Internet Engineering Task Force K. K. Ramakrishnan
INTERNET DRAFT TeraOptic Networks
draft-ietf-tsvwg-ecn-00.txt Sally Floyd
ACIRI
D. Black
EMC
November, 2000
Expires: May, 2001
The Addition of Explicit Congestion Notification (ECN) to IP
Status of this Memo
This document is an Internet-Draft and is in full conformance with
all provisions of Section 10 of RFC2026.
Internet-Drafts are working documents of the Internet Engineering
Task Force (IETF), its areas, and its working groups. Note that
other groups may also distribute working documents as Internet-
Drafts.
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.
Abstract
This document specifies the incorporation of ECN (Explicit Congestion
Notification) to TCP and IP, including ECN's use of two bits in the
IP header's DS field. We begin by describing TCP's use of packet
drops as an indication of congestion. Next we explain that with the
addition of active queue management (e.g., RED) to the Internet
infrastructure, where routers detect congestion before the queue
overflows, routers are no longer limited to packet drops as an
indication of congestion. Routers can instead set the Congestion
Experienced (CE) bit in the IP header of packets from ECN-capable
transports. We describe when the CE bit is to be set in routers, and
Ramakrishnan and Floyd Proposed Standard [Page 1]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
describe modifications needed to TCP to make it ECN-capable.
Modifications to other transport protocols (e.g., unreliable unicast
or multicast, reliable multicast, other reliable unicast transport
protocols) could be considered as those protocols are developed and
advance through the standards process.
We also describe in this document the issues involving the use of ECN
within IP tunnels, and within IPsec tunnels in particular.
One of the guiding principles for this document is that all the
mechanisms specified here are incrementally deployable.
Table of Contents
Ramakrishnan and Floyd Proposed Standard [Page 2]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
1. Introduction
2. Conventions and Acronyms
3. Assumptions and General Principles
4. Active Queue Management (AQM)
5. Explicit Congestion Notification in IP
5.1. ECN as an indication of persistent congestion
5.2. Dropped or Corrupted Packets
6. Support from the Transport Protocol
6.1. TCP
6.1.1. TCP Initialization
6.1.1.1. Robust TCP Initialization with an Echoed Reserve Field
6.1.1.2. Robust TCP Initialization with no response to the SYN
6.1.2. The TCP Sender
6.1.3. The TCP Receiver
6.1.4. Congestion on the ACK-path
6.1.5. Retransmitted TCP packets
6.1.6. TCP Window Probes.
7. Non-compliance by the End Nodes
8. Non-compliance in the Network
8.1. Complications Introduced by Split Paths
9. Encapsulated Packets
9.1. IP packets encapsulated in IP
9.1.1. The limited-functionality and full-functionality options within
9.1.2. Changes to the ECN Field within an IP Tunnel.
9.2. IPsec Tunnels
9.2.1. Negotiation between Tunnel Endpoints
9.2.1.1. ECN Tunnel Security Association Database Field
9.2.1.2. ECN Tunnel Security Association Attribute
9.2.1.3. Changes to IPsec Tunnel Header Processing
9.2.2. Changes to the ECN Field within an IPsec Tunnel.
9.2.3. Comments for IPsec Support
9.3. IP packets encapsulated in non-IP packet headers.
10. Issues Raised by Monitoring and Policing Devices
11. Evaluations of ECN
12. Summary of changes required in IP and TCP
13. Conclusions
14. Acknowledgements
15. References
16. Security Considerations
17. IPv4 Header Checksum Recalculation
18. Possible Changes to the ECN Field in the Network
18.1. Possible Changes to the IP Header
18.1.1. Erasing the Congestion Indication
18.1.2. Falsely Reporting Congestion
18.1.3. Disabling ECN-Capability
18.1.4. Falsely Indicating ECN-Capability
18.1.5. Changes with No Functional Effect
18.2. Information carried in the Transport Header
Ramakrishnan and Floyd Proposed Standard [Page 3]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
18.3. Split Paths
19. Implications of Subverting End-to-End Congestion Control
19.1. Implications for the Network and for Competing Flows
19.2. Implications for the Subverted Flow
19.3. Non-ECN-Based Methods of Subverting End-to-end Congestion Control
20. The motivation for the ECT bit.
21. Why use two bits in the IP header?
22. Historical definitions for the IPv4 TOS octet
1. Introduction
TCP's congestion control and avoidance algorithms are based on the
notion that the network is a black-box [Jacobson88, Jacobson90]. The
network's state of congestion or otherwise is determined by end- sys-
tems probing for the network state, by gradually increasing the load
on the network (by increasing the window of packets that are out-
standing in the network) until the network becomes congested and a
packet is lost. Treating the network as a "black-box" and treating
loss as an indication of congestion in the network is appropriate for
pure best-effort data carried by TCP, with little or no sensitivity
to delay or loss of individual packets. In addition, TCP's conges-
tion management algorithms have techniques built-in (such as Fast
Retransmit and Fast Recovery) to minimize the impact of losses, from
a throughput perspective. However, these mechanisms are not intended
to help applications that are in fact sensitive to the delay or loss
of one or more individual packets. Interactive traffic such as tel-
net, web-browsing, and transfer of audio and video data can be sensi-
tive to packet losses (especially when using an unreliable data
delivery transport such as UDP) or to the increased latency of the
packet caused by the need to retransmit the packet after a loss (with
the reliable data delivery semantics provided by TCP).
Since TCP determines the appropriate congestion window to use by
gradually increasing the window size until it experiences a dropped
packet, this causes the queues at the bottleneck router to build up.
With most packet drop policies at the router that are not sensitive
to the load placed by each individual flow (e.g., tail-drop on queue
overflow), this means that some of the packets of latency-sensitive
flows may be dropped. In addition, such drop policies lead to syn-
chronization of loss across multiple flows.
Active queue management mechanisms detect congestion before the queue
overflows, and provide an indication of this congestion to the end
nodes. Thus, active queue management can reduce unnecessary queueing
delay for all traffic sharing that queue. The advantages of active
queue management are discussed in RFC 2309 [RFC2309]. Active queue
management avoids some of the bad properties of dropping on queue
Ramakrishnan and Floyd Proposed Standard [Page 4]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
overflow, including the undesirable synchronization of loss across
multiple flows. More importantly, active queue management means that
transport protocols with mechanisms for congestion control (e.g.,
TCP) do not have to rely on buffer overflow as the only indication of
congestion.
Active queue management mechanisms may use one of several methods for
indicating congestion to end-nodes. One is to use packet drops, as is
currently done. However, active queue management allows the router to
separate policies of queueing or dropping packets from the policies
for indicating congestion. Thus, active queue management allows
routers to use the Congestion Experienced (CE) bit in a packet header
as an indication of congestion, instead of relying solely on packet
drops. This has the potential of reducing the impact of loss on
latency-sensitive flows.
2. Conventions and Acronyms
The keywords MUST, MUST NOT, REQUIRED, SHALL, SHALL NOT, SHOULD,
SHOULD NOT, RECOMMENDED, MAY, and OPTIONAL, when they appear in this
document, are to be interpreted as described in [B97].
3. Assumptions and General Principles
In this section, we describe some of the important design principles
and assumptions that guided the design choices in this proposal.
* Because ECN is likely to be adopted gradually, accommodating migra-
tion is essential. Some routers may still only drop packets to indi-
cate congestion, and some end-systems may not be ECN- capable. The
most viable strategy is one that accommodates incremental deployment
without having to resort to "islands" of ECN-capable and non-ECN-
capable environments.
* New mechanisms for congestion control and avoidance need to co-
exist and cooperate with existing mechanisms for congestion control.
In particular, new mechanisms have to co-exist with TCP's current
methods of adapting to congestion and with routers' current practice
of dropping packets in periods of congestion.
* Congestion may persist over different time-scales. The time scales
that we are concerned with are congestion events that may last longer
than a round-trip time.
* The number of packets in an individual flow (e.g., TCP connection
or an exchange using UDP) may range from a small number of packets to
quite a large number. We are interested in managing the congestion
caused by flows that send enough packets so that they are still
active when network feedback reaches them.
* Asymmetric routing is likely to be a normal occurrence in the
Internet. The path (sequence of links and routers) followed by data
Ramakrishnan and Floyd Proposed Standard [Page 5]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
packets may be different from the path followed by the acknowledgment
packets in the reverse direction.
* Many routers process the "regular" headers in IP packets more effi-
ciently than they process the header information in IP options. This
suggests keeping congestion experienced information in the regular
headers of an IP packet.
* It must be recognized that not all end-systems will cooperate in
mechanisms for congestion control. However, new mechanisms shouldn't
make it easier for TCP applications to disable TCP congestion con-
trol. The benefit of lying about participating in new mechanisms
such as ECN-capability should be small.
4. Active Queue Management (AQM)
Random Early Detection (RED) is one mechanism for Active Queue Man-
agement (AQM) that has been proposed to detect incipient congestion
[FJ93], and is currently being deployed in the Internet [RFC2309].
AQM is meant to be a general mechanism using one of several alterna-
tives for congestion indication, but in the absence of ECN, AQM is
restricted to using packet drops as a mechanism for congestion indi-
cation. AQM drops packets based on the average queue length exceed-
ing a threshold, rather than only when the queue overflows. However,
because AQM may drop packets before the queue actually overflows, AQM
is not always forced by memory limitations to discard the packet.
AQM can set a Congestion Experienced (CE) bit in the packet header
instead of dropping the packet, when such a bit is provided in the IP
header and understood by the transport protocol. The use of the CE
bit with ECN allows the receiver(s) to receive the packet, avoiding
the potential for excessive delays due to retransmissions after
packet losses. We use the term 'CE packet' to denote a packet that
has the CE bit set.
5. Explicit Congestion Notification in IP
This document specifies that the Internet provide a congestion indi-
cation for incipient congestion (as in RED and earlier work [RJ90])
where the notification can sometimes be through marking packets
rather than dropping them. This uses an ECN field in the IP header
with two bits. The ECN-Capable Transport (ECT) bit is set by the
data sender to indicate that the end-points of the transport protocol
are ECN-capable. The CE bit is set by the router to indicate conges-
tion to the end nodes. Routers that have a packet arriving at a full
queue drop the packet, just as they do in the absence of ECN.
Bits 6 and 7 in the IPv4 TOS octet are designated as the ECN field.
Bit 6 is designated as the ECT bit, and bit 7 is designated as the CE
bit. The IPv4 TOS octet corresponds to the Traffic Class octet in
Ramakrishnan and Floyd Proposed Standard [Page 6]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
IPv6. The definitions for the IPv4 TOS octet [RFC791] and the IPv6
Traffic Class octet have been superseded by the DS (Differentiated
Services) Field [RFC2474]. Bits 6 and 7 are listed in [RFC2474] as
Currently Unused. Section 19 gives a brief history of the TOS octet.
0 1 2 3 4 5 6 7
+-----+-----+-----+-----+-----+-----+-----+-----+
| | ECN FIELD |
| DSCP | |
| | ECT | CE |
+-----+-----+-----+-----+-----+-----+-----+-----+
DSCP: differentiated services codepoint
ECN: Explicit Congestion Notification
Figure 1: The Differentiated Services Field in IP.
Because of the unstable history of the TOS octet, the use of the ECN
field as specified in this document cannot be guaranteed to be back-
wards compatible with all past uses of these two bits. The potential
dangers of this lack of backwards compatibility are discussed in Sec-
tion 19.
Upon the receipt by an ECN-Capable transport of a single CE packet,
the congestion control algorithms followed at the end-systems MUST be
essentially the same as the congestion control response to a *single*
dropped packet. For example, for ECN-Capable TCP the source TCP is
required to halve its congestion window for any window of data con-
taining either a packet drop or an ECN indication.
One reason for requiring that the congestion-control response to the
CE packet be essentially the same as the response to a dropped packet
is to accommodate the incremental deployment of ECN in both end-sys-
tems and in routers. Some routers may drop ECN-Capable packets
(e.g., using the same AQM policies for congestion detection) while
other routers set the CE bit, for equivalent levels of congestion.
Similarly, a router might drop a non-ECN-Capable packet but set the
CE bit in an ECN-Capable packet, for equivalent levels of congestion.
If there were different congestion control responses to a CE bit
indication than to a packet drop, this could result in unfair treat-
ment for different flows.
An additional goal is that the end-systems should react to congestion
at most once per window of data (i.e., at most once per round-trip
time), to avoid reacting multiple times to multiple indications of
congestion within a round-trip time.
For a router, the CE bit of an ECN-Capable packet should only be set
Ramakrishnan and Floyd Proposed Standard [Page 7]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
if the router would otherwise have dropped the packet as an indica-
tion of congestion to the end nodes. When the router's buffer is not
yet full and the router is prepared to drop a packet to inform end
nodes of incipient congestion, the router should first check to see
if the ECT bit is set in that packet's IP header. If so, then
instead of dropping the packet, the router MAY instead set the CE bit
in the IP header.
An environment where all end nodes were ECN-Capable could allow new
criteria to be developed for setting the CE bit, and new congestion
control mechanisms for end-node reaction to CE packets. However,
this is a research issue, and as such is not addressed in this docu-
ment.
When a CE packet (i.e., a packet that has the CE bit set) is received
by a router, the CE bit is left unchanged, and the packet is trans-
mitted as usual. When severe congestion has occurred and the router's
queue is full, then the router has no choice but to drop some packet
when a new packet arrives. We anticipate that such packet losses
will become relatively infrequent when a majority of end-systems
become ECN- Capable and participate in TCP or other compatible con-
gestion control mechanisms. In an ECN-Capable environment that is
adequately-provisioned network, packet losses should occur primarily
during transients or in the presence of non-cooperating sources.
We expect that routers will set the CE bit in response to incipient
congestion as indicated by the average queue size, using the RED
algorithms suggested in [FJ93, RFC2309]. To the best of our knowl-
edge, this is the only proposal currently under discussion in the
IETF for routers to drop packets proactively, before the buffer over-
flows. However, this document does not attempt to specify a particu-
lar mechanism for active queue management, leaving that endeavor, if
needed, to other areas of the IETF. While ECN is inextricably tied
up with the need to have a reasonable active queue management mecha-
nism at the router, the reverse does not hold; active queue manage-
ment mechanisms have been developed and deployed independent of ECN,
using packet drops as indications of congestion in the absence of ECN
in the IP architecture.
