RFC 5562, June 2009.

[RFC5670] Eardley, P., "Metering and Marking
Behaviour of PCN-Nodes", RFC 5670,
November 2009.

[RFC5681] Allman, M., Paxson, V., and E.
Blanton, "TCP Congestion Control",
RFC 5681, September 2009.

[RFC5690] Floyd, S., Arcia, A., Ros, D., and J.
Iyengar, "Adding Acknowledgement
Congestion Control to TCP", RFC 5690,
February 2010.

[Rate_fair_Dis] Briscoe, B., "Flow Rate Fairness:
Dismantling a Religion",
ACM CCR 37(2)63--74, April 2007,
<http://portal.acm.org/citation.cfm?id=1232926>.

[WindowPropFair]
Siris, V., "Service Differentiation Religion", ACM
CCR 37(2)63--74, April 2007, <http://
portal.acm.org/
citation.cfm?id=1232926>.

[WindowPropFair] Siris, V., "Service Differentiation
and Performance of Weighted Window-
Based Congestion Control and Packet
Marking Algorithms in ECN Networks",
Computer Communications 26(4) 314--
326, 2002, <http://www.ics.forth.gr/
netgroup/publications/
weighted_window_control.html>.

[gentle_RED] Floyd, S., "Recommendation on using
the "gentle_" variant of RED", Web
page , March 2000, <http://
www.icir.org/floyd/red/gentle.html>.

[pBox] Floyd, S. and K. Fall, "Promoting the
Use of End-to-End Congestion Control
in the Internet", IEEE/ACM
Transactions on Networking 7(4) 458--
472, August 1999, <http://
www.aciri.org/floyd/
end2end-paper.html>.

[pktByteEmail] Yes and J. Doe, "Missing for now",
RFC 0000, May 2006.

[xcp-spec] Falk, A., "Specification for the
Explicit Control Protocol (XCP)",
draft-falk-xcp-spec-03 (work in
progress), July 2007.

Appendix A. Congestion Notification Definition: Further Justification

In Section 1.1 on the definition of congestion notification, load not
capacity was used as the denominator. This also has a subtle
significance in the related debate over the design of new transport
protocols--typical new protocol designs (e.g. in XCP [xcp-spec] &
Quickstart [RFC4782]) expect the sending transport to communicate its
desired flow rate to the network and network elements to
progressively subtract from this so that the achievable flow rate
emerges at the receiving transport.

Congestion notification with total load in the denominator can serve
a similar purpose (though in retrospect not in advance like XCP &
QuickStart). Congestion notification is a dimensionless fraction but
each source can extract necessary rate information from it because it
already knows what its own rate is. Even though congestion
notification doesn't communicate a rate explicitly, from each
source's point of view congestion notification represents the
fraction of the rate it was sending a round trip ago that couldn't
(or wouldn't) be served by available resources.

Appendix B. Idealised Wire Protocol

We will start by inventing an idealised congestion notification
protocol before discussing how to make it practical. The idealised
protocol is shown to be correct using examples later in this
appendix.

B.1. Protocol Coding

Congestion notification involves the congested resource coding a
congestion notification signal into the packet stream and the
transports decoding it. The idealised protocol uses two different
(imaginary) fields in each datagram to signal congestion: one for
byte congestion and one for packet congestion.

We are not saying two ECN fields will be needed (and we are not
saying that somehow a resource should be able to drop a packet in one
of two different ways so that the transport can distinguish which
sort of drop it was!). These two congestion notification channels
are just a conceptual device. They allow us to defer having to
decide whether to distinguish between byte and packet congestion when
the network resource codes the signal or when the transport decodes
it.

However, although this idealised mechanism isn't intended for
implementation, we do want to emphasise that we may need to find a
way to implement it, because it could become necessary to somehow
distinguish between bit and packet congestion [RFC3714]. Currently,
packet-congestion is not the common case, but there is no guarantee
that it will not become common with future technology trends.

