One document matched: draft-nir-ipsecme-ike-tcp-01.xml
<?xml version="1.0"?>
<?xml-stylesheet type='text/xsl' href='./rfc2629.xslt' ?>
<?rfc toc="no"?>
<?rfc symrefs="no"?>
<?rfc compact="yes" ?>
<?rfc sortrefs="yes" ?>
<?rfc strict="yes" ?>
<?rfc linkmailto="yes" ?>
<!DOCTYPE rfc SYSTEM "rfc2629.dtd" [
<!ENTITY RFC2119 SYSTEM "http://xml.resource.org/public/rfc/bibxml/reference.RFC.2119.xml">
<!ENTITY RFC5996 SYSTEM "http://xml.resource.org/public/rfc/bibxml/reference.RFC.5996.xml">
<!ENTITY RFC2407 SYSTEM "http://xml.resource.org/public/rfc/bibxml/reference.RFC.2407.xml">
<!ENTITY RFC2408 SYSTEM "http://xml.resource.org/public/rfc/bibxml/reference.RFC.2408.xml">
<!ENTITY RFC0793 SYSTEM "http://xml.resource.org/public/rfc/bibxml/reference.RFC.0793.xml">
<!ENTITY RFC4787 SYSTEM "http://xml.resource.org/public/rfc/bibxml/reference.RFC.4787.xml">
<!ENTITY I-D.behave-lsn-requirements SYSTEM "http://xml.resource.org/public/rfc/bibxml3/reference.I-D.ietf-behave-lsn-requirements.xml">
]>
<?xml-stylesheet type='text/xsl' href='./rfc2629.xslt' ?>
<?rfc toc="no"?>
<?rfc symrefs="yes"?>
<?rfc compact="yes" ?>
<?rfc subcompact="yes" ?>
<?rfc sortrefs="yes" ?>
<rfc ipr="trust200902" docName="draft-nir-ipsecme-ike-tcp-01" category="std">
<front>
<title abbrev="IKE over TCP">A TCP transport for the Internet Key Exchange</title>
<author initials="Y." surname="Nir" fullname="Yoav Nir">
<organization abbrev="Check Point">Check Point Software Technologies Ltd.</organization>
<address>
<postal>
<street>5 Hasolelim st.</street>
<city>Tel Aviv</city>
<code>67897</code>
<country>Israel</country>
</postal>
<email>ynir@checkpoint.com</email>
</address>
</author>
<date year="2012"/>
<area>Security Area</area>
<keyword>Internet-Draft</keyword>
<abstract>
<t> This document describes using TCP for IKE messages. This facilitates the transport of
large messages over paths where fragments are either dropped, or packet loss makes them
unreliable.</t>
</abstract>
</front>
<middle>
<!-- ====================================================================== -->
<section anchor="introduction" title="Introduction">
<t> The Internet Key Exchange (IKE) specified in <xref target="RFC2407" /> and
<xref target="RFC2408" />, and IKEv2 as specified in <xref target="RFC5996"/> uses UDP to
transport the exchange messages. Some of those messages may be fairly large. Specifically,
the 5th and 6th messages of IKEv1 Main Mode, the first and second messages of IKEv1
Aggressive Mode, and the messages of IKEv2 IKE_AUTH exchange can become quite large, as
they may contain a chain of certificates, a signature payload (called "Auth" in IKEv2),
CRLs, and in the case of IKEv2, some configuration information that is carried in the CFG
payload.</t>
<t> When such UDP packets exceed the path MTU, they get fragmented. This increases the
probability of packets getting dropped, but the retransmission mechanisms in IKE (as
described in section 2.1 of RFC 5996) takes care of that. More recently we have seen a
number of service providers dropping fragmented packets. Firewalls and NAT devices need to
keep state for each packet where some but not all of the fragments have been received. This
creates a burden in terms of memory, especially for high capacity devices such as
Carrier-Grade NAT (CGN) or high capacity firewalls.</t>
<t> The BEHAVE working group has an Internet Draft describing required behavior of CGNs
(<xref target="I-D.ietf-behave-lsn-requirements"/>). It requires CGNs to comply with
<xref target="RFC4787"/>, which in section 11 requires NAT devices to support fragments.