5.1. ECN as an indication of persistent congestion
We emphasize that a *single* packet with the CE bit set in an IP
packet causes the transport layer to respond, in terms of congestion
control, as it would to a packet drop. The instantaneous queue size
is likely to see considerable variations even when the router does
not experience persistent congestion. As such, it is important that
transient congestion at a router, reflected by the instantaneous
queue size reaching a threshold much smaller than the capacity of the
Ramakrishnan and Floyd Proposed Standard [Page 8]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
queue, not trigger a reaction at the transport layer. Therefore, the
CE bit should not be set by a router based on the instantaneous queue
size.
For example, since the ATM and Frame Relay mechanisms for congestion
indication have typically been defined without an associated notion
of average queue size as the basis for determining that an intermedi-
ate node is congested, we believe that they provide a very noisy sig-
nal. The TCP-sender reaction specified in this document for ECN is
NOT the appropriate reaction for such a noisy signal of congestion
notification. However, if the routers that interface to the ATM net-
work have a way of maintaining the average queue at the interface,
and use it to come to a reliable determination that the ATM subnet is
congested, they may use the ECN notification that is defined here.
We continue to encourage experiments in techniques at layer 2 (e.g.,
in ATM switches or Frame Relay switches) to take advantage of ECN.
For example, using a scheme such as RED (where packet marking is
based on the average queue length exceeding a threshold), layer 2
devices could provide a reasonably reliable indication of congestion.
When all the layer 2 devices in a path set that layer's own Conges-
tion Experienced bit (e.g., the EFCI bit for ATM, the FECN bit in
Frame Relay) in this reliable manner, then the interface router to
the layer 2 network could copy the state of that layer 2 Congestion
Experienced bit into the CE bit in the IP header. We recognize that
this is not the current practice, nor is it in current standards.
However, encouraging experimentation in this manner may provide the
information needed to enable evolution of existing layer 2 mechanisms
to provide a more reliable means of congestion indication, when they
use a single bit for indicating congestion.
5.2. Dropped or Corrupted Packets
For the proposed use for ECN in this document (that is, for a trans-
port protocol such as TCP for which a dropped data packet is an indi-
cation of congestion), end nodes detect dropped data packets, and the
congestion response of the end nodes to a dropped data packet is at
least as strong as the congestion response to a received CE packet.
To ensure the reliable delivery of the congestion indication of the
CE bit, the ECT bit MUST NOT be set in a packet unless the loss of
that packet in the network would be detected by the end nodes and
interpreted as an indication of congestion.
Transport protocols such as TCP do not necessarily detect all packet
drops, such as the drop of a "pure" ACK packet; for example, TCP does
not reduce the arrival rate of subsequent ACK packets in response to
an earlier dropped ACK packet. Any proposal for extending ECN-Capa-
bility to such packets would have to address issues such as the case
Ramakrishnan and Floyd Proposed Standard [Page 9]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
of an ACK packet that was marked with the CE bit but was later
dropped in the network. We believe that this aspect is still the sub-
ject of research, so this document specifies that at this time,
"pure" ACK packets MUST NOT indicate ECN-Capability.
Similarly, if a CE packet is dropped later in the network due to cor-
ruption (bit errors), the end nodes should still invoke congestion
control, just as TCP would today in response to a dropped data
packet. This issue of corrupted CE packets would have to be consid-
ered in any proposal for the network to distinguish between packets
dropped due to corruption, and packets dropped due to congestion or
buffer overflow. In particular, the ubiquitous deployment of ECN
would not, in and of itself, be a sufficient development to allow
end-nodes to interpret packet drops as indications of corruption
rather than congestion.
6. Support from the Transport Protocol
ECN requires support from the transport protocol, in addition to the
functionality given by the ECN field in the IP packet header. The
transport protocol might require negotiation between the endpoints
during setup to determine that all of the endpoints are ECN-capable,
so that the sender can set the ECT bit in transmitted packets. Sec-
ond, the transport protocol must be capable of reacting appropriately
to the receipt of CE packets. This reaction could be in the form of
the data receiver informing the data sender of the received CE packet
(e.g., TCP), of the data receiver unsubscribing to a layered multi-
cast group (e.g., RLM [MJV96]), or of some other action that ulti-
mately reduces the arrival rate of that flow on that congested link.
This document only addresses the addition of ECN Capability to TCP,
leaving issues of ECN in other transport protocols to further
research. For TCP, ECN requires three new pieces of functionality:
negotiation between the endpoints during connection setup to deter-
mine if they are both ECN-capable; an ECN-Echo (ECE) flag in the TCP
header so that the data receiver can inform the data sender when a CE
packet has been received; and a Congestion Window Reduced (CWR) flag
in the TCP header so that the data sender can inform the data
receiver that the congestion window has been reduced. The support
required from other transport protocols is likely to be different,
particularly for unreliable or reliable multicast transport proto-
cols, and will have to be determined as other transport protocols are
brought to the IETF for standardization.
6.1. TCP
The following sections describe in detail the proposed use of ECN in
TCP. This proposal is described in essentially the same form in
Ramakrishnan and Floyd Proposed Standard [Page 10]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
[Floyd94]. We assume that the source TCP uses the standard congestion
control algorithms of Slow-start, Fast Retransmit and Fast Recovery
[RFC 2001].
This proposal specifies two new flags in the Reserved field of the
TCP header. The TCP mechanism for negotiating ECN-Capability uses
the ECN-Echo (ECE) flag in the TCP header. Bit 9 in the Reserved
field of the TCP header is designated as the ECN-Echo flag. The
location of the 6-bit Reserved field in the TCP header is shown in
Figure 3 of RFC 793 [RFC793] (and is reproduced below for complete-
ness). This specification of the ECN Field leaves the Reserved field
as a 4-bit field using bits 4-7.
To enable the TCP receiver to determine when to stop setting the ECN-
Echo flag, we introduce a second new flag in the TCP header, the CWR
flag. The CWR flag is assigned to Bit 8 in the Reserved field of the
TCP header.
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+
| | | U | A | P | R | S | F |
| Header Length | Reserved | R | C | S | S | Y | I |
| | | G | K | H | T | N | N |
+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+
Figure 2: The old definition of bytes 13 and 14 of the TCP
header.
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+
| | | C | E | U | A | P | R | S | F |
| Header Length | Reserved | W | C | R | C | S | S | Y | I |
| | | R | E | G | K | H | T | N | N |
+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+
Figure 3: The new definition of bytes 13 and 14 of the TCP
Header.
Thus, ECN uses the ECT and CE flags in the IP header (as shown in
Figure 1) for signaling between routers and connection endpoints, and
uses the ECN-Echo and CWR flags in the TCP header (as shown in Figure
3) for TCP-endpoint to TCP-endpoint signaling. For a TCP connection,
a typical sequence of events in an ECN-based reaction to congestion
is as follows:
* The ECT bit is set in packets transmitted by the sender to indi-
cate that ECN is supported by the transport entities for these
packets.
* An ECN-capable router detects impending congestion and detects
Ramakrishnan and Floyd Proposed Standard [Page 11]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
that the ECT bit is set in the packet it is about to drop.
Instead of dropping the packet, the router chooses to set the CE
bit in the IP header and forwards the packet.
* The receiver receives the packet with the CE bit set, and sets
the ECN-Echo flag in its next TCP ACK sent to the sender.
* The sender receives the TCP ACK with ECN-Echo set, and reacts to
the congestion as if a packet had been dropped.
* The sender sets the CWR flag in the TCP header of the next
packet sent to the receiver to acknowledge its receipt of and
reaction to the ECN-Echo flag.
The negotiation for using ECN by the TCP transport entities and the
use of the ECN-Echo and CWR flags is described in more detail in the
sections below.
6.1.1 TCP Initialization
In the TCP connection setup phase, the source and destination TCPs
exchange information about their willingness to use ECN. Subsequent
to the completion of this negotiation, the TCP sender sets the ECT
bit in the IP header of data packets to indicate to the network that
the transport is capable and willing to participate in ECN for this
packet. This indicates to the routers that they may mark this packet
with the CE bit, if they would like to use that as a method of con-
gestion notification. If the TCP connection does not wish to use ECN
notification for a particular packet, the sending TCP sets the ECT
bit equal to 0 (i.e., not set), and the TCP receiver ignores the CE
bit in the received packet.
For this discussion, we designate the initiating host as Host A and
the responding host as Host B. We call a SYN packet with the ECE and
CWR flags set an "ECN-setup SYN packet", and we call a SYN packet
with the ECE and CWR flags not set a "non-ECN-setup SYN packet".
Similarly, we call a SYN-ACK packet with only the ECE flag set but
the CWR flag not set an "ECN-setup SYN-ACK packet", and we call a
SYN-ACK packet with both the ECE and CWR flags not set a "non-ECN-
setup SYN-ACK packet".
Before a TCP connection can use ECN, Host A sends an ECN-setup SYN
packet, and Host B sends an ECN-setup SYN-ACK packet. For a SYN
packet, the setting of both ECE and CWR in the ECN-setup SYN packet
is defined as an indication that the sending TCP is ECN-Capable,
rather than as an indication of congestion or of response to conges-
tion. More precisely, an ECN-setup SYN packet indicates that the TCP
implementation transmitting the SYN packet will participate in ECN as
both a sender and receiver. Specifically, as a receiver, it will
respond to incoming data packets that have the CE bit set in the IP
header by setting ECE in outgoing TCP Acknowledgement (ACK) packets.
Ramakrishnan and Floyd Proposed Standard [Page 12]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
As a sender, it will respond to incoming packets that have ECE set by
reducing the congestion window and setting CWR when appropriate. An
ECN-setup SYN packet does not commit the TCP sender to setting the
ECT bit in any or all of the packets it may transmit. However, the
commitment to respond appropriately to incoming packets with the CE
bit set remains even if the TCP sender in a later transmission,
within this TCP connection, sends a SYN packet without ECE and CWR
set.
When Host B sends an ECN-setup SYN-ACK packet, it sets the ECE flag
but not the CWR flag. An ECN-setup SYN-ACK packet is defined as an
indication that the TCP transmitting the SYN-ACK packet is ECN-Capa-
ble. As with the SYN packet, an ECN-setup SYN-ACK packet does not
commit the TCP host to setting the ECT bit in transmitted packets.
The following rules apply to the sending of ECN-setup packets:
* If a host has received an ECN-setup SYN packet, then it MAY send an
ECN-setup SYN-ACK packet. Otherwise, it MUST NOT send an ECN-setup
SYN-ACK packet.
* A host MUST NOT set ECT on data packets unless it has sent at least
one ECN-setup SYN or ECN-setup SYN-ACK packet, and has received at
least one ECN-setup SYN or ECN-setup SYN-ACK packet, and has sent no
non-ECN-setup SYN or non-ECN-setup SYN-ACK packet. If a host has
received at least one non-ECN-setup SYN or non-ECN-setup SYN-ACK
packet, then it SHOULD NOT set ECT on data packets.
* If a host ever sets the ECT bit on a data packet, then that host
MUST correctly set/clear the CWR TCP bit on all subsequent packets in
the connection.
* If a host has sent at least one ECN-setup SYN or ECN-setup SYN-ACK
packet, and has received no non-ECN-setup SYN or non-ECN-setup SYN-
ACK packet, then if that host receives TCP data packets with ECT and
CE bits set in the IP header, then that host MUST process these pack-
ets as specified for an ECN-capable connection.
6.1.1.1. Robust TCP Initialization with an Echoed Reserve Field
There is the question of why we chose to have the TCP sending the SYN
set two ECN-related flags in the Reserved field of the TCP header for
the SYN packet, while the responding TCP sending the SYN-ACK sets
only one ECN-related flag in the SYN-ACK packet. This asymmetry is
necessary for the robust negotiation of ECN-capability with some
deployed TCP implementations. There exists at least one faulty TCP
implementation in which TCP receivers set the Reserved field of the
TCP header in ACK packets (and hence the SYN-ACK) simply to reflect
the Reserved field of the TCP header in the received data packet.
Because the TCP SYN packet sets the ECN-Echo and CWR flags to indi-
cate ECN-capability, while the SYN-ACK packet sets only the ECN-Echo
Ramakrishnan and Floyd Proposed Standard [Page 13]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
flag, the sending TCP correctly interprets a receiver's reflection of
its own flags in the Reserved field as an indication that the
receiver is not ECN-capable. The sending TCP is not mislead by a
faulty TCP implementation sending a SYN-ACK packet that simply
reflects the Reserved field of the incoming SYN packet.
6.1.1.2. Robust TCP Initialization with no response to the SYN
ECN introduces the use of the ECN-Echo and CWR flags in the TCP
header (as shown in Figure 3) for initialization. There exists some
faulty equipment in the Internet that either ignores an ECN-setup SYN
packet or responds with a RST, in the belief that such a packet (with
these bits set) is a signature for a port-scanning tool that could be
used in a denial-of-service attack. To provide robust connectivity
even in the presence of such faulty equipment, a host that receives a
RST in response to the transmission of an ECN-setup SYN packet MAY
resend a SYN with CWR and ECE cleared. This could result in a TCP
connection being established without using ECN. Similarly, a host
that receives no reply to an ECN-setup SYN within the normal SYN
retransmission timeout interval MAY resend the SYN and any subsequent
SYN retransmissions with CWR and ECE cleared. To overcome normal
packet loss that results in the original SYN being lost, the origi-
nating host may retransmit one or more ECN-setup SYN packets before
giving up and retransmitting the SYN with the CWR and ECE bits
cleared.
We note that in this case, the following example scenario is possi-
ble:
(1) Host A: Sends an ECN-setup SYN.
(2) Host B: Sends an ECN-setup SYN/ACK, packet is dropped or
delayed.
(3) Host A: Sends a non-ECN-setup SYN.
(4) Host B: Sends a non-ECN-setup SYN/ACK.
We note that in this case, following the procedures above, neither
Host A nor Host B may set the ECT bit on data packets, We further
note that a host NEVER uses the reception of ECT data packets as an
implicit signal that the other host is ECN-capable.