The idealised wire protocol is given below. It accounts for packet
sizes at the transport layer, not in the network, and then only in
the case of bit-congestible resources. This avoids the perverse
incentive to send smaller packets and the DoS vulnerability that
would otherwise result if the network were to bias towards them (see
the motivating argument about avoiding perverse incentives in
Section 2.2):

1. A packet-congestible resource trying to code congestion level p_p
into a packet stream should mark the idealised `packet
congestion' field in each packet with probability p_p
irrespective of the packet's size. The transport should then
take a packet with the packet congestion field marked to mean
just one mark, irrespective of the packet size.

2. A bit-congestible resource trying to code time-varying byte-
congestion level p_b into a packet stream should mark the `byte
congestion' field in each packet with probability p_b, again
irrespective of the packet's size. Unlike before, the transport
should take a packet with the byte congestion field marked to
count as a mark on each byte in the packet.

The worked examples in Appendix B.2 show that transports can extract
sufficient and correct congestion notification from these protocols
for cases when two flows with different packet sizes have matching
bit rates or matching packet rates. Examples are also given that mix
these two flows into one to show that a flow with mixed packet sizes
would still be able to extract sufficient and Performance of
Weighted Window-Based Congestion Control correct information.

Sufficient and Packet
Marking Algorithms in ECN Networks", Computer
Communications 26(4) 314--326, 2002, <http://
www.ics.forth.gr/netgroup/publications/
weighted_window_control.html>.

[gentle_RED]
Floyd, S., "Recommendation on using correct congestion information means that there is
sufficient information for the "gentle_" variant two different types of RED", Web page , March 2000,
<http://www.icir.org/floyd/red/gentle.html>.

[pBox] Floyd, S. and K. Fall, "Promoting transport
requirements:

Ratio-based: Established transport congestion controls like TCP's
[RFC5681] aim to achieve equal segment rates per RTT through the Use of End-to-End
Congestion Control in
same bottleneck--TCP friendliness [RFC3448]. They work with the Internet", IEEE/ACM Transactions
on Networking 7(4) 458--472, August 1999,
<http://www.aciri.org/floyd/end2end-paper.html>.

[pktByteEmail]
Floyd, S., "RED: Discussions
ratio of Byte and Packet Modes",
email , March 1997,
<http://www-nrg.ee.lbl.gov/floyd/REDaveraging.txt>.

[xcp-spec]
Falk, A., "Specification for the Explicit Control Protocol
(XCP)", draft-falk-xcp-spec-03 (work dropped to delivered segments (or marked to unmarked
segments in progress),
July 2007.

(Expired)

Editorial Comments

[Note_Variation] The algorithm of the byte-mode drop variant case of RED
switches off any bias towards small packets
whenever the smoothed queue length dictates ECN). The example scenarios show that
these ratio-based transports are effectively the drop probability of large packets should be
100%. In the example same whether
counting in bytes or packets, because the Introduction, as the
large units cancel out.
(Incidentally, this is why TCP's bit rate is still proportional to
packet drop probability varies around 25% size even when byte-counting is used, as recommended for
TCP in [RFC5681], mainly for orthogonal security reasons.)

Absolute-target-based: Other congestion controls proposed in the
small packet drop probability will vary around 1%,
but with occasional jumps
research community aim to 100% whenever limit the
instantaneous queue (after drop) manages volume of congestion caused to sustain
a length above the 100% drop point constant weight parameter. [MulTCP][WindowPropFair] are
examples of weighted proportionally fair transports designed for longer than
cost-fair environments [Rate_fair_Dis]. In this case, the queue averaging period.

Appendix A.
transport requires a count (not a ratio) of dropped/marked bytes
in the bit-congestible case and of dropped/marked packets in the
packet congestible case.

B.2. Example Scenarios

A.1.