However, some people deploying IKE have found that some ISPs have begun to drop fragments
in preparation for deploying CGNs. While we all hope for a future where all devices comply
with the emerging standards, or even a future where CGNs are not required, we have to make
IKE work today.</t>
<t> The solution described in this document is to transport the IKE messages over a TCP
(<xref target="RFC0793"/>) rather than over UDP. IKE packets (both versions) describe their
own length, so they are well-suited for transport over a stream-based connection such as
TCP. The Initiator opens a TCP connection to the Responder's port 500, sends the requests
and receives the responses, and then closes the connection. TCP can handle arbitrary-length
messages, works well with any sized data, and is well supported by all ISP infrastructure.</t>
<section anchor="nongoals" title="Non-Goals of this Specification">
<t> Firewall traversal is not a goal of this specification. If a firewall has a policy to
block IKE and/or IPsec, hiding the IKE exchange in TCP is not expected to help. Some
implementations hide both IKE and IPsec in a TCP connection, usually pretending to be
HTTPS by using port 443. This has a significant impact on bandwidth and gateway capacity,
and even this is defeated by better firewalls. SSL VPNs tunnel IP packets over TLS, but
the latest firewalls are also TLS proxies, and are able to defeat this as well.</t>
<t> This document is not part of that arms race. It is only meant to allow IKE to work When
faced with broken infrastructure that drops large IP packets.</t>
</section>
<section anchor="mustshouldmay" title="Conventions Used in This Document">
<t>The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT",
"RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described
in <xref target="RFC2119"/>.</t>
</section>
</section>
<section anchor="proto" title="The Protocol">
<section anchor="init" title="Initiator">
<t> An Initiator MAY try IKE using TCP for any request. It opens a TCP connection from an
arbitrary port to port 500 of the Responder. When the three-way handshake completes, the
Initiator MUST send the request. If the Initiator knows that this request is the last
request needed at this time, it SHOULD half-close the TCP connection, although it MAY wait
until the last response is received. When all responses have been received, the Initiator
MUST close the connection. If the peer has closed the connection before all requests have
been transmitted or responded to, the Initiator SHOULD either open a new TCP connection or
transmit them over UDP again.</t>
<t> It MUST accept responses sent over IKE within the same connection, but MUST also accept
responses over other transports, if the request had been sent over them as well.</t>
</section>
<section anchor="resp" title="Responder">
<t> A Responder MAY accept TCP connections to port 500, and if it does, it MUST accept IKE
requests over this connection. Responses to requests received over this connection MUST
also go over this connection. If the connection has closed before the Responder had had a
chance to respond, it MUST NOT respond over UDP, but MUST instead wait for a retransmission
over UDP or over another TCP connection.</t>
<t> The responder MUST accept different requests on different transports. Specifically, the
Responder MUST NOT rely on subsequent requests coming over the same transport. For example,
it is entirely acceptable to have the first two requests on IKE Main Mode come over UDP
port 500, while the last request comes over TCP, and the following Quick Mode request might
come over UDP port 4500 (because NAT has been detected).</t>
<t> A responder that receives an IKEv2 Initial request over any other transport MUST send an
IKE_TCP_SUPPORTED notification (<xref target="tcp_notif" />) in the Initial response. the
responder MAY send this notification even if the Initial request was received over TCP.</t>
<t> If the responder has some requests of its own to send, it MUST NOT use a connection that
has been opened by a peer. Instead, it MUST either use UDP or else open a new TCP connection
to the original Initiator's TCP port 500.</t>
<t> The normal flow of things is that the Initiator opens a connection and closes its side
first. The responder closes after sending the last response where the initiator has already
half-closed the connection. If, however, a significant amount of time has passed, and
neither new requests arrive nor the connection is closed by the initiator, the Responder
MAY close or even reset the connection.</t>
<t> This specification makes no recommendation as to how long such a timeout should be, but
a few seconds should be enough.</t>
</section>
<section anchor="trans" title="Transmitter">
<t> The transmitter, whether an initiator transmitting a request or a responder transmitting
a response MUST NOT retransmit over the same connection. TCP takes care of that. It SHOULD
send the IKE header and the IKE payloads with a single command or in rapid succession,
because the receiver might block on reading from the socket.</t>
</section>
<section anchor="recv" title="Receiver">
<t> The IKE header is copied from RFC 5996 below for reference:</t>
<figure>
<artwork><![CDATA[
1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| IKE SA Initiator's SPI |
| |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| IKE SA Responder's SPI |
| |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Next Payload | MjVer | MnVer | Exchange Type | Flags |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Message ID |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Length |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
Figure 1: IKE Header Format
]]></artwork>
</figure>
<t> The receiver MUST first read in the 28 bytes that make up the IKE header. The Responder
then subtracts 28 from the length field, and reads the resulting number of bytes. The
combined message, comprised on 28 header bytes and whatever number of payload bytes is
processed the same way as regular UDP messages. That includes retransmission detection,
with one slight difference: if a retransmitted request is detected, the response is
retransmitted as well, but using the current TCP connection rather than whatever other
transport had been used for the original transmission of the request.</t>
</section>
<section anchor="tcp_notif" title="IKE_TCP_SUPPORTED Notification">
<t> This notification is sent by a responder over non-TCP transports to inform the
initiator that this specification is supported.</t>
<t> The Notify payload is formatted as follows:</t>
<figure>
<artwork><![CDATA[
1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
! Next Payload !C! RESERVED ! Payload Length !
