One document matched: draft-ietf-behave-turn-16.xml
<?xml version="1.0" encoding="US-ASCII"?>
<!DOCTYPE rfc SYSTEM "rfc2629.dtd">
<?rfc toc="yes"?>
<?rfc compact='yes'?>
<?rfc subcompact='no'?>
<?rfc symrefs="yes"?>
<rfc category="std" docName="draft-ietf-behave-turn-16"
ipr="pre5378Trust200902">
<front>
<title abbrev="TURN">Traversal Using Relays around NAT (TURN): Relay
Extensions to Session Traversal Utilities for NAT (STUN)</title>
<author fullname="Jonathan Rosenberg" initials="J." surname="Rosenberg">
<organization abbrev="Cisco">Cisco Systems, Inc.</organization>
<address>
<postal>
<street></street>
<city>Edison</city>
<region>NJ</region>
<country>USA</country>
</postal>
<email>jdrosen@cisco.com</email>
<uri>http://www.jdrosen.net</uri>
</address>
</author>
<author fullname="Rohan Mahy" initials="R." surname="Mahy">
<organization>(Unaffiliated)</organization>
<address>
<email>rohan@ekabal.com</email>
</address>
</author>
<author fullname="Philip Matthews" initials="P." surname="Matthews">
<organization>Alcatel-Lucent</organization>
<address>
<postal>
<street>600 March Road</street>
<city>Ottawa</city>
<region>Ontario</region>
<code></code>
<country>Canada</country>
</postal>
<phone></phone>
<facsimile></facsimile>
<email>philip_matthews@magma.ca</email>
<uri></uri>
</address>
</author>
<date year="2009" />
<area>Transport</area>
<workgroup>BEHAVE WG</workgroup>
<keyword>NAT</keyword>
<keyword>TURN</keyword>
<keyword>STUN</keyword>
<keyword>ICE</keyword>
<abstract>
<t>If a host is located behind a NAT, then in certain situations it can
be impossible for that host to communicate directly with other hosts
(peers). In these situations, it is necessary for the host to use the
services of an intermediate node that acts as a communication relay.
This specification defines a protocol, called TURN (Traversal Using
Relays around NAT), that allows the host to control the operation of the
relay and to exchange packets with its peers using the relay. TURN
differs from some other relay control protocols in that it allows a
client to communicate with multiple peers using a single relay
address.</t>
<t>The TURN protocol was designed to be used as part of the ICE
(Interactive Connectivity Establishment) approach to NAT traversal,
though it can be also used without ICE.</t>
</abstract>
</front>
<middle>
<section title="Introduction">
<t>A host behind a NAT may wish to exchange packets with other hosts,
some of which may also be behind NATs. To do this, the hosts involved
can use 'Hole Punching' techniques (see <xref target="RFC5128"></xref>)
in an attempt discover a direct communication path; that is, a
communication path that goes from host to another through intervening
NATs and routers, but does not traverse any relays.</t>
<t>As described in <xref target="RFC5128"></xref> and <xref
target="RFC4787"></xref>, hole punching techniques will fail if both
hosts are behind NATs that are not well-behaved. For example, if both
hosts are behind NATs that have a mapping behavior of "address dependent
mapping" or "address and port dependent mapping", then hole punching
techniques generally fail.</t>
<t>When a direct communication path cannot be found, it is necessary to
use the services of an intermediate host that acts as a relay for the
packets. This relay typically sits in the public Internet and relays
packets between two hosts that both sit behind NATs.</t>
<t>This specification defines a protocol, called TURN, that allows a
host behind a NAT (called the TURN client) to request that another host
(called the TURN server) act as a relay. The client can arrange for the
server to relay packets to and from certain other hosts (called peers)
and can control aspects of how the relaying is done. The client does
this by obtaining an IP address and port on the server, called the
relayed-transport-address. When a peer sends a packet to the
relayed-transport-address, the server relays the packet to the client.