B.2.1. Notation

To prove our idealised wire protocol (Section 5) (Appendix B.1) is correct, we
will compare two flows with different packet sizes, s_1 and s_2 [bit/pkt], [bit/
pkt], to make sure their transports each see the correct congestion
notification. Initially, within each flow we will take all packets
as having equal sizes, but later we will generalise to flows within
which packet sizes vary. A flow's bit rate, x [bit/s], is related to
its packet rate, u [pkt/s], by

x(t) = s.u(t).

We will consider a 2x2 matrix of four scenarios:

+-----------------------------+------------------+------------------+
| resource type and | A) Equal bit | B) Equal pkt |
| congestion level | rates | rates |
+-----------------------------+------------------+------------------+
| i) bit-congestible, p_b | (Ai) | (Bi) |
| ii) pkt-congestible, p_p | (Aii) | (Bii) |
+-----------------------------+------------------+------------------+

Table 3

A.2.

B.2.2. Bit-congestible resource, equal bit rates (Ai)

Starting with the bit-congestible scenario, for two flows to maintain
equal bit rates (Ai) the ratio of the packet rates must be the
inverse of the ratio of packet sizes: u_2/u_1 = s_1/s_2. So, for
instance, a flow of 60B packets would have to send 25x more packets
to achieve the same bit rate as a flow of 1500B packets. If a
congested resource marks proportion p_b of packets irrespective of
size, the ratio of marked packets received by each transport will
still be the same as the ratio of their packet rates, p_b.u_2/p_b.u_1
= s_1/s_2. So of the 25x more 60B packets sent, 25x more will be
marked than in the 1500B packet flow, but 25x more won't be marked
too.

In this scenario, the resource is bit-congestible, so it always uses
our idealised bit-congestion field when it marks packets. Therefore
the transport should count marked bytes not packets. But it doesn't
actually matter for ratio-based transports like TCP (Section 5). (Appendix B.1).
The ratio of marked to unmarked bytes seen by each flow will be p_b,
as will the ratio of marked to unmarked packets. Because they are
ratios, the units cancel out.

If a flow sent an inconsistent mixture of packet sizes, we have said
it should count the ratio of marked and unmarked bytes not packets in
order to correctly decode the level of congestion. But actually, if
all it is trying to do is decode p_b, it still doesn't matter. For
instance, imagine the two equal bit rate flows were actually one flow
at twice the bit rate sending a mixture of one 1500B packet for every
thirty 60B packets. 25x more small packets will be marked and 25x
more will be unmarked. The transport can still calculate p_b whether
it uses bytes or packets for the ratio. In general, for any
algorithm which works on a ratio of marks to non-marks, either bytes
or packets can be counted interchangeably, because the choice cancels
out in the ratio calculation.

However, where an absolute target rather than relative volume of
congestion caused is important (Section 5), (Appendix B.1), as it is for
congestion accountability [Rate_fair_Dis], the transport must count
marked bytes not packets, in this bit-congestible case. Aside from
the goal of congestion accountability, this is how the bit rate of a
transport can be made independent of packet size; by ensuring the
rate of congestion caused is kept to a constant weight
[WindowPropFair], rather than merely responding to the ratio of
marked and unmarked bytes.

Note the unit of byte-congestion-volume is the byte.

A.3.

B.2.3. Bit-congestible resource, equal packet rates (Bi)

If two flows send different packet sizes but at the same packet rate,
their bit rates will be in the same ratio as their packet sizes, x_2/
x_1 = s_2/s_1. For instance, a flow sending 1500B packets at the
same packet rate as another sending 60B packets will be sending at
25x greater bit rate. In this case, if a congested resource marks
proportion p_b of packets irrespective of size, the ratio of packets
received with the byte-congestion field marked by each transport will
be the same, p_b.u_2/p_b.u_1 = 1.