6.1.2. The TCP Sender
For a TCP connection using ECN, data packets are transmitted with the
ECT bit set in the IP header (set to a "1"). If the sender receives
an ECN-Echo (ECE) ACK packet (that is, an ACK packet with the ECN-
Echo flag set in the TCP header), then the sender knows that conges-
tion was encountered in the network on the path from the sender to
the receiver. The indication of congestion should be treated just as
Ramakrishnan and Floyd Proposed Standard [Page 14]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
a congestion loss in non-ECN-Capable TCP. That is, the TCP source
halves the congestion window "cwnd" and reduces the slow start
threshold "ssthresh". The sending TCP SHOULD NOT increase the con-
gestion window in response to the receipt of an ECN-Echo ACK packet.
TCP should not react to congestion indications more than once every
window of data (or more loosely, more than once every round-trip
time). That is, the TCP sender's congestion window should be reduced
only once in response to a series of dropped and/or CE packets from a
single window of data. In addition, the TCP source should not
decrease the slow-start threshold, ssthresh, if it has been decreased
within the last round trip time. However, if any retransmitted pack-
ets are dropped, then this is interpreted by the source TCP as a new
instance of congestion.
After the source TCP reduces its congestion window in response to a
CE packet, incoming acknowledgements that continue to arrive can
"clock out" outgoing packets as allowed by the reduced congestion
window. If the congestion window consists of only one MSS (maximum
segment size), and the sending TCP receives an ECN-Echo ACK packet,
then the sending TCP should in principle still reduce its congestion
window in half. However, the value of the congestion window is
bounded below by a value of one MSS. If the sending TCP were to con-
tinue to send, using a congestion window of 1 MSS, this results in
the transmission of one packet per round-trip time. It is necessary
to still reduce the sending rate of the TCP sender even further, on
receipt of an ECN-Echo packet when the congestion window is one. We
use the retransmit timer as a means of reducing the rate further in
this circumstance. Therefore, the sending TCP MUST reset the
retransmit timer on receiving the ECN-Echo packet when the congestion
window is one. The sending TCP will then be able to send a new
packet only when the retransmit timer expires.
[Floyd94] discusses TCP's response to ECN in more detail. [Floyd98]
discusses the validation test in the ns simulator, which illustrates
a wide range of ECN scenarios. These scenarios include the following:
an ECN followed by another ECN, a Fast Retransmit, or a Retransmit
Timeout; a Retransmit Timeout or a Fast Retransmit followed by an
ECN; and a congestion window of one packet followed by an ECN.
TCP follows existing algorithms for sending data packets in response
to incoming ACKs, multiple duplicate acknowledgements, or retransmit
timeouts [RFC2581]. TCP also follows the normal procedures for
increasing the congestion window when it receives ACK packets without
the ECN-Echo bit set [RFC2581].
Ramakrishnan and Floyd Proposed Standard [Page 15]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
6.1.3. The TCP Receiver
When TCP receives a CE data packet at the destination end-system, the
TCP data receiver sets the ECN-Echo flag in the TCP header of the
subsequent ACK packet. If there is any ACK withholding implemented,
as in current "delayed-ACK" TCP implementations where the TCP
receiver can send an ACK for two arriving data packets, then the ECN-
Echo flag in the ACK packet will be set to the OR of the CE bits of
all of the data packets being acknowledged. That is, if any of the
received data packets are CE packets, then the returning ACK has the
ECN-Echo flag set.
To provide robustness against the possibility of a dropped ACK packet
carrying an ECN-Echo flag, the TCP receiver sets the ECN-Echo flag in
a series of ACK packets sent subsequently. The TCP receiver uses the
CWR flag received from the TCP sender to determine when to stop set-
ting the ECN-Echo flag.
When an ECN-Capable TCP sender reduces its congestion window for any
reason (because of a retransmit timeout, a Fast Retransmit, or in
response to an ECN Notification), the TCP sender sets the CWR flag in
the TCP header of the first new data packet sent after the window
reduction. If that data packet is dropped in the network, then the
sending TCP will have to reduce the congestion window again and
retransmit the dropped packet.
We ensure that the "Congestion Window Reduced" information is reli-
ably delivered to the TCP receiver. This comes about from the fact
that if the new data packet carrying the CWR flag is dropped, then
the TCP sender will have to again reduce its congestion window, and
send another new data packet with the CWR flag set. Thus, the CWR
bit in the TCP header SHOULD NOT be set on retransmitted packets.
When the TCP data sender is ready to set the CWR bit after reducing
the congestion window, it SHOULD set the CWR bit only on the first
new data packet that it transmits.
After a TCP receiver sends an ACK packet with the ECN-Echo bit set,
that TCP receiver continues to set the ECN-Echo flag in all the ACK
packets it sends (whether they acknowledge CE data packets or non-CE
data packets) until it receives a CWR packet (a packet with the CWR
flag set). After the receipt of the CWR packet, acknowledgements for
subsequent non-CE data packets do not have the ECN-Echo flag set. If
another CE packet is received by the data receiver, the receiver
would once again send ACK packets with the ECN-Echo flag set. While
the receipt of a CWR packet does not guarantee that the data sender
received the ECN-Echo message, this does suggest that the data sender
reduced its congestion window at some point *after* it sent the data
packet for which the CE bit was set.
Ramakrishnan and Floyd Proposed Standard [Page 16]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
We have already specified that a TCP sender is not required to reduce
its congestion window more than once per window of data. Some care
is required if the TCP sender is to avoid unnecessary reductions of
the congestion window when a window of data includes both dropped
packets and (marked) CE packets. This is illustrated in [Floyd98].
6.1.4. Congestion on the ACK-path
For the current generation of TCP congestion control algorithms, pure
acknowledgement packets (e.g., packets that do not contain any accom-
panying data) should be sent with the ECT bit off. Current TCP
receivers have no mechanisms for reducing traffic on the ACK-path in
response to congestion notification. Mechanisms for responding to
congestion on the ACK-path are areas for current and future research.
(One simple possibility would be for the sender to reduce its conges-
tion window when it receives a pure ACK packet with the CE bit set).
For current TCP implementations, a single dropped ACK generally has
only a very small effect on the TCP's sending rate.
6.1.5. Retransmitted TCP packets
This document specifies that for ECN-capable TCP implementations, the
ECT bit (ECN-Capable Transport) in the IP header MUST NOT be set on
retransmitted data packets, and that the TCP data receiver SHOULD
ignore the ECN field on arriving data packets that are outside of the
receiver's current window. This is for greater security against
denial-of-service attacks, as well as for robustness of the ECN con-
gestion indication with packets that are dropped later in the net-
work.
First, we note that if the TCP sender were to set the ECT bit on a
retransmitted packet, then if an unnecessarily-retransmitted packet
was later dropped in the network, the end nodes would never receive
the indication of congestion from the router setting the CE bit.
Thus, setting the ECT bit on retransmitted data packets is not con-
sistent with the robust delivery of the congestion indication even
for packets that are later dropped in the network.
In addition, an attacker capable of spoofing the IP source address of
the TCP sender could send data packets with arbitrary sequence num-
bers, with both the ECT and CE bits set in the IP header. On receiv-
ing this spoofed data packet, the TCP data receiver would determine
that the data does not lie in the current receive window, and return
a duplicate acknowledgement. We define an out-of-window packet at
the TCP data receiver as a data packet that lies outside the
receiver's current window. On receiving an out-of-window packet, the
TCP data receiver has to decide whether or not to treat the CE bit in
the packet header as a valid indication of congestion, and therefore
Ramakrishnan and Floyd Proposed Standard [Page 17]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
whether to return ECN-Echo indications to the TCP data sender. If
the TCP data receiver ignored the CE bit in an out-of-window packet,
then the TCP data sender would not receive this possibly-legitimate
indication of congestion from the network, resulting in a violation
of end-to-end congestion control. On the other hand, if the TCP data
receiver honors the CE indication in the out-of-window packet, and
reports the indication of congestion to the TCP data sender, then the
malicious node that created the spoofed, out-of-window packet has
successfully "attacked" the TCP connection by forcing the data sender
to unnecessarily reduce (halve) its congestion window. To prevent
such a denial-of-service attack, we specify that a legitimate TCP
data sender MUST NOT set the ECT bit on retransmitted data packets,
and that the TCP data receiver SHOULD ignore the CE bit on out-of-
window packets.
One drawback of not setting ECT on retransmitted packets denies ECN
protection for retransmitted packets. However, for an ECN-capable
TCP connection in a fully-ECN-capable environment with mild conges-
tion, packets should rarely be dropped due to congestion in the first
place, and so instances of retransmitted packets should rarely arise.
If packets are being retransmitted, then there are already packet
losses (from corruption or from congestion) that ECN has been unable
to prevent.
We note that if the router sets the CE bit for an ECN-capable data
packet within a TCP connection, then the TCP connection is guaranteed
to receive that indication of congestion, or to receive some other
indication of congestion within the same window of data, even if this
packet is dropped or reordered in the network. We consider two
cases, when the packet is later retransmitted, and when the packet is
not later retransmitted.
In the first case, if the packet is either dropped or delayed, and at
some point retransmitted by the data sender, then the retransmission
is a result of a Fast Retransmit or a Retransmit Timeout for either
that packet or for some prior packet in the same window of data. In
this case, because the data sender already has retransmitted this
packet, we know that the data sender has already responded to an
indication of congestion for some packet within the same window of
data as the original packet. Thus, even if the first transmission of
the packet is dropped in the network, or is delayed, if it had the CE
bit set, and is later ignored by the data receiver as an out-of-win-
dow packet, this is not a problem, because the sender has already
responded to an indication of congestion for that window of data.
In the second case, if the packet is never retransmitted by the data
sender, then this data packet is the only copy of this data received
by the data receiver, and therefore arrives at the data receiver as
Ramakrishnan and Floyd Proposed Standard [Page 18]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
an in-window packet, regardless of how much the packet might be
delayed or reordered. In this case, if the CE bit is set on the
packet within the network, this will be treated by the data receiver
as a valid indication of congestion.
6.1.6. TCP Window Probes.
When the TCP data receiver advertises a zero window, the TCP data
sender sends window probes to determine if the receiver's window has
increased. Window probe packets do not contain any user data except
for the sequence number, which is a byte. If a window probe packet
is dropped in the network, this loss is not detected by the receiver.
Therefore, the TCP data sender MUST NOT set either the ECT or CWR
bits on window probe packets.
However, because window probes use exact sequence numbers, they can-
not be easily spoofed in denial-of-service attacks. Therefore, if a
window probe arrives with ECT and CE set, then the receiver SHOULD
respond to the ECN indications.
7. Non-compliance by the End Nodes
This section discusses concerns about the vulnerability of ECN to
non-compliant end-nodes (i.e., end nodes that set the ECT bit in
transmitted packets but do not respond to received CE packets). We
argue that the addition of ECN to the IP architecture will not sig-
nificantly increase the current vulnerability of the architecture to
unresponsive flows.
Even for non-ECN environments, there are serious concerns about the
damage that can be done by non-compliant or unresponsive flows (that
is, flows that do not respond to congestion control indications by
reducing their arrival rate at the congested link). For example, an
end-node could "turn off congestion control" by not reducing its con-
gestion window in response to packet drops. This is a concern for the
current Internet. It has been argued that routers will have to
deploy mechanisms to detect and differentially treat packets from
non-compliant flows [RFC2309,FF99]. It has also been suggested that
techniques such as end-to-end per-flow scheduling and isolation of
one flow from another, differentiated services, or end-to-end reser-
vations could remove some of the more damaging effects of unrespon-
sive flows.
It might seem that dropping packets in itself is an adequate deter-
rent for non-compliance, and that the use of ECN removes this deter-
rent. We would argue in response that (1) ECN-capable routers pre-
serve packet-dropping behavior in times of high congestion; and (2)
even in times of high congestion, dropping packets in itself is not
Ramakrishnan and Floyd Proposed Standard [Page 19]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
an adequate deterrent for non-compliance.
First, ECN-Capable routers will only mark packets (as opposed to
dropping them) when the packet marking rate is reasonably low. During
periods where the average queue size exceeds an upper threshold, and
therefore the potential packet marking rate would be high, our recom-
mendation is that routers drop packets rather then set the CE bit in
packet headers.
During the periods of low or moderate packet marking rates when ECN
would be deployed, there would be little deterrent effect on unre-
sponsive flows of dropping rather than marking those packets. For
example, delay-insensitive flows using reliable delivery might have
an incentive to increase rather than to decrease their sending rate
in the presence of dropped packets. Similarly, delay-sensitive flows
using unreliable delivery might increase their use of FEC in response
to an increased packet drop rate, increasing rather than decreasing
their sending rate. For the same reasons, we do not believe that
packet dropping itself is an effective deterrent for non-compliance
even in an environment of high packet drop rates, when all flows are
sharing the same packet drop rate.
Several methods have been proposed to identify and restrict non- com-
pliant or unresponsive flows. The addition of ECN to the network
environment would not in any way increase the difficulty of designing
and deploying such mechanisms. If anything, the addition of ECN to
the architecture would make the job of identifying unresponsive flows
slightly easier. For example, in an ECN-Capable environment routers
are not limited to information about packets that are dropped or have
the CE bit set at that router itself; in such an environment, routers
could also take note of arriving CE packets that indicate congestion
encountered by that packet earlier in the path.
8. Non-compliance in the Network
This section considers the issues when a router is operating, possi-
bly maliciously, to modify either of the bits in the ECN field. In
this section we represent the ECN field in the IP header by the tuple
(ECT bit, CE bit).
By tampering with the bits in the ECN field, an adversary (or a bro-
ken router) could do one or more of the following: falsely report
congestion, disable ECN-Capability for an individual packet, erase
the ECN congestion indication, or falsely indicate ECN-Capability.
Appendix X systematically examines the various cases by which the ECN
field could be modified. The important criterion considered in
determining the consequences of such modifications is whether it is
likely to lead to poorer behavior in any dimension (throughput,
Ramakrishnan and Floyd Proposed Standard [Page 20]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
delay, fairness or functionality) than if a router were to drop a
packet.
The first two possible changes, falsely reporting congestion or dis-
abling ECN-Capability for an individual packet, are no worse than if
the router were to simply drop the packet. From a congestion control
point of view, setting the CE bit in the absence of congestion by a
non-compliant router would be no worse than a router dropping a
packet unnecessarily. By "erasing" the ECT bit of a packet that is
later dropped in the network, a router's actions could result in an
unnecessary packet drop for that packet later in the network.