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
! Protocol ID ! SPI Size !IKE_TCP_SUPPORTED Message Type !
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
]]></artwork>
</figure>
<t><list style="symbols">
<t>Protocol ID (1 octet) MUST be 0.</t>
<t>SPI Size (1 octet) MUST be zero, in conformance with section 3.10 of
<xref target="RFC5996"/>.</t>
<t>IKE_TCP_SUPPORTED Notify Message Type (2 octets) - MUST be xxxxx, the value assigned
for IKE_TCP_SUPPORTED. TBA by IANA.</t></list></t>
</section>
</section>
<section anchor="ops" title="Operational Considerations">
<t> Most IKE messages are relatively short. Quick Mode in IKEv1, and all but the IKE_AUTH
exchange in IKEv2 are comprised of short messages that fit in a single packet on most
networks. It is only the IKE_AUTH exchange in IKEv2, and the two last messages of Main Mode
that are long. UDP has advantages in lower latency and lower resource consumption, so it
makes sense to use UDP whenever TCP is not required.</t>
<t> The requirements in <xref target="resp"/> mean that different requests may be sent over
different transports. So the initiator can choose the transport on a per-request basis. So
one obvious policy would be to do everything over UDP except the specific requests that
tend to become too big. This way the first messages use UDP, and the Initiator can set up
the TCP connection at the same time, eliminating the latency penalty of using TCP. This may
not always be the most efficient policy, though. It means that the first messages sent over
TCP are relatively large ones, and TCP slow start may cause an extra roundtrip, because the
message may exceed the transmission window. An initiator using this policy MUST NOT go to
TCP if the responder has not indicated support by sending the IKE_TCP_SUPPORTED
notification (<xref target="tcp_notif" />) in the Initial response.</t>
<t> An alternative method, that is probably easier for the Initiator to implement, is to do
an entire "mission" using the same transport. So if TCP is needed and an IKE SA has not
yet been created, the Initiator will open a TCP connection, and perform all 2-4 requests
needed to set up a child SA over the same connection.</t>
<t> Yet another policy would be to begin by using UDP, and at the same time set up the TCP
connection. If at any point the TCP handshake completes, the next requests go over that
connection. This method can be used to auto-discover support of TCP on the responder. This
is easier for the user than configuring which peers support TCP, but has the potential of
wasting resources, as TCP connections may finish the three-way handshake just when IKE over
UDP has finished. The requirements from the responder ensure that all these policies will
work.</t>
<section anchor="dpd" title="Liveness Check">
<t> The TCP connections described in this document are short-lived. We do not expect them
to stay for the lifetime of the SA, but to get torn down by either side within seconds
of the SA being set up. Because of this, they are not well-suited for the transport of
short requests such as those for liveness check.</t>
<t> Although liveness checks MAY be sent over TCP, this is not recommended.</t>
</section>
</section>
<section anchor="security" title="Security Considerations">
<t> Most of the security considerations for IKE over TCP are the same as those for UDP as in
RFC 5996.</t>
<t> For the Responder, listening to TCP port 500 involves all the risks of maintaining any
TCP server. Precautions against DoS attacks, such as SYN cookies are RECOMMENDED.</t>
</section>
<section anchor="iana" title="IANA Considerations">
<t> IANA is requested to assign a notify message type from the status types range
(16418-40959) of the "IKEv2 Notify Message Types" registry with name "IKE_TCP_SUPPORTED"</t>
<t> No IANA action is required for the TCP port, as TCP port 500 is already allocated to
"ISAKMP".</t>
</section>
</middle>
<!-- ====================================================================== -->
<back>
<references title="Normative References">
&RFC2119;
&RFC5996;
&RFC2407;
&RFC2408;
</references>
<references title="Informative References">
&RFC0793;
&RFC4787;
&I-D.behave-lsn-requirements;
</references>
<!-- ====================================================================== -->
</back>
</rfc>
| PAFTECH AB 2003-2026 | 2026-04-24 03:30:15 |