Because the byte-congestion field is marked, the transport should
count marked bytes not packets. But because each flow sends
consistently sized packets it still doesn't matter for ratio-based
transports. The ratio of marked to unmarked bytes seen by each flow
will be p_b, as will the ratio of marked to unmarked packets.
Therefore, if the congestion control algorithm is only concerned with
the ratio of marked to unmarked packets (as is TCP), both flows will
be able to decode p_b correctly whether they count packets or bytes.

But if the absolute volume of congestion is important, e.g. for
congestion accountability, the transport must count marked bytes not
packets. Then the lower bit rate flow using smaller packets will
rightly be perceived as causing less byte-congestion even though its
packet rate is the same.

If the two flows are mixed into one, of bit rate x1+x2, with equal
packet rates of each size packet, the ratio p_b will still be
measurable by counting the ratio of marked to unmarked bytes (or
packets because the ratio cancels out the units). However, if the
absolute volume of congestion is required, the transport must count
the sum of congestion marked bytes, which indeed gives a correct
measure of the rate of byte-congestion p_b(x_1 + x_2) caused by the
combined bit rate.

A.4.

B.2.4. Pkt-congestible resource, equal bit rates (Aii)

Moving to the case of packet-congestible resources, we now take two
flows that send different packet sizes at the same bit rate, but this
time the pkt-congestion field is marked by the resource with
probability p_p. As in scenario Ai with the same bit rates but a
bit-congestible resource, the flow with smaller packets will have a
higher packet rate, so more packets will be both marked and unmarked,
but in the same proportion.

This time, the transport should only count marks without taking into
account packet sizes. Transports will get the same result, p_p, by
decoding the ratio of marked to unmarked packets in either flow.

If one flow imitates the two flows but merged together, the bit rate
will double with more small packets than large. The ratio of marked
to unmarked packets will still be p_p. But if the absolute number of
pkt-congestion marked packets is counted it will accumulate at the
combined packet rate times the marking probability, p_p(u_1+u_2), 26x
faster than packet congestion accumulates in the single 1500B packet
flow of our example, as required.

But if the transport is interested in the absolute number of packet
congestion, it should just count how many marked packets arrive. For
instance, a flow sending 60B packets will see 25x more marked packets
than one sending 1500B packets at the same bit rate, because it is
sending more packets through a packet-congestible resource.

Note the unit of packet congestion is a packet.

A.5.

B.2.5. Pkt-congestible resource, equal packet rates (Bii)

Finally, if two flows with the same packet rate, pass through a
packet-congestible resource, they will both suffer the same
proportion of marking, p_p, irrespective of their packet sizes. On
detecting that the pkt-congestion field the pkt-congestion field is marked, the transport
should count packets, and it will be able to extract the ratio p_p of
marked to unmarked packets from both flows, irrespective of packet
sizes.

Even if the transport is monitoring the absolute amount of packets
congestion over a period, still it will see the same amount of packet
congestion from either flow.

And if the two equal packet rates of different size packets are mixed
together in one flow, the packet rate will double, so the absolute
volume of packet-congestion will accumulate at twice the rate of
either flow, 2p_p.u_1 = p_p(u_1+u_2).

Appendix C. Byte-mode Drop Complicates Policing Congestion Response

This appendix explains why the ability of networks to police the
response of _any_ transport to congestion depends on bit-congestible
network resources only doing packet-mode not byte-mode drop.

To be able to police a transport's response to congestion when
fairness can only be judged over time and over all an individual's
flows, the policer has to have an integrated view of all the
congestion an individual (not just one flow) has caused due to all
traffic entering the Internet from that individual. This is marked, the transport
should count packets, and it will be able termed
congestion accountability.

But a byte-mode drop algorithm has to extract depend on the ratio p_p local MTU of
marked the
line - an algorithm needs to unmarked packets from both flows, irrespective use some concept of a 'normal' packet
size. Therefore, one dropped or marked packet
sizes.