When the client sends a data packet to the server, the server relays it
to the appropriate peer using the relayed-transport-address as the
source.</t>
<t>A client using TURN must have some way to communicate the
relayed-transport-address to its peers, and to learn each peer's IP
address and port (more precisely, each peer's server-reflexive transport
address, see <xref target="sec-overview"></xref>). How this is done is
out of the scope of the TURN protocol. One way this might be done is for
the client and peers to exchange e-mail messages. Another way is for the
client and its peers to use a special-purpose 'introduction' or
'rendezvous' protocol (see <xref target="RFC5128"></xref> for more
details).</t>
<t>If TURN is used with ICE <xref target="I-D.ietf-mmusic-ice"></xref>,
then the relayed-transport-address and the IP addresses and ports of the
peers are included in the ICE candidate information which the rendezvous
protocol must carry. For example, if TURN and ICE are used as part of a
multimedia solution using SIP <xref target="RFC3261"></xref>, then SIP
serves the role of the rendezvous protocol, carrying the ICE candidate
information inside the body of SIP messages. If TURN and ICE are used
with some other rendezvous protocol, then <xref
target="I-D.rosenberg-mmusic-ice-nonsip"></xref> provides guidance on
the services the rendezvous protocol must perform.</t>
<t>Though the use of a TURN server to enable communication between two
hosts behind NATs is very likely to work, it comes at a high cost to the
provider of the TURN server, since the server typically needs a high
bandwidth connection to the Internet . As a consequence, it is best to
use a TURN server only when a direct communication path cannot be found.
When the client and a peer use ICE to determine the communication path,
ICE will use hole punching techniques to search for a direct path first
and only use a TURN server when a direct path cannot be found.</t>
<t>TURN was originally invented to support multimedia sessions signaled
using SIP. Since SIP supports forking, TURN supports multiple peers per
relayed-transport-address; a feature not supported by other approaches
(e.g., SOCKS <xref target="RFC1928"></xref>). However, care has been
taken to make sure that TURN is suitable for other types of
applications.</t>
<t>TURN was designed as one piece in the larger ICE approach to NAT
traversal. Implementors of TURN are urged to investigate ICE and
seriously consider using it for their application. However, it is
possible to use TURN without ICE.</t>
<t>TURN is an extension to the STUN (Session Traversal Utilities for NAT
<xref target="RFC5389"></xref>) protocol. Most, though not all, TURN
messages are STUN-formatted messages. A reader of this document should
be familiar with STUN.</t>
</section>
<section anchor="sec-overview" title="Overview of Operation">
<t>This section gives an overview of the operation of TURN. It is
non-normative.</t>
<t>In a typical configuration, a TURN client is connected to a <xref
target="RFC1918">private network</xref> and through one or more NATs to
the public Internet. On the public Internet is a TURN server. Elsewhere
in the Internet are one or more peers that the TURN client wishes to
communicate with. These peers may or may not be behind one or more NATs.
The client uses the server as a relay to send packets to these peers and
to receive packets from these peers.</t>
<figure anchor="fig-turn-model">
<artwork><![CDATA[ Peer A
Server-Reflexive +---------+
Transport Address | |
192.0.2.150:32102 | |
| /| |
TURN | / ^| Peer A |
Client's Server | / || |
Host Transport Transport | // || |
Address Address | // |+---------+
10.1.1.2:49721 192.0.2.15:3478 |+-+ // Peer A
| | ||N| / Host Transport
| +-+ | ||A|/ Address
| | | | v|T| 192.168.100.2:49582
| | | | /+-+
+---------+| | | |+---------+ / +---------+
| || |N| || | // | |
| TURN |v | | v| TURN |/ | |
| Client |----|A|----------| Server |------------------| Peer B |
| | | |^ | |^ ^| |
| | |T|| | || || |
+---------+ | || +---------+| |+---------+
| || | |
| || | |
+-+| | |
| | |
| | |
Client's | Peer B
Server-Reflexive Relayed Transport
Transport Address Transport Address Address
192.0.2.1:7000 192.0.2.15:50000 192.0.2.210:49191
]]></artwork>
</figure>
<t></t>
<t><xref target="fig-turn-model"></xref> shows a typical deployment. In
this figure, the TURN client and the TURN server are separated by a NAT,
with the client on the private side and the server on the public side of
the NAT. This NAT is assumed to be a “bad” NAT; for example,
it might have a mapping property of address-and-port-dependent mapping
(see <xref target="RFC4787"></xref> for a description of what this
means).</t>
<t>The client talks to the server from a (IP address, port) combination
called the client's HOST TRANSPORT ADDRESS. (The combination of an IP
address and port is called a TRANSPORT ADDRESS).</t>
<t>The client sends TURN messages from its host transport address to a
transport address on the TURN server which is known as the TURN SERVER
TRANSPORT ADDRESS. The client learns the server’s transport
address through some unspecified means (e.g., configuration), and this
address is typically used by many clients simultaneously.</t>
<t>Since the client is behind a NAT, the server sees packets from the
client as coming from a transport address on the NAT itself. This
address is known as the client’s SERVER-REFLEXIVE transport
address; packets sent by the server to the client’s
server-reflexive transport address will be forwarded by the NAT to the
client’s host transport address.</t>
<t>The client uses TURN commands to create and manipulate an ALLOCATION
on the server. An allocation is a data structure on the server, an
important component of which is a RELAYED TRANSPORT ADDRESS. The relayed
transport address for the allocation is a transport address on the
server which is used to send and receive packets to the peers.</t>
<t>Once an allocation is created, the client can send application data
to the server along with an indication of which peer the data is to be
sent to, and the server will relay this data to the appropriate peer.