However, as discussed in Section X in the Appendix, a router that
erases the ECN congestion indication or falsely indicates ECN-Capa-
bility could potentially do more damage to the flow that if it has
simply dropped the packet. A rogue or broken router that "erased"
the CE bit in arriving CE packets would prevent that indication of
congestion from reaching downstream receivers. This could result in
the failure of congestion control for that flow and a resulting
increase in congestion in the network, ultimately resulting in subse-
quent packets dropped for this flow as the average queue size
increased at the congested gateway.
Appendix X considers the potential repercussions of subverting end-
to-end congestion control by either falsely indicating ECN-Capabil-
ity, or by erasing the congestion indication in ECN (the CE-bit). We
observe in the Appendix that the consequence of subverting ECN-based
congestion control may lead to potential unfairness, but this is
likely to be no worse than the subversion of either ECN-based or
packet-based congestion control by the end nodes.
8.1. Complications Introduced by Split Paths
If a router or other network element has access to all of the packets
of a flow, then that router could do no more damage to a flow by
altering the ECN field than it could by simply dropping all of the
packets from that flow. However, in some cases, a malicious or bro-
ken router might have access to only a subset of the packets from a
flow. The question is as follows: can this router, by altering the
ECN field in this subset of the packets, do more damage to that flow
than if it has simply dropped that set of the packets?
This is also discussed in detail in the Appendix, which concludes as
follows: It is true that the adversary that has access only to a
subset of packets in an aggregate might, by subverting ECN-based con-
gestion control, be able to deny the benefits of ECN to the other
packets in the aggregate. While this is undesirable, this is not a
sufficient concern to result in disabling ECN within an IP tunnel.
Ramakrishnan and Floyd Proposed Standard [Page 21]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
9. Encapsulated Packets
9.1. IP packets encapsulated in IP
The encapsulation of IP packet headers in tunnels is used in many
places, including IPsec and IP in IP [RFC2003]. Currently, the ECN
specification does not accommodate the constraints imposed by some of
these pre-existing specifications for tunnels. This document consid-
ers issues related to interactions between ECN and IP tunnels, and
specifies two alternative solutions.
Some IP tunnel modes are based on adding a new "outer" IP header that
encapsulates the original, or "inner" IP header and its associated
packet. In many cases, the new "outer" IP header may be added and
removed at intermediate points along a connection, enabling the net-
work to establish a tunnel without requiring endpoint participation.
We denote tunnels that specify that the outer header be discarded at
tunnel egress as "simple tunnels".
ECN uses the ECT and CE flags in the IP header for signaling between
routers and connection endpoints. ECN interacts with IP tunnels
because of the ECT and CE flags in the DS field octet in the IP
header [RFC2474] (also referred to as the IPv4 TOS octet or IPv6
Traffic Class octet). [RFC2983] discusses interactions of Differen-
tiated Services with IP tunnels of various forms. In simple IP tun-
nels the DS field octet is copied or mapped from the inner IP header
to the outer IP header at IP tunnel ingress, and the outer header's
copy of this field is discarded at IP tunnel egress. If the outer
header were to be simply discarded without taking care to deal with
the ECN related flags, and an ECN-capable router were to set the CE
(Congestion Experienced) bit within a packet in a simple IP tunnel,
this indication would be discarded at tunnel egress, losing the indi-
cation of congestion.
Thus, the use of ECN over simple IP tunnels would result in routers
attempting to use the outer IP header to signal congestion to end-
points, but those congestion warnings never arriving because the
outer header is discarded at the tunnel egress point. This problem
was encountered with ECN and IPsec in tunnel mode, and RFC 2481 rec-
ommended that ECN not be used with the older simple IPsec tunnels in
order to avoid this behavior and its consequences. When ECN becomes
widely deployed, then simple tunnels likely to carry ECN-capable
traffic will have to be changed.
From a security point of view, the use of ECN in the outer header of
an IP tunnel might raise security concerns because an adversary could
tamper with the ECN information that propagates beyond the tunnel
endpoint. Based on an analysis in the Appendix of these concerns and
Ramakrishnan and Floyd Proposed Standard [Page 22]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
the resultant risks, our overall approach is to make support for ECN
an option for IP tunnels, so that an IP tunnel can be specified or
configured either to use ECN or not to use ECN in the outer header of
the tunnel. Thus, in environments or tunneling protocols where the
risks of using ECN are judged to outweigh its benefits, the tunnel
can simply not use ECN in the outer header. Then the only indication
of congestion experienced at routers within the tunnel would be
through packet loss.
The result is that there are two viable options for the behavior of
ECN-capable connections over an IP tunnel, especially IPSec tunnels:
* A limited-functionality option in which ECN is preserved in the
inner header, but disabled in the outer header. The only mecha-
nism available for signaling congestion occurring within the tun-
nel in this case is dropped packets.
* A full-functionality option that supports ECN in both the inner
and outer headers, and propagates congestion warnings from nodes
within the tunnel to endpoints.
Support for these options requires varying amounts of changes to IP
header processing at tunnel ingress and egress. A small subset of
these changes sufficient to support only the limited-functionality
option would be sufficient to eliminate any incompatibility between
ECN and IP tunnels.
One goal of this document is to give guidance about the tradeoffs
between the limited-functionality and full-functionality options. A
full discussion of the potential effects of an adversary's modifica-
tions of the CE and ECT bits is given in the Appendix.
9.1.1. The limited-functionality and full-functionality options within
IP Tunnels
The limited-functionality option for ECN encapsulation in IP tunnels
is for the ECT bit in the outside (encapsulating) header to be off
(i.e., set to 0), regardless of the value of the ECT bit in the
inside (encapsulated) header. With this option, the ECN field in the
inner header is not altered upon de-capsulation. The disadvantage of
this approach is that the flow does not have ECN support for that
part of the path that is using IP tunneling, even if the encapsulated
packet (from the original TCP sender) is ECN-Capable. That is, if
the encapsulated packet arrives at a congested router that is ECN-
capable, and the router can decide to drop or mark the packet as an
indication of congestion to the end nodes, the router will not be
permitted to set the CE bit in the packet header, but instead will
have to drop the packet.
The IP full-functionality option for ECN encapsulation is to copy the
Ramakrishnan and Floyd Proposed Standard [Page 23]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
ECT bit of the inside header to the outside header on encapsulation,
and to OR the CE bit from the outer header with the CE bit of the
inside header on decapsulation. That is, for full ECN support the
encapsulation and decapsulation processing for the DS field octet
involves the following: At tunnel ingress, the full-functionality
option copies the value of ECT (bit 6) in the inner header to the
outer header. CE (bit 7) is set to 0 in the outer header. Upon
decapsulation at the tunnel egress, the full-functionality option
sets CE to 1 in the inner header if the value of ECT (bit 6) in the
inner header is 1, and the value of CE (bit 7) in the outer header is
1. Otherwise, no change is made to this field of the inner header.
With the full-functionality option, a flow can take advantage of ECN
in those parts of the path that might use IP tunneling. The disad-
vantage of the full-functionality option from a security perspective
is that the IP tunnel cannot protect the flow from certain modifica-
tions to the ECN bits in the IP header within the tunnel. The poten-
tial dangers from modifications to the ECN bits in the IP header are
described in detail in the Appendix.
(1) An IP tunnel MUST modify the handling of the DS field octet at
IP tunnel endpoints by implementing either the limited-functional-
ity or the full-functionality option.
(2) Optionally, an IP tunnel MAY enable the endpoints of an IP
tunnel to negotiate the choice between the limited-functionality
and the full-functionality option for ECN in the tunnel.
The minimum required to make ECN usable with IP tunnels is the lim-
ited-functionality option, which prevents ECN from being enabled in
the outer header of an IPsec tunnel. Full support for ECN requires
the use of the full-functionality option. If there are no optional
mechanisms for the tunnel endpoints to negotiate a choice between the
limited-functionality or full-functionality option, there can be a
pre-existing agreement between the tunnel endpoints about whether to
support the limited-functionality or the full-functionality ECN
option.
In addition, it is RECOMMENDED that packets with ECT and CE both set
to 1 in the outer header be dropped if they arrive at the tunnel
egress point for a tunnel that uses the limited-functionality option,
or for a tunnel that uses the full-functionality option but for which
the ECT bit in the inner header is set to zero. This is motivated by
backwards compatibility and to ensure that no unauthorized modifica-
tions of the ECN field take place, and is discussed further in the
Appendix.
Ramakrishnan and Floyd Proposed Standard [Page 24]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
9.1.2. Changes to the ECN Field within an IP Tunnel.
The presence of a copy of the ECN field in the inner header of an IP
tunnel mode packet provides an opportunity for detection of unautho-
rized modifications to the ECT bit in the outer header. Comparison
of the ECT bits in the inner and outer headers falls into two cate-
gories for implementations that conform to this document:
* If the IP tunnel uses the full-functionality option, then the
values of the ECT bits in the inner and outer headers should be
identical.
* If the tunnel uses the limited-functionality option, then the
ECT bit in the outer header should be 0.
Receipt of a packet not satisfying the appropriate condition could be
a cause of concern.
Consider the case of an IP tunnel where the tunnel ingress point has
not been updated to this document's requirements, while the tunnel
egress point has been updated to support ECN. In this case, the IP
tunnel is not explicitly configured to support the full-functionality
ECN option. However, the tunnel ingress point is behaving identically
to a tunnel ingress point that supports the full-functionality
option. If packets from an ECN-capable connection use this tunnel,
ECT will be set to 1 in the outer header at the tunnel ingress point.
Congestion within the tunnel may then result in ECN-capable routers
setting CE in the outer header. Because the tunnel has not been
explicitly configured to support the full-functionality option, the
tunnel egress point expects the ECT bit in the outer header to be 0.
When an ECN-capable tunnel egress point receives a packet with the
ECT bit in the outer header set to 1, in a tunnel that has not been
configured to support the full-functionality option, that packet
should be processed, according to whether CE bit was set, as follows.
It is RECOMMENDED that such packets, with the ECT bit in the outer
header set to 1 on a tunnel that has not been configured to support
the full-functionality option, be dropped at the egress point if CE
is set to 1 in the outer header but 0 in the inner header, and for-
warded otherwise.
An IP tunnel cannot provide protection against erasure of congestion
indications based on resetting the value of the CE bit in packets for
which ECT is set in the outer header. The erasure of congestion
indications may impact the network and other flows in ways that would
not be possible in the absence of ECN. It is important to note that
erasure of congestion indications can only be performed to congestion
indications placed by nodes within the tunnel; the copy of the CE bit
in the inner header preserves congestion notifications from nodes
upstream of the tunnel ingress. If erasure of congestion notifica-
tions is judged to be a security risk that exceeds the congestion
Ramakrishnan and Floyd Proposed Standard [Page 25]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
management benefits of ECN, then tunnels could be specified or con-
figured to use the limited-functionality option.
9.2. IPsec Tunnels
IPsec supports secure communication over potentially insecure network
components such as intermediate routers. IPsec protocols support two
operating modes, transport mode and tunnel mode, that span a wide
range of security requirements and operating environments. Transport
mode security protocol header(s) are inserted between the IP (IPv4 or
IPv6) header and higher layer protocol headers (e.g., TCP), and hence
transport mode can only be used for end-to-end security on a connec-
tion. IPsec tunnel mode is based on adding a new "outer" IP header
that encapsulates the original, or "inner" IP header and its associ-
ated packet. Tunnel mode security headers are inserted between these
two IP headers. In contrast to transport mode, the new "outer" IP
header and tunnel mode security headers can be added and removed at
intermediate points along a connection, enabling security gateways to
secure vulnerable portions of a connection without requiring endpoint
participation in the security protocols. An important aspect of tun-
nel mode security is that in the original specification, the outer
header is discarded at tunnel egress, ensuring that security threats
based on modifying the IP header do not propagate beyond that tunnel
endpoint. Further discussion of IPsec can be found in [RFC 2401].
The IPsec protocol as originally defined in [ESP, AH] required that
the inner header's ECN field not be changed by IPsec decapsulation
processing at a tunnel egress node; this would have ruled out the
possibility of full-functionality mode for ECN. At the same time,
this would ensure that an adversary's modifications to the ECN field
cannot be used to launch theft- or denial-of-service attacks across
an IPsec tunnel endpoint, as any such modifications will be discarded
at the tunnel endpoint.
In principle, permitting the use of ECN functionality in the outer
header of an IPsec tunnel raises security concerns because an adver-
sary could tamper with the information that propagates beyond the
tunnel endpoint. Based on an analysis (included in the Appendix) of
these concerns and the associated risks, our overall approach has
been to provide configuration support for IPsec changes to remove the
conflict with ECN.
In particular, in tunnel mode the IPsec tunnel MUST support either
the limited-functionality or the full-functionality mode outlined in
Section X.
This makes permission to use ECN functionality in the outer header of
an IPsec tunnel a configurable part of the corresponding IPsec
Ramakrishnan and Floyd Proposed Standard [Page 26]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
Security Association (SA), so that it can be disabled in situations
where the risks are judged to outweigh the benefits. The result is
that an IPsec security administrator is presented with two alterna-
tives for the behavior of ECN-capable connections within an IPsec
tunnel, the limited-functionality alternative and full-functionality
alternative described earlier. All IPsec implementations MUST imple-
ment either the limited-functionality or the full-functionality
alternative in order to eliminate incompatibility between ECN and
IPsec tunnels, but implementers MAY choose to implement either alter-
native.
In addition, this document specifies how the endpoints of an IPsec
tunnel could negotiate enabling ECN functionality in the outer head-
ers of that tunnel based on security policy. The ability to negoti-
ate ECN usage between tunnel endpoints would enable a security admin-
istrator to disable ECN in situations where she believes the risks
(e.g., of lost congestion notifications) outweigh the benefits of
ECN.
The IPsec protocol, as defined in [ESP, AH], does not include the IP
header's ECN field in any of its cryptographic calculations (in the
case of tunnel mode, the outer IP header's ECN field is not
included). Hence modification of the ECN field by a network node has
no effect on IPsec's end-to-end security, because it cannot cause any
IPsec integrity check to fail. As a consequence, IPsec does not pro-
vide any defense against an adversary's modification of the ECN field
(i.e., a man-in-the-middle attack), as the adversary's modification
will also have no effect on IPsec's end-to-end security. In some
environments, the ability to modify the ECN field without affecting
IPsec integrity checks may constitute a covert channel; if it is nec-
essary to eliminate such a channel or reduce its bandwidth, then the
IPsec tunnel should be run in limited-functionality mode.