Even if the transport is monitoring not necessarily
equivalent to another unless you know the absolute amount MTU at the queue where it
was dropped/marked. To have an integrated view of packets
congestion over a period, still user, we believe
congestion policing has to be located at an individual's attachment
point to the Internet [I-D.briscoe-tsvwg-re-ecn-tcp]. But from there
it will see cannot know the same amount MTU of packet each remote queue that caused each drop/
mark. Therefore it cannot take an integrated approach to policing
all the responses to congestion from either flow.

And if of all the two equal packet rates transports of different size packets are mixed
together in one flow, the packet rate will double, so
individual. Therefore it cannot police anything.

The security/incentive argument _for_ packet-mode drop is similar.
Firstly, confining RED to packet-mode drop would not preclude
bottleneck policing approaches such as [pBox] as it seems likely they
could work just as well by monitoring the absolute volume of packet-congestion will accumulate at twice dropped bytes
rather than packets. Secondly packet-mode dropping/marking naturally
allows the rate congestion notification of
either flow, 2p_p.u_1 = p_p(u_1+u_2).

Appendix B. Congestion Notification Definition: Further Justification

In Section 3 packets to be globally
meaningful without relying on MTU information held elsewhere.

Because we recommend that a dropped/marked packet should be taken to
mean that all the definition of congestion notification, load not
capacity was used as bytes in the denominator. This also has packet are dropped/marked, a subtle
significance policer
can remain robust against bits being re-divided into different size
packets or across different size flows [Rate_fair_Dis]. Therefore
policing would work naturally with just simple packet-mode drop in
RED.

In summary, making drop probability depend on the related debate over the design size of new transport
protocols--typical new protocol designs (e.g. in XCP [xcp-spec] &
Quickstart [RFC4782]) expect the sending transport to communicate its
desired flow rate packets
that bits happen to be divided into simply encourages the network and network elements bits to
progressively subtract be
divided into smaller packets. Byte-mode drop would therefore
irreversibly complicate any attempt to fix the Internet's incentive
structures.

Appendix D. Changes from this so that Previous Versions

To be removed by the achievable flow rate
emerges RFC Editor on publication.

Congestion notification with total load in rfcdiff tool):

From -01 to -02 (this version):

* Restructured the denominator can serve
a similar purpose (though in retrospect not whole document for (hopefully) easier reading
and clarity. The concrete recommendation, in advance like XCP &
QuickStart). Congestion notification RFC2119 language,
is a dimensionless fraction but
each source can extract necessary rate information from it because it
already knows what its own rate is. Even though congestion
notification doesn't communicate a rate explicitly, from each
source's point of view congestion notification represents now in Section 5.

From -00 to -01:

* Minor clarifications throughout and updated references

From briscoe-byte-pkt-mark-02 to ietf-byte-pkt-congest-00:

* Added note on relationship to existing RFCs
* Posed the
fraction question of the rate whether packet-congestion could become
common and deferred it was sending a round trip ago that couldn't
(or wouldn't) be served by available resources. After they were
sent, all these fractions of each source's offered load added up to the aggregate fraction of offered load seen by the congested
resource. So, the source can also know the total excess rate by
multiplying total load by congestion level. Therefore congestion
notification, as one scale-free dimensionless fraction, implicitly
communicates IRTF ICCRG. Added ref to the instantaneous excess flow rate, albeit a RTT ago.

Appendix C. Byte-mode Drop Complicates Policing Congestion Response

This appendix explains why
dual-resource queue (DRQ) proposal.

* Changed PCN references from the ability of networks PCN charter & architecture to police
the
response of _any_ transport PCN marking behaviour draft most likely to congestion depends on bit-congestible
network resources only doing packet-mode not byte-mode drop.

To be able imminently
become the standards track WG item.

From -01 to police a transport's response -02:

* Abstract reorganised to congestion when
fairness can only be judged over time and over all an individual's
flows, align with clearer separation of issue
in the policer has memo.

* Introduction reorganised with motivating arguments removed to have an integrated view
new Section 2.