The client sends the application data to the server inside a TURN
message; at the server, the data is extracted from the TURN message and
sent to the peer in a UDP datagram. In the reverse direction, a peer can
send application data in a UDP datagram to the relayed transport address
for the allocation; the server will then encapsulate this data inside a
TURN message and send it to the client along with an indication of which
peer sent the data. Since the TURN message always contains an indication
of which peer the client is communicating with, the client can use a
single allocation to communicate with multiple peers.</t>
<t>When the peer is behind a NAT, then the client must identify the peer
using its server-reflexive transport address rather than its host
transport address. For example, to send application data to peer A in
the example above, the client must specify 192.0.2.150:32102 (peer A's
server-reflexive transport address) rather than 192.168.100.2:49582
(peer A's host transport address).</t>
<t>Each allocation on the server belongs to a single client and has
exactly one relayed transport address which is used only by that
allocation. Thus when a packet arrives at a relayed transport address on
the server, the server knows which client the data is intended for.
However, the client may have multiple allocations on a server at the
same time.</t>
<section anchor="sec-transports" title="Transports">
<t>TURN as defined in this specification always uses UDP between the
server and the peer. However, this specification allows the use of any
one of UDP, TCP, or TLS over TCP to carry the TURN messages between
the client and the server.</t>
<texttable>
<ttcol align="center">TURN client to TURN server</ttcol>
<ttcol align="center">TURN server to peer</ttcol>
<c>UDP</c>
<c>UDP</c>
<c>TCP</c>
<c>UDP</c>
<c>TLS over TCP</c>
<c>UDP</c>
</texttable>
<t>If TCP or TLS over TCP is used between the client and the server,
then the server will convert between these transports and UDP
transport when relaying data to/from the peer.</t>
<t>Since this version of TURN only supports UDP between the server and
the peer, it is expected that most clients will prefer to also use UDP
between the client and the server. That being the case, some readers
may wonder: Why also support TCP and TLS over TCP?</t>
<t>TURN supports TCP transport between the client and the server
because some firewalls are configured to block UDP entirely. These
firewalls block UDP but not TCP in part because TCP has properties
that make the intention of the nodes being protected by the firewall
more obvious to the firewall. For example, TCP has a three-way
handshake that makes in clearer that the protected node really wishes
to have that particular connection established, while for UDP the best
the firewall can do is guess which flows are desired by using
filtering rules. Also, TCP has explicit connection teardown, while for
UDP the firewall has to use timers to guess when the flow is
finished.</t>
<t>TURN supports TLS over TCP transport between the client and the
server because TLS provides additional security properties not
provided by TURN's default digest authentication; properties which
some clients may wish to take advantage of. In particular, TLS
provides a way for the client to ascertain that it is talking to the
server that it intended to, and also provides for confidentiality of
TURN control messages. TURN does not require TLS because the overhead
of using TLS is higher than that of digest authentication; for
example, using TLS likely means that most application data will be
doubly encrypted (once by TLS and once to ensure it is still encrypted
in the UDP datagram).</t>
<t>There is a planned extension to TURN to add support for TCP between
the server and the peers <xref
target="I-D.ietf-behave-turn-tcp"></xref>. For this reason,
allocations that use UDP between the server and the peers are known as
UDP allocations, while allocations that use TCP between the server and
the peers are known as TCP allocations. This specification describes
only UDP allocations.</t>
<t>TURN as defined in this specification only supports IPv4. All IP
addresses in this specification must be IPv4 addresses. However, there
is a planned extension to TURN to add support for IPv6 and for
relaying between IPv4 and IPv6 <xref
target="I-D.ietf-behave-turn-ipv6"></xref>.</t>
<t>In some applications for TURN, the client may send and receive
packets other than TURN packets on the host transport address it uses
to communicate with the server. This can happen, for example, when
using TURN with ICE. In these cases, the client can distinguish TURN
packets from other packets by examining the source address of the
arriving packet: those arriving from the TURN server will be TURN
packets.</t>
</section>
<section title="Allocations">
<t>To create an allocation on the server, the client uses an Allocate