9.2.1. Negotiation between Tunnel Endpoints
This section describes the detailed changes to enable usage of ECN
over IPsec tunnels, including the negotiation of ECN support between
tunnel endpoints. This is supported by three changes to IPsec:
* An optional Security Association Database (SAD) field indicating
whether tunnel encapsulation and decapsulation processing allows
or forbids ECN usage in the outer IP header.
* An optional Security Association Attribute that enables negotia-
tion of this SAD field between the two endpoints of an SA that
supports tunnel mode.
* Changes to tunnel mode encapsulation and decapsulation process-
ing to allow or forbid ECN usage in the outer IP header based on
the value of the SAD field. When ECN usage is allowed in the
outer IP header, ECT is set in the outer header for ECN-capable
Ramakrishnan and Floyd Proposed Standard [Page 27]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
connections and congestion notifications (indicated by the CE bit)
from such connections are propagated to the inner header at tunnel
egress.
If negotiation of ECN usage is implemented, then the SAD field SHOULD
also be implemented. On the other hand, negotiation of ECN usage is
OPTIONAL in all cases, even for implementations that support the SAD
field. The encapsulation and decapsulation processing changes are
REQUIRED, but MAY be implemented without the other two changes by
assuming that ECN usage is always forbidden. The full-functionality
alternative for ECN usage over IPsec tunnels consists of the SAD
field and the full version of encapsulation and decapsulation pro-
cessing changes, with or without the OPTIONAL negotiation support.
The limited-functionality alternative consists of a subset of the
encapsulation and decapsulation changes that always forbids ECN
usage.
These changes are covered further in the following three subsections.
9.2.1.1. ECN Tunnel Security Association Database Field
Full ECN functionality adds a new field to the SAD (see [RFC2401]):
ECN Tunnel: allowed or forbidden.
Indicates whether ECN-capable connections using this SA in tunnel
mode are permitted to receive ECN congestion notifications for
congestion occurring within the tunnel. The allowed value enables
ECN congestion notifications. The forbidden value disables such
notifications, causing all congestion to be indicated via dropped
packets.
[OPTIONAL. The value of this field SHOULD be assumed to be "for-
bidden" in implementations that do not support it.]
If this attribute is implemented, then the SA specification in a
Security Policy Database (SPD) entry MUST support a corresponding
attribute, and this SPD attribute MUST be covered by the SPD adminis-
trative interface (currently described in Section 4.4.1 of
[RFC2401]).
9.2.1.2. ECN Tunnel Security Association Attribute
A new IPsec Security Association Attribute is defined to enable the
support for ECN congestion notifications based on the outer IP header
to be negotiated for IPsec tunnels (see [RFC2407]). This attribute
is OPTIONAL, although implementations that support it SHOULD also
support the SAD field defined in Section 3.1.
Ramakrishnan and Floyd Proposed Standard [Page 28]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
Attribute Type
class value type
-------------------------------------------------
ECN Tunnel 10 Basic
The IPsec SA Attribute value 10 has been allocated by IANA to indi-
cate that the ECN Tunnel SA Attribute is being negotiated; the type
of this attribute is Basic (see Section 4.5 of [RFC2407]). The Class
Values are used to conduct the negotiation. See [RFC2407, RFC2408,
RFC2409] for further information including encoding formats and
requirements for negotiating this SA attribute.
Class Values
ECN Tunnel
Specifies whether ECN functionality is allowed to
be used with Tunnel Encapsulation Mode.
This affects tunnel encapsulation and decapsulation processing -
see Section 3.3.
RESERVED 0
Allowed 1
Forbidden 2
Values 3-61439 are reserved to IANA. Values 61440-65535 are for
private use.
If unspecified, the default shall be assumed to be Forbidden.
ECN Tunnel is a new SA attribute, and hence initiators that use it
can expect to encounter responders that do not understand it, and
therefore reject proposals containing it. For backwards compatibil-
ity with such implementations initiators SHOULD always also include a
proposal without the ECN Tunnel attribute to enable such a responder
to select a transform or proposal that does not contain the ECN Tun-
nel attribute. RFC 2407 currently requires responders to reject all
proposals if any proposal contains an unknown attribute; this
requirement is expected to be changed to require a responder not to
select proposals or transforms containing unknown attributes.
9.2.1.3. Changes to IPsec Tunnel Header Processing
Subsequent to the publication of [RFC 2401], the TOS octet of IPv4
and the Traffic Class octet of IPv6 have been superseded by the six-
bit DS Field [RFC2474, RFC2780] and a two-bit "currently unused" (CU)
field [RFC2780], and this document supersedes the CU field by tne ECN
Ramakrishnan and Floyd Proposed Standard [Page 29]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
Field.
For full ECN support, the encapsulation and decapsulation processing
for the IPv4 TOS field and the IPv6 Traffic Class field are changed
from that specified in [RFC2401] to the following:
<-- How Outer Hdr Relates to Inner Hdr -->
Outer Hdr at Inner Hdr at
IPv4 Encapsulator Decapsulator
Header fields: -------------------- ------------
DS Field copied from inner hdr (5) no change
ECN Field constructed (7) constructed (8)
IPv6
Header fields:
DS Field copied from inner hdr (6) no change
ECN Field constructed (7) constructed (8)
(5)(6) If the packet will immediately enter a domain for which the
DSCP value in the outer header is not appropriate, that value MUST
be mapped to an appropriate value for the domain [RFC 2474]. Also
see [RFC 2475] for further information.
(7) If the value of the ECN Tunnel field in the SAD entry for this
SA is "allowed" and the value of ECT (bit 0) is 1 in the inner
header, set ECT to 1 in the outer header, else set ECT to 0 in the
outer header. Set CE (bit 1) to 0 in the outer header.
(8) If the value of the ECN tunnel field in the SAD entry for this
SA is "allowed" and the value of ECT (bit 0) in the inner header
is 1, then set the CE bit (bit 1) in the inner header to the logi-
cal OR of the CE bit in the inner header with the CE bit in the
outer header, else make no change to the ECN field.
(5) and (6) are identical to match usage in [RFC2401], although
they are different in [RFC2401].
The above description applies to implementations that support the ECN
Tunnel field in the SAD; such implementations MUST implement this
processing of the DS field instead of the processing of the IPv4 TOS
octet and IPv6 Traffic Class octet defined in [RFC2401]. This con-
stitutes the full-functionality alternative for ECN usage with IPsec
tunnels.
An implementation that does not support the ECN Tunnel field in the
SAD MUST implement processing of the DS Field by assuming that the
value of the ECN Tunnel field of the SAD is "forbidden" for every SA.
In this case, the processing of the ECN field reduces to:
Ramakrishnan and Floyd Proposed Standard [Page 30]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
(7) Set the ECN field (ECT and CE bits) to zero in the outer
header.
(8) Make no change to the ECN field in the inner header.
This constitutes the limited functionality alternative for ECN usage
with IPsec tunnels.
For backwards compatibility, packets with ECT and CE both set to 1 in
the outer header SHOULD be dropped if they arrive on an SA that is
using the limited-functionality option, or that is using the full-
functionality option (i.e., and has set the ECT flag in the outer
header to 1) for a packet with the ECT flag set to 0 in the inner
header.
9.2.2. Changes to the ECN Field within an IPsec Tunnel.
If the ECN Field is changed inappropriately within an IPsec tunnel,
and this change is detected at the tunnel egress, then the receipt of
a packet not satisfying the appropriate condition for its SA is an
auditable event. An implementation MAY create audit records with
per-SA counts of incorrect packets over some time period rather than
creating an audit record for each erroneous packet. Any such audit
record SHOULD contain the headers from at least one erroneous packet,
but need not contain the headers from every packet represented by the
entry.
9.2.3. Comments for IPsec Support
Substantial comments were received on two areas of this document dur-
ing review by the IPsec working group. This section describes these
comments and explains why the proposed changes were not incorporated.
The first comment indicated that per-node configuration is easier to
implement than per-SA configuration. After serious thought and
despite some initial encouragement of per-node configuration, it no
longer seems to be a good idea. The concern is that as IPsec is pro-
gressively deployed, many ECN-aware IPsec implementations will find
themselves communicating with a mixture of ECN-aware and ECN-unaware
IPsec tunnel endpoints. In such an environment with per-node config-
uration, the only reasonable thing to do is forbid ECN usage for all
IPsec tunnels, which is not the desired outcome.
In the second area, several reviewers noted that SA negotiation is
complex, and adding to it is non-trivial. One reviewer suggested
using ICMP after tunnel setup as a possible alternative. The addi-
tion to SA negotiation in the document is OPTIONAL and will remain
so; implementers are free to ignore it. The authors believe that the
assurance it provides can be useful in a number of situations. In
Ramakrishnan and Floyd Proposed Standard [Page 31]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
practice, if this is not implemented, it can be deleted at a subse-
quent stage in the standards process. Extending ICMP to negotiate
ECN after tunnel setup is more complex than extending SA attribute
negotiation. Some tunnels do not permit traffic to be addressed to
the tunnel egress endpoint, hence the ICMP packet would have to be
addressed to somewhere else, scanned for by the egress endpoint, and
discarded there or at its actual destination. In addition, ICMP
delivery is unreliable, and hence there is a possibility of an ICMP
packet being dropped, entailing the invention of yet another
ack/retransmit mechanism. It seems better simply to specify an
OPTIONAL extension to the existing SA negotiation mechanism.
9.3. IP packets encapsulated in non-IP packet headers.
A different set of issues are raised, relative to ECN, when IP pack-
ets are encapsulated in tunnels with non-IP packet headers. This
occurs with MPLS [MPLS], GRE [GRE], L2TP [L2TP], and PPTP [PPTP].
For these protocols, there is no conflict with ECN; it is just that
ECN cannot be used within the tunnel unless an ECN codepoint can be
specified for the header of the encapsulating protocol. [RFD99] con-
sidered a preliminary proposal for incorporating ECN into MPLS, and
proposals for incorporating ECN into GRE, L2TP, or PPTP will be con-
sidered as the need arises.
10. Issues Raised by Monitoring and Policing Devices
One possibility is that monitoring and policing devices (or more
informally, "penalty boxes") will be installed in the network to mon-
itor whether best-effort flows are appropriately responding to con-
gestion, and to preferentially drop packets from flows determined not
to be using adequate end-to-end congestion control procedures. This
is discussed in more detail in the Appendix.
We recommend that any "penalty box" that detects a flow or an aggre-
gate of flows that is not responding to end-to-end congestion control
first change from marking to dropping packets from that flow, before
taking any additional action to restrict the bandwidth available to
that flow. Thus, initially, the router may drop packets in which the
router would otherwise would have set the CE bit. This could include
dropping those arriving packets for that flow that are ECN-Capable
and that already have the CE bit set. In this way, any congestion
indications seen by that router for that flow will be guaranteed to
also be seen by the end nodes, even in the presence of malicious or
broken routers elsewhere in the path. If we assume that the first
action taken at any "penalty box" for an ECN-capable flow will be to
drop packets instead of marking them, then there is no way that an
adversary that subverts ECN-based end-to-end congestion control can
cause a flow to be characterized as being non-cooperative and placed
Ramakrishnan and Floyd Proposed Standard [Page 32]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
into a more severe action within the "penalty box".
The monitoring and policing devices that are actually deployed could
fall short of the `ideal' monitoring device described above, in that
the monitoring is applied not to a single flow or to a single IPsec
tunnel, but to an aggregate of flows. In this case, the switch from
marking to dropping would apply to all of the flows in that aggre-
gate, denying the benefits of ECN to the other flows in the aggregate
also. At the highest level of aggregation, another form of the dis-
abling of ECN happens even in the absence of monitoring and policing
devices, when ECN-Capable RED queues switch from marking to dropping
packets as an indication of congestion when the average queue size
has exceeded some threshold.
If there were serious operational problems with routers inappropri-
ately erasing the CE bit in packet headers, one potential fix would
be to include a one-bit ECN nonce in packet headers, and for routers
to erase the nonce when they set the CE bit [SCWA99]. Routers that
erased the CE bit would be unable to consistently reconstruct the
original nonce, and thus repeated erasure of the CE bit would be
detected by the end-nodes. (This could in fact be done without
adding any extra bits for ECN in the IP header, by using the ECN
codepoints (ECT=1, CE=0) and (ECT=0, CE=1) as the two values for the
nonce, and by defining the codepoint (ECT=0, CE=1) to mean exactly
the same as the codepoint (ECT=1, CE=0).) However, at this point the
potential danger of misbehaving routers does not seem of sufficient
concern to warrant this additional complication of adding an ECN
nonce to protect against the erasure of the CE bit.
An ECN nonce would also address the problem of misbehaving transport
receivers lying to the transport sender about whether or not the CE
bit was set in a packet. However, another possibility is for the
data sender to test for a misbehaving receiver directly, by occasion-
ally sending a data packet with ECT and CE set, to see if the
receiver reports receiving the CE bit. Of course, if these packets
encountered congestion in the network, the TCP sender would not
receive this indication of congestion, so setting the ECT and CE bits
at the sender would have to be done very sparingly. In addition, the
TCP sender would have to remember which packets were sent with the
ECT and CE bits set, so that it doesn't react to them as if there was
congestion in the network. We believe that further research is
needed on possible transport-based mechanisms for verifying that the
transport receiver does not lie to the transport sender about the
receipt of congestion indications.
Ramakrishnan and Floyd Proposed Standard [Page 33]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
11. Evaluations of ECN
This section discusses some of the related work evaluating the use of
ECN. The ECN Web Page [ECN] has pointers to other papers, as well as
to implementations of ECN.
[Floyd94] considers the advantages and drawbacks of adding ECN to the
TCP/IP architecture. As shown in the simulation-based comparisons,
one advantage of ECN is to avoid unnecessary packet drops for short
or delay-sensitive TCP connections. A second advantage of ECN is in
avoiding some unnecessary retransmit timeouts in TCP. This paper
discusses in detail the integration of ECN into TCP's congestion con-
trol mechanisms. The possible disadvantages of ECN discussed in the
paper are that a non-compliant TCP connection could falsely advertise
itself as ECN-capable, and that a TCP ACK packet carrying an ECN-Echo
message could itself be dropped in the network. The first of these
two issues is discussed in the appendix of this document, and the
second is addressed by the addition of the CWR flag in the TCP
header.