* Clarified avoiding lock-out of all large packets is not the
congestion an individual (not just one flow) has caused due main or
only motivation for RED.

* Mentioned choice of drop or marking explicitly throughout,
rather than trying to all
traffic entering the Internet from that individual. This is termed
congestion accountability.

But coin a byte-mode drop algorithm has word to depend on the local MTU of mean either.

* Generalised the
line - an algorithm needs discussion throughout to use some concept of a 'normal' packet
size. Therefore, one dropped or marked any packet is forwarding
function on any network equipment, not necessarily
equivalent to another unless you know the MTU at just routers.

* Clarified the queue that where
it was dropped/marked. To have an integrated view of last point about why this is a user, we
believe congestion policing has good time to sort
out this issue: because it will be located at an individual's
attachment point hard / impossible to design
new transports unless we decide whether the Internet [I-D.briscoe-tsvwg-re-ecn-tcp]. But
from there it cannot know network or the MTU
transport is allowing for packet size.

* Added statement explaining the horizon of each remote queue that caused
each drop/mark. Therefore it cannot take an integrated approach to
policing all the responses to memo is long
term, but with short term expediency in mind.

* Added material on scaling congestion control with packet size
(Section 2.1).

* Separated out issue of all the transports normalising TCP's bit rate from issue of one
individual. Therefore it cannot police anything.

The security/incentive argument _for_ packet-mode
preference to control packets (Section 2.3).

* Divided up Congestion Measurement section for clarity,
including new material on fixed size packet buffers and buffer
carving (Section 3.1.1 & Section 3.2.1) and on congestion
measurement in wireless link technologies without queues
(Section 3.1.2).

* Added section on 'Making Transports Robust against Control
Packet Losses' (Section 3.2.3) with existing & new material
included.

* Added tabulated results of vendor survey on byte-mode drop is similar.
Firstly, confining
variant of RED (Table 2).

From -00 to -01:

* Clarified applicability to packet-mode drop would not preclude
bottleneck policing approaches such as [pBox] as it seems likely they
could work just as well by monitoring the volume of dropped bytes
rather than packets. Secondly packet-mode dropping/marking naturally
allows the congestion notification of packets to be globally
meaningful without relying on MTU information held elsewhere.

Because we recommend as ECN.

* Highlighted DoS vulnerability.

* Emphasised that a dropped/marked packet drop-tail suffers from similar problems to
byte-mode drop, so only byte-mode drop should be taken to
mean that all the bytes in turned off,
not RED itself.

* Clarified the packet are dropped/marked, a policer
can remain robust against bits being re-divided into different size
packets or across different size flows [Rate_fair_Dis]. Therefore
policing would work naturally with just simple packet-mode drop in
RED.

In summary, making original apparent motivations for recommending
byte-mode drop probability depend on included protecting SYNs and pure ACKs more than
equalising the size bit rates of the packets
that bits happen TCPs with different segment sizes.
Removed some conjectured motivations.

* Added support for updates to be divided TCP in progress (ackcc & ecn-syn-
ack).

* Updated survey results with newly arrived data.

* Pulled all recommendations together into simply encourages the bits to be
divided conclusions.

* Moved some detailed points into smaller packets. Byte-mode drop would therefore
irreversibly complicate any attempt to fix the Internet's incentive
structures.

Author's Address two additional appendices and a
note.

* Considerable clarifications throughout.

* Updated references

Authors' Addresses

Bob Briscoe
BT
B54/77, Adastral Park
Martlesham Heath
Ipswich IP5 3RE
UK

Phone: +44 1473 645196
Email:
EMail: bob.briscoe@bt.com
URI: http://bobbriscoe.net/
Jukka Manner
Aalto University
Department of Communications and Networking (Comnet)
P.O. Box 13000
FIN-00076 Aalto
Finland

Phone: +358 9 470 22481
EMail: jukka.manner@tkk.fi
URI: http://www.netlab.tkk.fi/~jmanner/