transaction. The client sends a Allocate request to the server, and
the server replies with an Allocate success response containing the
allocated relayed transport address. The client can include attributes
in the Allocate request that describe the type of allocation it
desires (e.g., the lifetime of the allocation). Since relaying data
may require lots of bandwidth, the server typically requires that the
client authenticate itself using STUN’s long-term credential
mechanism, to show that it is authorized to use the server.</t>
<t>Once a relayed transport address is allocated, a client must keep
the allocation alive. To do this, the client periodically sends a
Refresh request to the server. TURN deliberately uses a different
method (Refresh rather than Allocate) for refreshes to ensure that the
client is informed if the allocation vanishes for some reason.</t>
<t>The frequency of the Refresh transaction is determined by the
lifetime of the allocation. The default lifetime of an allocation is
10 minutes -- this value was chosen to be long enough so that
refreshing is not typically a burden on the client, while expiring
allocations where the client has unexpectedly quit in a timely manner.
However, the client can request a longer lifetime in the Allocate
request and may modify its request in a Refresh request, and the
server always indicates the actual lifetime in the response. The
client must issue a new Refresh transaction within 'lifetime' seconds
of the previous Allocate or Refresh transaction. Once a client no
longer wishes to use an Allocation, it should delete the allocation
using a Refresh request with a requested lifetime of 0.</t>
<t>Both the server and client keep track of a value known as the
5-TUPLE. At the client, the 5-tuple consists of the client's host
transport address, the server transport address, and the transport
protocol used by the client to communicate with the server. At the
server, the 5-tuple value is the same except that the client's host
transport address is replaced by the client's server-reflexive
address, since that is the client's address as seen by the server.</t>
<t>Both the client and the server remember the 5-tuple used in the
Allocate request. Subsequent messages between the client and the
server uses the same 5-tuple. In this way, the client and server know
which allocation is being referred to. If the client wishes to
allocate a second relayed transport address, it must create a second
allocation using a different 5-tuple (e.g., by using a different
client host address or port).</t>
<t><list>
<t>NOTE: While the terminology used in this document refers to
5-tuples, the TURN server can store whatever identifier it likes
that yields identical results. Specifically, an implementation may
use a file-descriptor in place of a 5-tuple to represent a TCP
connection</t>
</list></t>
<figure anchor="fig-allocate">
<artwork><![CDATA[TURN TURN Peer Peer
client server A B
|-- Allocate request --------------->| | |
| | | |
|<--------------- Allocate failure --| | |
| (401 Unauthorized) | | |
| | | |
|-- Allocate request --------------->| | |
| | | |
|<---------- Allocate success resp --| | |
| (192.0.2.15:50000) | | |
// // // //
| | | |
|-- Refresh request ---------------->| | |
| | | |
|<----------- Refresh success resp --| | |
| | | |
]]></artwork>
<postamble></postamble>
</figure>
<t>In <xref target="fig-allocate"></xref>, the client sends an
Allocate request to the server without credentials. Since the server
requires that all requests be authenticated using STUN's long-term
credential mechanism, the server rejects the request with a 401
(Unauthorized) error code. The client then tries again, this time
including credentials (not shown). This time, the server accepts the
Allocate request and returns an Allocate success response containing
(amongst other things) the relayed transport address assigned to the
allocation. Sometime later the client decides to refresh the
allocation and thus sends a Refresh request to the server. The refresh
is accepted and the server replies with a Refresh success
response.</t>
</section>
<section anchor="sec-permission-overview" title="Permissions">
<t>To ease concerns amongst enterprise IT administrators that TURN
could be used to bypass corporate firewall security, TURN includes the
notion of permissions. TURN permissions mimic the address-restricted
filtering mechanism of NATs that comply with <xref
target="RFC4787"></xref>.</t>
<t>An allocation can have zero or more permissions. Each permission
consists of an IP address and a lifetime. When the server receives a
UDP datagram on the allocation's relayed transport address, it first
checks the list of permissions. If the source IP address of the
datagram matches a permission, the application data is relayed to the
client, otherwise the UDP datagram is silently discarded.</t>