Experimental evaluations of ECN include [RFC2884,K98]. The conclu-
sions of [K98] and [RFC2884] are that ECN TCP gets moderately better
throughput than non-ECN TCP; that ECN TCP flows are fair towards non-
ECN TCP flows; and that ECN TCP is robust with two-way traffic (with
congestion in both directions) and with multiple congested gateways.
Experiments with many short web transfers show that, while most of
the short connections have similar transfer times with or without
ECN, a small percentage of the short connections have very long
transfer times for the non-ECN experiments as compared to the ECN
experiments.
12. Summary of changes required in IP and TCP
This document specified two bits in the IP header, the ECN-Capable
Transport (ECT) bit and the Congestion Experienced (CE) bit, to be
used for ECN. The ECT bit set to "0" indicates that the transport
protocol will ignore the CE bit. This is the default value for the
ECT bit. The ECT bit set to "1" indicates that the transport proto-
col is willing and able to participate in ECN.
The default value for the CE bit is "0". The router sets the CE bit
to "1" to indicate congestion to the end nodes. The CE bit in a
packet header MUST NOT be reset by a router from "1" to "0".
When viewed in terms of code points, this document has defined three
code points for the ECN field, for "not ECT" (ECT=0, CE=0), "ECT but
not CE" (ECT=1, CE=0), and "ECT and CE" (ECT=1, CE=1). The code
point of (ECT=0, CE=1) is not defined in this document. One
Ramakrishnan and Floyd Proposed Standard [Page 34]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
possibility would be for this code point to be used, some time in the
future, for some other function for non-ECN-capable packets. A sec-
ond possibility would be for this code point to be used as an ECN
nonce, as described earlier in the paper. A third possibility would
be for the code point (ECT=0, CE=1) to be used to indicate that the
packet is ECN-capable for an alternate semantics for the Congestion
Experienced indication. However, at this time the code point (ECT=0,
CE=1) remains undefined.
TCP requires three changes for ECN, a setup phase and two new flags
in the TCP header. The ECN-Echo flag is used by the data receiver to
inform the data sender of a received CE packet. The Congestion Win-
dow Reduced (CWR) flag is used by the data sender to inform the data
receiver that the congestion window has been reduced.
When ECN (Explicit Congestion Notification [RFC2481]) is used, it is
required that congestion indications generated within an IP tunnel
not be lost at the tunnel egress. We specified a minor modification
to the IP protocol's handling of the ECN field during encapsulation
and de-capsulation to allow flows that will undergo IP tunneling to
use ECN.
Two options for ECN in tunnels were specified:
1) A limited-functionality option that does not use ECN inside the IP
tunnel, by turning the ECT bit in the outer header off, and not
altering the inner header at the time of decapsulation.
2) The full-functionality option, which copies the ECT bit of the
inner header to the encapsulating header. At decapsulation, if the
ECT bit is set in the inner header, the CE bit on the outer header is
ORed with the CE bit of the inner header to update the CE bit of the
packet.
All IP tunnels MUST implement one of the two alternative approaches
described above. For IPsec tunnels, this document also defines an
optional IPsec SA attribute that enables negotiation of ECN usage
within IPsec tunnels and an optional field in the Security Associa-
tion Database to indicate whether ECN is permitted in tunnel mode on
a SA.
This document is intended to obsolete RFC 2481, "A Proposal to add
Explicit Congestion Notification (ECN) to IP", which defined ECN as
an Experimental Protocol for the Internet Community, as well as to
obsolete three subsequent internet-drafts on ECN, "IPsec Interactions
with ECN", "ECN Interactions with IP Tunnels", and "TCP with ECN: The
Treatment of Retransmitted Data Packets". This document is intended
largely to merge the earlier documents all into a single document,
for greater clarity, in preparation to becoming a Proposed Standard.
The rest of this section describes the relationship between this
Ramakrishnan and Floyd Proposed Standard [Page 35]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
document and its predecessors.
RFC 2481 included a brief discussion of the use of ECN with encapsu-
lated packets, and noted that for the IPsec specifications at the
time (January 1999), flows could not safely use ECN if they were to
traverse IPsec tunnels. RFC 2481 also described the changes that
could be made to IPsec tunnel specifications to made them compatible
with ECN. "IPsec Interactions with ECN" outlined these changes to
IPsec tunnels in detail, and included an extensive discussion of the
security implications of ECN (now included as Sections 18 and 19 of
this document). The draft of "ECN Interactions with IP Tunnels"
extended the discussion of IPsec tunnels to include all IP tunnels.
Because older IP tunnels are not compatible with a flow's use of ECN,
the deployment of ECN in the Internet will create strong pressure for
older IP tunnels to be updated to an ECN-compatible version, using
either the limited-functionality or the full-functionality option.
This document does not address the issue of including ECN in non-IP
tunnels such as MPLS, GRE, L2TP, or PPTP. An earlier preliminary
document about adding ECN support to MPLS has since expired.
This document expands on one area not addressed in RFC 2481, the use
of ECN with retransmitted data packets. That is, this document
includes the material from "TCP with ECN: The Treatment of Retrans-
mitted Data Packets" specifying that the ECT bit should not be set on
retransmitted data packets. The motivation for this additional spec-
ification is to eliminate a possible avenue for denial-of-service
attacks on an existing TCP connection. Some prior deployments of
ECN-capable TCP might not conform to the (new) requirement not to set
the ECT bit on retransmitted packets; we do not believe this will
cause significant problems in practice.
This document also expands on the specification of the use of SYN
packets for the negotiation of ECN, and specifies some optional
behavior for this. In particular, the document allows a TCP host to
send a non-ECN-setup SYN packet after sending a failed ECN-setup SYN
packet, and precisely specifies the required behavior when both ECN-
setup SYN packets and non-ECN-setup SYN packets are sent in the same
connection. While some prior deployments of ECN-capable TCP might
not conform to the requirements specified in this document, we do not
believe that this will lead to any performance or compatibility prob-
lems for TCP connections with a combination of TCP implementations at
the endpoints.
13. Conclusions
Given the current effort to implement AQM, we believe this is the
right time to deploy congestion avoidance mechanisms that do not
Ramakrishnan and Floyd Proposed Standard [Page 36]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
depend on packet drops alone. With the increased deployment of
applications and transports sensitive to the delay and loss of a sin-
gle packet (e.g., realtime traffic, short web transfers), depending
on packet loss as a normal congestion notification mechanism appears
to be insufficient (or at the very least, non-optimal).
We examined the consequence of modifications of the ECN field within
the network, analyzing all the opportunities for an adversary to
change the ECN field. In many cases, the change to the ECN field is
no worse than dropping a packet. However, we noted that some changes
have the more serious consequence of subverting end-to-end congestion
control. However, we point out that even then the potential damage
is limited, and is similar to the threat posed by end-systems inten-
tionally failing to cooperate with end-to-end congestion control.
14. Acknowledgements
Many people have made contributions to this work and this document,
including many that we have not managed to directly acknowledge in
this document. In addition, we would like to thank Kenjiro Cho for
the proposal for the TCP mechanism for negotiating ECN-Capability,
Kevin Fall for the proposal of the CWR bit, Steve Blake for material
on IPv4 Header Checksum Recalculation, Jamal Hadi-Salim for discus-
sions of ECN issues, and Steve Bellovin, Jim Bound, Brian Carpenter,
Paul Ferguson, Stephen Kent, Greg Minshall, and Vern Paxson for dis-
cussions of security issues. We also thank the Internet End-to-End
Research Group for ongoing discussions of these issues.
Email discussions with a number of people, including Alexey
Kuznetsov, Jamal Hadi-Salim, and Venkat Venkatsubra, have addressed
the issues raised by non-conformant equipment in the Internet that
does not respond to TCP SYN packets with the ECE and CWR flags set.
We thank Mark Handley, Jitentra Padhye, and others for contributions
to the TCP initialization procedures.
The discussion of ECN and IP tunnel considerations draws heavily on
related discussions and documents from the Differentiated Services
Working Group. We thank Tabassum Bint Haque from Dhaka, Bangladesh,
for feedback on IP tunnels. We thank Derrell Piper and Kero Tivinen
for proposing modifications to RFC 2407 that improve the usability of
negotiating the ECN Tunnel SA attribute.
15. References
[AH] Kent, S. and R. Atkinson, "IP Authentication Header", RFC 2402,
November 1998.
[B97] Bradner, S., "Key words for use in RFCs to Indicate Requirement
Ramakrishnan and Floyd Proposed Standard [Page 37]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
Levels", BCP 14, RFC 2119, March 1997.
[ECN] "The ECN Web Page", URL "http://www.aciri.org/floyd/ecn.html".
[ESP] Kent, S. and R. Atkinson, "IP Encapsulating Security Payload",
RFC 2406, November 1998.
[FJ93] Floyd, S., and Jacobson, V., "Random Early Detection gateways
for Congestion Avoidance", IEEE/ACM Transactions on Networking, V.1
N.4, August 1993, p. 397-413.
[Floyd94] Floyd, S., "TCP and Explicit Congestion Notification", ACM
Computer Communication Review, V. 24 N. 5, October 1994, p. 10-23.
[Floyd98] Floyd, S., "The ECN Validation Test in the NS Simulator",
URL "http://www-mash.cs.berkeley.edu/ns/", test tcl/test/test-all-
ecn.
[FF99] Floyd, S., and Fall, K., "Promoting the Use of End-to-End Con-
gestion Control in the Internet", IEEE/ACM Transactions on Network-
ing, August 1999.
[FRED] Lin, D., and Morris, R., "Dynamics of Random Early Detection",
SIGCOMM '97, September 1997.
[GRE] S. Hanks, T. Li, D. Farinacci, and P. Traina, Generic Routing
Encapsulation (GRE), RFC 1701, October 1994.
[Jacobson88] V. Jacobson, "Congestion Avoidance and Control", Proc.
ACM SIGCOMM '88, pp. 314-329.
[Jacobson90] V. Jacobson, "Modified TCP Congestion Avoidance Algo-
rithm", Message to end2end-interest mailing list, April 1990. URL
"ftp://ftp.ee.lbl.gov/email/vanj.90apr30.txt".
[K98] Krishnan, H., "Analyzing Explicit Congestion Notification (ECN)
benefits for TCP", Master's thesis, UCLA, 1998, URL
"http://www.cs.ucla.edu/~hari/software/ecn/ ecn_report.ps.gz".
[L2TP] W. Townsley, A. Valencia, A. Rubens, G. Pall, G. Zorn, and B.
Palter Layer Two Tunneling Protocol "L2TP", RFC 2661, August 1999.
[MJV96] S. McCanne, V. Jacobson, and M. Vetterli, "Receiver- driven
Layered Multicast", SIGCOMM '96, August 1996, pp. 117-130.
[MPLS] D. Awduche, J. Malcolm, J. Agogbua, M. O'Dell, J. McManus,
Requirements for Traffic Engineering Over MPLS, RFC 2702, September
1999.
Ramakrishnan and Floyd Proposed Standard [Page 38]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
[PPTP] Hamzeh, K., Pall, G., Verthein, W., Taarud, J., Little, W.
and G. Zorn, "Point-to-Point Tunneling Protocol (PPTP)", RFC 2637,
July 1999.
[RFC791] Postel, J., "Internet Protocol", STD 5, RFC 791, September
1981.
[RFC793] Postel, J., "Transmission Control Protocol", STD 7, RFC 793,
September 1981.
[RFC1141] Mallory, T. and A. Kullberg, "Incremental Updating of the
Internet Checksum", RFC 1141, January 1990.
[RFC1349] Almquist, P., "Type of Service in the Internet Protocol
Suite", RFC 1349, July 1992.
[RFC1455] Eastlake, D., "Physical Link Security Type of Service", RFC
1455, May 1993.
[RFC1701] Hanks, S., Li, T., Farinacci, D., and P. Traina, Generic
Routing Encapsulation (GRE), RFC 1701, October 1994.
[RFC1702] Hanks, S., Li, T., Farinacci, D., and P. Traina, Generic
Routing Encapsulation over IPv4 networks, RFC 1702, October 1994.
[RFC2003] Perkins, C., IP Encapsulation within IP, RFC 2003, October
1996.
[RFC 2119] S. Bradner, Key words for use in RFCs to Indicate Require-
ment Levels, RFC 2119, March 1997.
[RFC2309] Braden, B., et al., "Recommendations on Queue Management
and Congestion Avoidance in the Internet", RFC 2309, April 1998.
[RFC 2401] S. Kent and R. Atkinson, Security Architecture for the
Internet Protocol, RFC 2401, November 1998.
[RFC2407] D. Piper, The Internet IP Security Domain of Interpretation
for ISAKMP, RFC 2407, November 1998.
[RFC2408] D. Maughan, M. Schertler, M. Schneider, and J. Turner,
Internet Security Association and Key Management Protocol (ISAKMP),
RFC 2409, November 1998.
[RFC2409] D. Harkins and D. Carrel, The Internet Key Exchange (IKE),
RFC 2409, November 1998.
[RFC2474] Nichols, K., Blake, S., Baker, F. and D. Black, "Definition
Ramakrishnan and Floyd Proposed Standard [Page 39]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
of the Differentiated Services Field (DS Field) in the IPv4 and IPv6
Headers", RFC 2474, December 1998.
[RFC2475] S. Blake, D. Black, M. Carlson, E. Davies, Z. Wang, and W.
Weiss, An Architecture for Differentiated Services, RFC 2475, Decem-
ber 1998.
[RFC2481] K. Ramakrishnan and S. Floyd, A Proposal to add Explicit
Congestion Notification (ECN) to IP, RFC 2481, January 1999.
[RFC2581] M. Allman, V. Paxson, W. Stevens, "TCP Congestion Control",
RFC 2581, April 1999.
[RFC2884] Jamal Hadi Salim and Uvaiz Ahmed, "Performance Evaluation
of Explicit Congestion Notification (ECN) in IP Networks", RFC 2884,
July 2000.
[RFC2983] D. Black, "Differentiated Services and Tunnels", RFC2983,
October 2000.
[RFC2780] S. Bradner and V. Paxson, IANA Allocation Guidelines For
Values In the Internet Protocol and Related Headers, RFC 2780, March
2000.
[RFD99] Ramakrishnan, Floyd, S., and Davie, B., A Proposal to Incor-
porate ECN in MPLS, work in progress, June 1999. URL
"http://www.aciri.org/floyd/papers/draft-ietf-mpls-ecn-00.txt".