<t>A permission expires after 5 minutes if it is not refreshed, and
there is no way to explicitly delete a permission. This behavior was
selected to match the behavior of a NAT that complies with <xref
target="RFC4787"></xref>.</t>
<t>The client can install or refresh a permission using either a
CreatePermission request or a ChannelBind request. Using the
CreatePermission request, multiple permissions can be installed or
refreshed with a single request -- this is important for applications
that use ICE. For security reasons, permissions can only be installed
or refreshed by transactions that can be authenticated; thus Send
indications and ChannelData messages (which are used to send data to
peers) do not install or refresh any permissions.</t>
<t>Note that permissions are within the context of an allocation, so
adding or expiring a permission in one allocation does not affect
other allocations.</t>
</section>
<section title="Send Mechanism">
<t>There are two mechanisms for the client and peers to exchange
application data using the TURN server. The first mechanism uses the
Send and Data methods, the second way uses channels. Common to both
ways is the ability of the client to communicate with multiple peers
using a single allocated relayed transport address; thus both ways
include a means for the client to indicate to the server which peer to
forward the data to, and for the server to indicate which peer sent
the data.</t>
<t>The Send mechanism uses Send and Data indications. Send indications
are used to send application data from the client to the server, while
Data indications are used to send application data from the server to
the client.</t>
<t>When using the Send mechanism, the client sends a Send indication
to the TURN server containing (a) an XOR-PEER-ADDRESS attribute
specifying the (server-reflexive) transport address of the peer and
(b) a DATA attribute holding the application data. When the TURN
server receives the Send indication, it extracts the application data
from the DATA attribute and sends it in a UDP datagram to the peer,
using the allocated relay address as the source address. Note that
there is no need to specify the relayed transport address, since it is
implied by the 5-tuple used for the Send indication.</t>
<t>In the reverse direction, UDP datagrams arriving at the relayed
transport address on the TURN server are converted into Data
indications and sent to the client, with the server-reflexive
transport address of the peer included in an XOR-PEER-ADDRESS
attribute and the data itself in a DATA attribute. Since the relayed
transport address uniquely identified the allocation, the server knows
which client to relay the data to.</t>
<t>Send and Data indications cannot be authenticated, since the
Long-Term Credential Mechanism of STUN does not support authenticating
indications. This is not as big an issue as it might first appear,
since the client-to-server leg is only half of the total path to the
peer; applications that want proper security need to use encryption or
similar to protect their data in the UDP datagrams between the server
and the peer. However, to prevent attackers from injecting rogue Send
indications to arbitrary destinations, TURN requires that a client
install a permission to a peer before sending data to it using a Send
indication.</t>
<t></t>
<figure anchor="fig-send-data">
<artwork><![CDATA[TURN TURN Peer Peer
client server A B
| | | |
|-- CreatePermission req (Peer A) -->| | |
|<-- CreatePermission success resp --| | |
| | | |
|--- Send ind (Peer A)-------------->| | |
| |=== data ===>| |
| | | |
| |<== data ====| |
|<-------------- Data ind (Peer A) --| | |
| | | |
| | | |
|--- Send ind (Peer B)-------------->| | |
| | dropped | |
| | | |
| |<== data ==================|
| dropped | | |
| | | |
]]></artwork>
<postamble></postamble>
</figure>
<t>In <xref target="fig-send-data"></xref>, the client has already
created an allocation and now wishes to send data to its peers. The
client first creates a permission by sending the server a
CreatePermission request specifying peer A's (server reflexive) IP
address in the XOR-PEER-ADDRESS attribute; if this was not done, the
server would not relay data between the client and the server. The
client then sends data to Peer A using a Send indication; at the
server, the application data is extracted and forwarded in a UDP
datagram to Peer A, using the relayed transport address as the source
transport address. When a UDP datagram from Peer A is received at the
relayed transport address, the contents are placed into a Data
indication and forwarded to the client. Later, the client attempts to
exchange data with Peer B, however no permission has been installed
for Peer B, so the Send indication from the client and the UDP
datagram from the peer are both dropped by the server.</t>
</section>
<section title="Channels">
<t>For some applications (e.g. Voice over IP), the 36 bytes of
overhead that a Send indication or Data indication adds to the
application data can substantially increase the bandwidth required
between the client and the server. To remedy this, TURN offers a
second way for the client and server to associate data with a specific
peer.</t>
<t>This second way uses an alternate packet format known as the
ChannelData message. The ChannelData message does not use the STUN
header used by other TURN messages, but instead has a 4-byte header
that includes a number known as a channel number. Each channel number
in use is bound to a specific peer and thus serves as a shorthand for
the peer's host transport address.</t>
<t>To bind a channel to a peer, the client sends a ChannelBind request
to the server, and includes an unbound channel number and the
transport address of the peer. Once the channel is bound, the client
can use a ChannelData message to send the server data destined for the
peer. Similarly, the server can relay data from that peer towards the
client using a ChannelData message.</t>
<t>Channel bindings last for 10 minutes unless refreshed -- this
lifetime was chosen to be longer than the permission lifetime. Channel
bindings are refreshed by sending another ChannelBind request
rebinding the channel to the peer. Like permissions (but unlike
allocations), there is no way to explicitly delete a channel binding;
the client must simply wait for it to time out.</t>
<figure anchor="fig-channels">
<artwork><![CDATA[TURN TURN Peer Peer
client server A B
| | | |
|-- ChannelBind req ---------------->| | |
| (Peer A to 0x4001) | | |
| | | |
|<---------- ChannelBind succ resp --| | |
| | | |
|-- [0x4001] data ------------------>| | |
| |=== data ===>| |
| | | |
| |<== data ====| |
|<------------------ [0x4001] data --| | |
| | | |
|--- Send ind (Peer A)-------------->| | |
| |=== data ===>| |
| | | |
| |<== data ====| |
|<------------------ [0x4001] data --| | |
| | | |
]]></artwork>
</figure>
<t><xref target="fig-channels"></xref> shows the channel mechanism in
use. The client has already created an allocation and now wishes to
bind a channel to peer A. To do this, the client sends a ChannelBind
request to the server, specifying the transport address of Peer A and
a channel number (0x4001). After that, the client can send application
data encapsulated inside ChannelData messages to Peer A: this is shown
as "[0x4001] data" where 0x4001 is the channel number. When the
ChannelData message arrives at the server, the server transfers the
data to a UDP datagram and sends it to the peer A, as indicated by the
channel number. When peer A sends a UDP datagram to the relayed
transport address, the data is placed inside a ChannelData message and
sent to the client.</t>
<t>Once a channel has been bound, the client is free to intermix
ChannelData messages and Send indications. In the figure, the client
later decides to use a Send indication rather than a ChannelData
message to send additional data to peer A. The client might decide to
do this, for example, so it can use the DONT-FRAGMENT attribute (see
the next section). However, once a channel is bound, the server will
always use a ChannelData message, as shown in the call flow.</t>
<t>Note that ChannelData messages can only be used for peers to which
the client has bound a channel. In the example above, Peer A has been
bound to a channel, but Peer B has not, so application data to and
from Peer B would use the Send mechanism.</t>
</section>
<section title="Unprivileged TURN Servers">
<t>This version of TURN is designed so that the server can be
implemented as an application that runs in user space under commonly
available operating systems without requiring special privileges. This
design decision was taken to make it easy to deploy a TURN server: for
example, to allow a TURN server to be integrated into a peer-to-peer
application so that one peer can offer NAT traversal services to
another peer.</t>
<t>This design decision has the following implications for data
relayed by a TURN server:<list style="symbols">
<t>The value of the Diff-Serv field may not be preserved across
the server;</t>
<t>The TTL field may be reset, rather than decremented, across the
server;</t>
<t>The ECN field may be reset by the server;</t>
<t>ICMP messages are not relayed by the server;</t>
<t>There is no end-to-end fragmentation, since the packet is
re-assembled at the server.</t>
</list>Future work may specify alternate TURN semantics that address
these limitations.</t>
</section>
<section title="Avoiding IP Fragmentation">
<t>For reasons described in <xref target="Frag-Harmful"></xref>,
applications, especially those sending large volumes of data, should
try hard to avoid having their packets fragmented. Applications using
TCP can more-or-less ignore this issue because fragmentation avoidance
is now a standard part of TCP, but applications using UDP (and thus