[RJ90] K. K. Ramakrishnan and Raj Jain, "A Binary Feedback Scheme for
Congestion Avoidance in Computer Networks", ACM Transactions on Com-
puter Systems, Vol.8, No.2, pp. 158-181, May 1990.
[SCWA99] Stefan Savage, Neal Cardwell, David Wetherall, and Tom
Anderson, TCP Congestion Control with a Misbehaving Receiver, ACM
Computer Communications Review, October 1999.
16. Security Considerations
Security considerations have been discussed in Sections 7 and 8.
17. IPv4 Header Checksum Recalculation
IPv4 header checksum recalculation is an issue with some high-end
router architectures using an output-buffered switch, since most if
not all of the header manipulation is performed on the input side of
the switch, while the ECN decision would need to be made local to the
output buffer. This is not an issue for IPv6, since there is no IPv6
header checksum. The IPv4 TOS octet is the last byte of a 16-bit
Ramakrishnan and Floyd Proposed Standard [Page 40]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
half-word.
RFC 1141 [RFC1141] discusses the incremental updating of the IPv4
checksum after the TTL field is decremented. The incremental updat-
ing of the IPv4 checksum after the CE bit was set would work as fol-
lows: Let HC be the original header checksum, and let HC' be the new
header checksum after the CE bit has been set. Then for header
checksums calculated with one's complement subtraction, HC' would be
recalculated as follows:
HC' = { HC - 1 HC > 1
{ 0x0000 HC = 1
For header checksums calculated on two's complement machines, HC' would
be recalculated as follows after the CE bit was set:
HC' = { HC - 1 HC > 0
{ 0xFFFE HC = 0
18. Possible Changes to the ECN Field in the Network
This section discusses in detail possible changes to the ECN field in
the network, such as falsely reporting congestion, disabling ECN-
Capability for an individual packet, erasing the ECN congestion indi-
cation, or falsely indicating ECN-Capability. We represent the ECN
bits in the IP header by the tuple (ECT bit, CE bit).
18.1. Possible Changes to the IP Header
18.1.1. Erasing the Congestion Indication
First, we consider the changes that a router could make that would
result in effectively erasing the congestion indication after it had
been set by a router upstream. The convention followed is:
(ECT, CE) of received packet -> (ECT, CE) of packet transmitted.
(1, 1) -> (1, 0): erase only the CE bit that was set.
(1, 1) -> (0, 0): erase both the ECT bit and the CE bit.
(1, 1) -> (0, 1): erase the ECT bit
The first change turns off the CE bit after it has been set by some
upstream router along the path. The consequence for the upstream
router is that there is a potential for congestion to build for a
time, because the congestion indication does not reach the source.
However, the packet would be received and acknowledged.
Ramakrishnan and Floyd Proposed Standard [Page 41]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
The potential effect of erasing the congestion indication is complex,
and is discussed in depth in Section 19 below. Note that the effect
of erasing the congestion indication is different from dropping a
packet in the network. When a data packet is dropped, the drop is
detected by the TCP sender, and interpreted as an indication of con-
gestion. Similarly, if a sufficient number of consecutive acknowl-
edgement packets are dropped, causing the cumulative acknowledgement
field not to be advanced at the sender, the sender is limited by the
congestion window from sending additional packets, and ultimately the
retransmit timer expires.
In contrast, a systematic erasure of the CE bit by a downstream
router can have the effect of causing a queue buildup at an upstream
router, including the possible loss of packets due to buffer over-
flow. There is a potential of unfairness in that another flow that
goes through the congested router could react to the CE bit set while
the flow that has the CE bit erased could see better performance.
The limitations on this potential unfairness are discussed in more
detail in Section 19 below.
The second change is to turn off both the ECT and the CE bits, thus
erasing the congestion indication and disabling ECN-Capability at the
same time. The third change turns off only the ECT bit, disabling
ECN-Capability.
Within an IP tunnel using the full-functionality option, the third
change would not erase the congestion indication, but would only dis-
able ECN-Capability for that packet within the rest of the tunnel.
However, when performed outside of an IP tunnel, the third change
would also effectively erase the congestion indication, because an
ECN field of (0, 1) is undefined.
The `erasure' of the congestion indication is only effective if the
packet does not end up being marked or dropped again by a downstream
router. With the first change, the packet remains ECN-Capable, and
could be either marked or dropped by a downstream router as an indi-
cation of congestion. With the second and third changes, the packet
is no longer ECN-capable, and can therefore be dropped but not marked
by a downstream router as an indication of congestion.
18.1.2. Falsely Reporting Congestion
(1, 0) -> (1, 1)
This change is to set the CE bit when the ECT bit was already set,
even though there was no congestion. This change does not affect the
treatment of that packet along the rest of the path. In particular,
a router does not examine the CE bit in deciding whether to drop or
Ramakrishnan and Floyd Proposed Standard [Page 42]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
mark an arriving packet.
However, this could result in the application unnecessarily invoking
end-to-end congestion control, and reducing its arrival rate. By
itself, this is no worse (for the application or for the network)
than if the tampering router had actually dropped the packet.
18.1.3. Disabling ECN-Capability
(1, 0) -> (0, *)
This change is to turn off the ECT bit of a packet that does not have
the CE bit set. (Section 18.1.1 discussed the case of turning off
the ECT bit of a packet that does have the CE bit set.) This means
that if the packet later encounters congestion (e.g., by arriving to
a RED queue with a moderate average queue size), it will be dropped
instead of being marked. By itself, this is no worse (for the appli-
cation) than if the tampering router had actually dropped the packet.
The saving grace in this particular case is that there is no con-
gested router upstream expecting a reaction from setting the CE bit.
18.1.4. Falsely Indicating ECN-Capability
This change would incorrectly label a packet as ECN-Capable. The
packet may have been sent either by an ECN-Capable transport or a
transport that is not ECN-Capable.
(0, *) -> (1, 0);
(0, *) -> (1, 1);
If the packet later encounters moderate congestion at an ECN-Capable
router, the router could set the CE bit instead of dropping the
packet. If the transport protocol in fact is not ECN-Capable, then
the transport will never receive this indication of congestion, and
will not reduce its sending rate in response. The potential conse-
quences of falsely indicating ECN-capability are discussed further in
Section 19 below.
If the packet never later encounters congestion at an ECN-Capable
router, then the first of these two changes would have no effect.
The second change, however, would have the effect of giving false
reports of congestion to a monitoring device along the path. If the
transport protocol is ECN-Capable, then the second of these two
changes (when, for example, (0,0) was changed to (1,1)) could also
have an effect at the transport level, by combining falsely indicat-
ing ECN-Capability with falsely reporting congestion. For an ECN-
capable transport, this would cause the transport to unnecessarily
react to congestion. In this particular case, the router that is
incorrectly changing the ECN field could have dropped the packet.
Ramakrishnan and Floyd Proposed Standard [Page 43]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
Thus for this case of an ECN-capable transport, the consequence of
this change to the ECN field is no worse than dropping the packet.
18.1.5. Changes with No Functional Effect
(0, *) -> (0, *)
The CE bit is ignored in a packet that does not have the ECT bit set.
Thus, this change would have no effect, in terms of ECN.
18.2. Information carried in the Transport Header
For TCP, an ECN-capable TCP receiver informs its TCP peer that it is
ECN-capable at the TCP level, using information in the TCP header at
the time the connection is setup. This document does not consider
potential dangers introduced by changes in the transport header
within the network. In the case of IPsec tunnels, the IPsec tunnel
protects the transport header.
18.3. Split Paths
In some cases, a malicious or broken router might have access to only
a subset of the packets from a flow. The question is as follows:
can this router, by altering the ECN field in this subset of the
packets, do more damage to that flow than if it had simply dropped
that set of packets?
We will classify the packets in the flow as A packets and B packets,
and assume that the adversary only has access to A packets. Assume
that the adversary is subverting end-to-end congestion control along
the path traveled by A packets only, by either falsely indicating
ECN-Capability upstream of the point where congestion occurs, or
erasing the congestion indication downstream. Consider also that
there exists a monitoring device that sees both the A and B packets,
and will "punish" both the A and B packets if the total flow is
determined not to be properly responding to indications of conges-
tion. Another key characteristic that we believe is likely to be
true is that the monitoring device, before `punishing' the A&B flow,
will first drop packets instead of setting the CE bit, and will drop
arriving packets of that flow that already have the ECT and CE bits
set. If the end nodes are in fact using end-to-end congestion con-
trol, they will see all of the indications of congestion seen by the
monitoring device, and will begin to respond to these indications of
congestion. Thus, the monitoring device is successful in providing
the indications to the flow at an early stage.
It is true that the adversary that has access only to the A packets
might, by subverting ECN-based congestion control, be able to deny
Ramakrishnan and Floyd Proposed Standard [Page 44]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
the benefits of ECN to the other packets in the A&B aggregate. While
this is unfortunate, this is not a reason to disable ECN within an
IPsec tunnel.
A variant of falsely reporting congestion occurs when there are two
adversaries along a path, where the first adversary falsely reports
congestion, and the second adversary `erases' those reports. (Unlike
packet drops, ECN congestion reports can be `reversed' later in the
network by a malicious or broken router.) While this would be trans-
parent to the end node, it is possible that a monitoring device
between the first and second adversaries would see the false indica-
tions of congestion. Keep in mind our recommendation in this docu-
ment, that before `punishing' a flow for not responding appropriately
to congestion, the router will first switch to dropping rather than
marking as an indication of congestion, for that flow. When this
includes dropping arriving packets from that flow that have the CE
bit set, this ensures that these indications of congestion are being
seen by the end nodes. Thus, there is no additional harm that we are
able to postulate as a result of multiple conflicting adversaries.
19. Implications of Subverting End-to-End Congestion Control
This section focuses on the potential repercussions of subverting
end-to-end congestion control by either falsely indicating ECN-Capa-
bility, or by erasing the congestion indication in ECN (the CE-bit).
Subverting end-to-end congestion control by either of these two meth-
ods can have consequences both for the application and for the net-
work. We discuss these separately below.
The first method to subvert end-to-end congestion control, that of
falsely indicating ECN-Capability, effectively subverts end-to-end
congestion control only if the packet later encounters congestion
that results in the setting of the CE bit. In this case, the trans-
port protocol (which may not be ECN-capable) does not receive the
indication of congestion from these downstream congested routers.
The second method to subvert end-to-end congestion control, `erasing'
the (set) CE bit in a packet, effectively subverts end-to-end conges-
tion control only when the CE bit in the packet was set earlier by a
congested router. In this case, the transport protocol does not
receive the indication of congestion from the upstream congested
routers.
Either of these two methods of subverting end-to-end congestion con-
trol can potentially introduce more damage to the network (and possi-
bly to the flow itself) than if the adversary had simply dropped
packets from that flow. However, as we discuss later in this section
and in Section 7, this potential damage is limited.
Ramakrishnan and Floyd Proposed Standard [Page 45]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
19.1. Implications for the Network and for Competing Flows
The CE bit of the ECN field is only used by routers as an indication
of congestion during periods of *moderate* congestion. ECN-capable
routers should drop rather than mark packets during heavy congestion
even if the router's queue is not yet full. For example, for routers
using active queue management based on RED, the router should drop
rather than mark packets that arrive while the average queue sizes
exceed the RED queue's maximum threshold.
One consequence for the network of subverting end-to-end congestion
control is that flows that do not receive the congestion indications
from the network might increase their sending rate until they drive
the network into heavier congestion. Then, the congested router
could begin to drop rather than mark arriving packets. For flows
that are not isolated by some form of per-flow scheduling or other
per-flow mechanisms, but are instead aggregated with other flows in a
single queue in an undifferentiated fashion, this packet-dropping at
the congested router would apply to all flows that share that queue.
Thus, the consequences would be to increase the level of congestion
in the network.
In some cases, the increase in the level of congestion will lead to a
substantial buffer buildup at the congested queue that will be suffi-
cient to drive the congested queue from the packet-marking to the
packet-dropping regime. This transition could occur either because
of buffer overflow, or because of the active queue management policy
described above that drops packets when the average queue is above
RED's maximum threshold. At this point, all flows, including the
subverted flow, will begin to see packet drops instead of packet
marks, and a malicious or broken router will no longer be able to
`erase' these indications of congestion in the network. If the end
nodes are deploying appropriate end-to-end congestion control, then
the subverted flow will reduce its arrival rate in response to con-
gestion. When the level of congestion is sufficiently reduced, the
congested queue can return from the packet-dropping regime to the
packet-marking regime. The steady-state pattern could be one of the
congested queue oscillating between these two regimes.
In other cases, the consequences of subverting end-to-end congestion
control will not be severe enough to drive the congested link into
sufficiently-heavy congestion that packets are dropped instead of
being marked. In this case, the implications for competing flows in
the network will be a slightly-increased rate of packet marking or
dropping, and a corresponding decrease in the bandwidth available to
those flows. This can be a stable state if the arrival rate of the
subverted flow is sufficiently small, relative to the link bandwidth,
that the average queue size at the congested router remains under
Ramakrishnan and Floyd Proposed Standard [Page 46]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
control. In particular, the subverted flow could have a limited
bandwidth demand on the link at this router, while still getting more
than its "fair" share of the link. This limited demand could be due
to a limited demand from the data source; a limitation from the TCP
advertised window; a lower-bandwidth access pipe; or other factors.
Thus the subversion of ECN-based congestion control can still lead to
unfairness, which we believe is appropriate to note here.
The threat to the network posed by the subversion of ECN-based con-
gestion control in the network is essentially the same as the threat
posed by an end-system that intentionally fails to cooperate with
end-to-end congestion control. The deployment of mechanisms in
routers to address this threat is an open research question, and is
discussed further in Section 10.
Let us take the example described in Section 18.1.1, where the CE bit
that was set in a packet is erased: {(1, 1) -> (1, 0)}. The conse-
quence for the congested upstream router that set the CE bit is that
this congestion indication does not reach the end nodes for that
flow. The source (even one which is completely cooperative and not
malicious) is thus allowed to continue to increase its sending rate
(if it is a TCP flow, by increasing its congestion window). The flow
potentially achieves better throughput than the other flows that also
share the congested router, especially if there are no policing mech-
anisms or per-flow queueing mechanisms at that router. Consider the
behavior of the other flows, especially if they are cooperative: that
is, the flows that do not experience subverted end-to-end congestion
control. They are likely to reduce their load (e.g., by reducing
their window size) on the congested router, thus benefiting our sub-
verted flow. This results in unfairness. As we discussed above, this
unfairness could either be transient (because the congested queue is
driven into the packet-marking regime), oscillatory (because the con-
gested queue oscillates between the packet marking and the packet
dropping regime), or more moderate but a persistent stable state
(because the congested queue is never driven to the packet dropping
regime).