any application using this version of TURN) must handle fragmentation
avoidance themselves.</t>
<t>The application running on the client and the peer can take one of
two approaches to avoid IP fragmentation.</t>
<t>The first approach is to avoid sending large amounts of application
data in the TURN messages/UDP datagrams exchanged between the client
and the peer. This is the approach taken by most VoIP (Voice-over-IP)
applications. In this approach, the application exploits the fact that
the IP specification <xref target="RFC0791"></xref> specifies that IP
packets up to 576 bytes should never need to be fragmented.</t>
<t>The exact amount of application data that can be included while
avoiding fragmentation depends the details of the TURN session between
the client and the server: whether UDP, TCP, or TLS transport is used,
whether ChannelData messages or Send/Data indications are used, and
whether any additional attributes (such as the DONT-FRAGMENT
attribute) are included. Another factor, which is hard to determine,
is whether the MTU is somewhere along the path is reduced for other
reasons, such as the use of IP-in-IP tunneling.</t>
<t>As a guideline, sending a maximum of 500 bytes of application data
in a single TURN message (by the client on the client-to-server leg)
or a UDP datagram (by the peer on the peer-to-server leg) will
generally avoid IP fragmentation. To further reduce the chance of
fragmentation, it is recommended that the client use ChannelData
messages when transferring significant volumes of data, since the
overhead of the ChannelData message is less than Send and Data
indications.</t>
<t>The second approach the client and peer can take to avoid
fragmentation is to use a path MTU discovery algorithm to determine
the maximum amount of application data than can be sent without
fragmentation.</t>
<t>Unfortunately, because servers implementing this version of TURN do
not relay ICMP messages, the classic Path MTU Discovery algorithm
defined in <xref target="RFC1191"></xref> is not able to discover the
MTU of the transmission path between the client and the peer. (Even if
they did relay ICMP messages, the algorithm would not always work
since ICMP messages are often filtered out by combined NAT/firewall
devices).</t>
<t>So the client and server need to use a path MTU discovery algorithm
that does not require ICMP messages. The Packetized Path MTU Discovery
algorithm defined in <xref target="RFC4821"></xref> is one such
algorithm.</t>
<t>The details of how to use the algorithm of <xref
target="RFC4821"></xref> with TURN are still under investigation.
However, as a step towards this goal, this version of TURN supports a
DONT-FRAGMENT attribute. When the client includes this attribute in a
Send indication, this tells the server to set the DF bit in the
resulting UDP datagram that it sends to the peer. Since some servers
may be unable to set the DF bit, the client should also include this
attribute in the Allocate request -- any server that does not support
the DONT-FRAGMENT attribute will indicate this by rejecting the
Allocate request.</t>
</section>
<section title="RTP Support">
<t>One of the envisioned uses of TURN is as a relay for clients and
peers wishing to exchange real-time data (e.g. voice or video) using
RTP. To facilitate the use of TURN for this purpose, TURN includes
some special support for older versions of RTP.</t>
<t>Old versions of RTP <xref target="RFC3550"></xref> required that
the RTP stream be on an even port number and the associated RTCP
stream, if present, be on the next highest port. To allow clients to
work with peers that still require this, TURN allows the client to
request that the server allocate a relayed-transport-address with an
even port number, and to optionally request the server reserve the
next-highest port number for a subsequent allocation.</t>
</section>
<section title="Anycast Discovery of Servers">
<t>This version of TURN has been designed to permit the future
specification of a method of doing anycast discovery of a TURN server
over UDP.</t>
<t>Specifically, a TURN server can reject an Allocate request with the
suggestion that the server try an alternate server. To avoid certain
types of attacks, the client must use the same credentials with the
alternate server as it would have with the initial server.</t>
</section>
</section>
<section title="Terminology">
<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">RFC 2119</xref>.</t>
<t>Readers are expected to be familiar with <xref
target="RFC5389"></xref> and the terms defined there.</t>
<t>The following terms are used in this document:</t>
<t><list style="hanging">
<t hangText="TURN:">The protocol spoken between a TURN client and a
TURN server. It is an extension to the STUN protocol <xref
target="RFC5389"></xref>. The protocol allows a client to allocate
and use a relayed transport address.</t>