The results would be similar if the subverted flow was intentionally
avoiding end-to-end congestion control. One difference is that a
flow that is intentionally avoiding end-to-end congestion control at
the end nodes can avoid end-to-end congestion control even when the
congested queue is in packet-dropping mode, by refusing to reduce its
sending rate in response to packet drops in the network. Thus the
problems for the network from the subversion of ECN-based congestion
control are less severe than the problems caused by the intentional
avoidance of end-to-end congestion control in the end nodes. It is
also the case that it is considerably more difficult to control the
behavior of the end nodes than it is to control the behavior of the
Ramakrishnan and Floyd Proposed Standard [Page 47]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
infrastructure itself. This is not to say that the problems for the
network posed by the network's subversion of ECN-based congestion
control are small; just that they are dwarfed by the problems for the
network posed by the subversion of either ECN-based or other cur-
rently known packet-based congestion control mechanisms by the end
nodes.
19.2. Implications for the Subverted Flow
When a source indicates that it is ECN-capable, there is an expecta-
tion that the routers in the network that are capable of participat-
ing in ECN will use the CE bit for indication of congestion. There is
the potential benefit of using ECN in reducing the amount of packet
loss (in addition to the reduced queueing delays because of active
queue management policies). When the packet flows through a tunnel
where the nodes that the tunneled packets traverse are untrusted in
some way, the expectation is that IPsec will protect the flow from
subversion that results in undesirable consequences.
In many cases, a subverted flow will benefit from the subversion of
end-to-end congestion control for that flow in the network, by
receiving more bandwidth than it would have otherwise, relative to
competing non-subverted flows. If the congested queue reaches the
packet-dropping stage, then the subversion of end-to-end congestion
control might or might not be of overall benefit to the subverted
flow, depending on that flow's relative tradeoffs between throughput,
loss, and delay.
One form of subverting end-to-end congestion control is to falsely
indicate ECN-capability by setting the ECT bit. This has the conse-
quence of downstream congested routers setting the CE bit in vain.
However, as we describe in the section below, if the ECT bit is
changed in the IPsec tunnel, this can be detected at the egress point
of the tunnel.
The second form of subverting end-to-end congestion control is to
erase the congestion indication, either by erasing the CE bit
directly, or by erasing the ECT bit when the CE bit is already set.
In this case, it is the upstream congested routers that set the CE
bit in vain.
If the ECT bit is erased within an IP tunnel, then this can be
detected at the egress point of the tunnel. If the CE bit is set
upstream of the IP tunnel, then any erasure of the outer header's CE
bit within the tunnel will have no effect because the inner header
preserves the set value of the CE bit. However, if the CE bit is set
within the tunnel, and erased either within or downstream of the tun-
nel, this is not necessarily detected at the egress point of the
Ramakrishnan and Floyd Proposed Standard [Page 48]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
tunnel.
With this subversion of end-to-end congestion control, an end-system
transport does not respond to the congestion indication. Along with
the increased unfairness for the non-subverted flows described in the
previous section, the congested router's queue could continue to
build, resulting in packet loss at the congested router - which is a
means for indicating congestion to the transport in any case. In the
interim, the flow might experience higher queueing delays, possibly
along with an increased bandwidth relative to other non-subverted
flows. But transports do not inherently make assumptions of consis-
tently experiencing carefully managed queueing in the path. We
believe that these forms of subverting end-to-end congestion control
are no worse for the subverted flow than if the adversary had simply
dropped the packets of that flow itself.
19.3. Non-ECN-Based Methods of Subverting End-to-end Congestion Control
We have shown that, in many cases, a malicious or broken router that
is able to change the bits in the ECN field can do no more damage
than if it had simply dropped the packet in question. However, this
is not true in all cases, in particular in the cases where the broken
router subverted end-to-end congestion control by either falsely
indicating ECN-Capability or by erasing the ECN congestion indication
(in the CE-bit). While there are many ways that a router can harm a
flow by dropping packets, a router cannot subvert end-to-end conges-
tion control by dropping packets. As an example, a router cannot
subvert TCP congestion control by dropping data packets, acknowledge-
ment packets, or control packets.
Even though packet-dropping cannot be used to subvert end-to-end con-
gestion control, there *are* non-ECN-based methods for subverting
end-to-end congestion control that a broken or malicious router could
use. For example, a broken router could duplicate data packets, thus
effectively negating the effects of end-to-end congestion control
along some portion of the path. (For a router that duplicated pack-
ets within an IPsec tunnel, the security administrator can cause the
duplicate packets to be discarded by configuring anti-replay protec-
tion for the tunnel.) This duplication of packets within the network
would have similar implications for the network and for the subverted
flow as those described in Sections 18.1.1 and 18.1.4 above.
20. The motivation for the ECT bit.
The need for the ECT bit is motivated by the fact that ECN will be
deployed incrementally in an Internet where some transport protocols
and routers understand ECN and some do not. With the ECT bit, the
router can drop packets from flows that are not ECN-capable, but can
Ramakrishnan and Floyd Proposed Standard [Page 49]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
*instead* set the CE bit in packets that *are* ECN-capable. Because
the ECT bit allows an end node to have the CE bit set in a packet
*instead* of having the packet dropped, an end node might have some
incentive to deploy ECN.
If there was no ECT indication, then the router would have to set the
CE bit for packets from both ECN-capable and non-ECN-capable flows.
In this case, there would be no incentive for end-nodes to deploy
ECN, and no viable path of incremental deployment from a non-ECN
world to an ECN-capable world. Consider the first stages of such an
incremental deployment, where a subset of the flows are ECN-capable.
At the onset of congestion, when the packet dropping/marking rate
would be low, routers would only set CE bits, rather than dropping
packets. However, only those flows that are ECN-capable would under-
stand and respond to CE packets. The result is that the ECN- capable
flows would back off, and the non-ECN-capable flows would be unaware
of the ECN signals and would continue to open their congestion win-
dows.
In this case, there are two possible outcomes: (1) the ECN-capable
flows back off, the non-ECN-capable flows get all of the bandwidth,
and congestion remains mild, or (2) the ECN-capable flows back off,
the non-ECN-capable flows don't, and congestion increases until the
router transitions from setting the CE bit to dropping packets.
While this second outcome evens out the fairness, the ECN-capable
flows would still receive little benefit from being ECN-capable,
because the increased congestion would drive the router to packet-
dropping behavior.
A flow that advertised itself as ECN-Capable but does not respond to
CE bits is functionally equivalent to a flow that turns off conges-
tion control, as discussed earlier in this document.
Thus, in a world when a subset of the flows are ECN-capable, but
where ECN-capable flows have no mechanism for indicating that fact to
the routers, there would be less effective and less fair congestion
control in the Internet, resulting in a strong incentive for end
nodes not to deploy ECN.
21. Why use two bits in the IP header?
Given the need for an ECT indication in the IP header, there still
remains the question of whether the ECT (ECN-Capable Transport) and
CE (Congestion Experienced) indications should have been overloaded
on a single bit. This overloaded-one-bit alternative, explored in
[Floyd94], would have involved a single bit with two values. One
value, "ECT and not CE", would represent an ECN-Capable Transport,
and the other value, "CE or not ECT", would represent either
Ramakrishnan and Floyd Proposed Standard [Page 50]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
Congestion Experienced or a non-ECN-Capable transport.
One difference between the one-bit and two-bit implementations con-
cerns packets that traverse multiple congested routers. Consider a
CE packet that arrives at a second congested router, and is selected
by the active queue management at that router for either marking or
dropping. In the one-bit implementation, the second congested router
has no choice but to drop the CE packet, because it cannot distin-
guish between a CE packet and a non-ECT packet. In the two-bit
implementation, the second congested router has the choice of either
dropping the CE packet, or of leaving it alone with the CE bit set.
Another difference between the one-bit and two-bit implementations
comes from the fact that with the one-bit implementation, receivers
in a single flow cannot distinguish between CE and non-ECT packets.
Thus, in the one-bit implementation an ECN-capable data sender would
have to unambiguously indicate to the receiver or receivers whether
each packet had been sent as ECN-Capable or as non-ECN-Capable. One
possibility would be for the sender to indicate in the transport
header whether the packet was sent as ECN-Capable. A second possi-
bility that would involve a functional limitation for the one- bit
implementation would be for the sender to unambiguously indicate that
it was going to send *all* of its packets as ECN-Capable or as non-
ECN-Capable. For a multicast transport protocol, this unambiguous
indication would have to be apparent to receivers joining an on-going
multicast session.
Another concern that was described earlier (and recommended in this
document) is that transports (particularly TCP) should not mark pure
ACK packets or retransmitted packets as being ECN-Capable. A pure
ACK packet from a non-ECN-capable transport could be dropped, without
necessarily having an impact on the transport from a congestion con-
trol perspective (because subsequent ACKs are cumulative). An ECN-
capable transport reacting to the CE bit set in a pure ACK packet by
reducing the window would be at a disadvantage in comparison to a
non-ECN-capable transport. For this reason (and for reasons described
earlier in relation to retransmitted packets), it is desirable to
have the ECN-Capable bit indication on a per-packet basis.
Another advantage of the two-bit approach is that it is somewhat more
robust. The most critical issue, discussed in Section 8, is that the
default indication should be that of a non-ECN-Capable transport. In
a two-bit implementation, this requirement for the default value sim-
ply means that the ECT bit should be `OFF' by default. In the one-
bit implementation, this means that the single overloaded bit should
by default be in the "CE or not ECT" position. This is less clear
and straightforward, and possibly more open to incorrect implementa-
tions either in the end nodes or in the routers.
Ramakrishnan and Floyd Proposed Standard [Page 51]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
In summary, while the one-bit implementation could be a possible
implementation, it has the following significant limitations relative
to the two-bit implementation. First, the one-bit implementation has
more limited functionality for the treatment of CE packets at a sec-
ond congested router. Second, the one-bit implementation requires
either that extra information be carried in the transport header of
packets from ECN-Capable flows (to convey the functionality of the
second bit elsewhere, namely in the transport header), or that
senders in ECN-Capable flows accept the limitation that receivers
must be able to determine a priori which packets are ECN-Capable and
which are not ECN-Capable. Third, the one-bit implementation is pos-
sibly more open to errors from faulty implementations that choose the
wrong default value for the ECN bit. We believe that the use of the
extra bit in the IP header for the ECT-bit is extremely valuable to
overcome these limitations.
22. Historical definitions for the IPv4 TOS octet
RFC 791 [RFC791] defined the ToS (Type of Service) octet in the IP
header. In RFC 791, bits 6 and 7 of the ToS octet are listed as
"Reserved for Future Use", and are shown set to zero. The first two
fields of the ToS octet were defined as the Precedence and Type of
Service (TOS) fields.
0 1 2 3 4 5 6 7
+-----+-----+-----+-----+-----+-----+-----+-----+
| PRECEDENCE | TOS | 0 | 0 | RFC 791
+-----+-----+-----+-----+-----+-----+-----+-----+
RFC 1122 included bits 6 and 7 in the TOS field, though it did not
discuss any specific use for those two bits:
0 1 2 3 4 5 6 7
+-----+-----+-----+-----+-----+-----+-----+-----+
| PRECEDENCE | TOS | RFC 1122
+-----+-----+-----+-----+-----+-----+-----+-----+
The IPv4 TOS octet was redefined in RFC 1349 [RFC1349] as follows:
0 1 2 3 4 5 6 7
+-----+-----+-----+-----+-----+-----+-----+-----+
| PRECEDENCE | TOS | MBZ | RFC 1349
+-----+-----+-----+-----+-----+-----+-----+-----+
Bit 6 in the TOS field was defined in RFC 1349 for "Minimize Monetary
Cost". In addition to the Precedence and Type of Service (TOS)
fields, the last field, MBZ (for "must be zero") was defined as
Ramakrishnan and Floyd Proposed Standard [Page 52]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
currently unused. RFC 1349 stated that "The originator of a datagram
sets [the MBZ] field to zero (unless participating in an Internet
protocol experiment which makes use of that bit)."
RFC 1455 [RFC 1455] defined an experimental standard that used all
four bits in the TOS field to request a guaranteed level of link
security.
RFC 1349 is obsoleted by "Definition of the Differentiated Services
Field (DS Field) in the IPv4 and IPv6 Headers" [RFC2474], in which
bits 6 and 7 of the DS field are listed as Currently Unused (CU).
The first six bits of the DS field are defined as the Differentiated
Services CodePoint (DSCP):
0 1 2 3 4 5 6 7
+-----+-----+-----+-----+-----+-----+-----+-----+
| DSCP | CU | RFC 2474
+-----+-----+-----+-----+-----+-----+-----+-----+
Because of this unstable history, the definition of the ECN field in
this document cannot be guaranteed to be backwards compatible with
all past uses of these two bits. The damage that could be done by a
non-ECN-capable router would be to "erase" the CE bit for an ECN-
capable packet that arrived at the router with the CE bit set, or set
the CE bit even in the absence of congestion. This has been dis-
cussed in the section on "Non-compliance in the Network".
The damage that could be done in an ECN-capable environment by a non-
ECN-capable end-node transmitting packets with the ECT bit set has
been discussed in the section on "Non-compliance by the End Nodes".
AUTHORS' ADDRESSES
K. K. Ramakrishnan
TeraOptic Networks, Inc.
Phone: +1 (408) 666-8650
Email: kk@teraoptic.com
Sally Floyd
Phone: +1 (510) 666-2989
ACIRI
Email: floyd@aciri.org
URL: http://www.aciri.org/floyd/
David L. Black
EMC Corporation
42 South St.
Ramakrishnan and Floyd Proposed Standard [Page 53]
draft-ietf-tsvwg-ecn-00 Addition of ECN to IP November 2000
Hopkinton, MA 01748
Phone: +1 (508) 435-1000 x75140
Email: black_david@emc.com
This draft was created in November 2000.
It expires May 2001.
Ramakrishnan and Floyd Proposed Standard [Page 54]