<t hangText="TURN client:">A STUN client that implements this
specification.</t>
<t hangText="TURN server:">A STUN server that implements this
specification. It relays data between a TURN client and its
peer(s).</t>
<t hangText="Peer:">A host with which the TURN client wishes to
communicate. The TURN server relays traffic between the TURN client
and its peer(s). The peer does not interact with the TURN server
using the protocol defined in this document; rather, the peer
receives data sent by the TURN server and the peer sends data
towards the TURN server.</t>
<t hangText="Transport Address:">The combination of an IP address
and a port.</t>
<t hangText="Host Transport Address:">A transport address on a
client or a peer.</t>
<t hangText="Server-Reflexive Transport Address:">A transport
address on the "public side" of a NAT. This address is allocated by
the NAT to correspond to a specific host transport address.</t>
<t hangText="Relayed Transport Address:">A transport address on the
TURN server that is used for relaying packets between the client and
a peer. A peer sends to this address on the TURN server, and the
packet is then relayed to the client.</t>
<t hangText="TURN Server Transport Address:">A transport address on
the TURN server that is used for sending TURN messages to the
server. This is the transport address that the client uses to
communicate with the server.</t>
<t hangText="Peer Transport Address:">The transport address of the
peer as seen by the server. When the peer is behind a NAT, this is
the peer's server-reflexive transport address.</t>
<t hangText="Allocation:">The relayed transport address granted to a
client through an Allocate request, along with related state, such
as permissions and expiration timers.</t>
<t hangText="5-tuple:">The combination (client IP address and port,
server IP address and port, and transport protocol (currently one of
UDP, TCP, or TLS)) used to communicate between the client and the
server. The 5-tuple uniquely identifies this communication stream.
The 5-tuple also uniquely identifies the Allocation on the
server.</t>
<t hangText="Channel:">A channel number and associated peer
transport address. Once a channel number is bound to a peer's
transport address, the client and server can use the more
bandwidth-efficient ChannelData message to exchange data.</t>
<t hangText="Permission:">The IP address and transport protocol (but
not the port) of a peer that is permitted to send traffic to the
TURN server and have that traffic relayed to the TURN client. The
TURN server will only forward traffic to its client from peers that
match an existing permission.</t>
<t hangText="Realm">A string used to describe the server or a
context within the server. The realm tells the client which username
and password combination to use to authenticate requests.</t>
<t hangText="Nonce">A string chosen at random by the server and
included in the message-digest. To prevent reply attacks, the server
should change the nonce regularly.</t>
</list></t>
</section>
<!-- Overview -->
<section anchor="sec-general-behavior" title="General Behavior">
<t>This section contains general TURN processing rules that apply to all
TURN messages.</t>
<t>TURN is an extension to STUN. All TURN messages, with the exception
of the ChannelData message, are STUN-formatted messages. All the base
processing rules described in <xref target="RFC5389"></xref> apply to
STUN-formatted messages. This means that all the message-forming and
-processing descriptions in this document are implicitly prefixed with
the rules of <xref target="RFC5389"></xref>.</t>
<t><xref target="RFC5389"></xref> specifies a Long-Term Credential
mechanism for STUN. TURN servers and clients MUST implement this
mechanism. The server SHOULD demand that all requests from the client be
authenticated using this mechanism, and the client MUST be prepared to
authenticate requests if required.</t>
<t><list>
<t>In general, it is strongly recommended that servers require
requests to be authenticated, as the security of TURN can otherwise
be quite weak. One reason that a server might not require requests
to be authenticated is that TURN is being used in a carefully
controlled environment in which the risks of unauthenticated
requests by hostile third-parties have been mitigated. See <xref
target="sec-security"></xref> for more discussion on this point.</t>
</list></t>
<t>Note that Long-Term Credential mechanism applies only to requests and
cannot be used to authenticate indications, thus indications in TURN are
never authenticated. If the server requires requests to be
authenticated, then the server's administrator MUST choose a realm value
that will uniquely identify the username and password combination that
the client must use, even if the client uses multiple servers under
different administrations. The server's administrator MAY choose to
allocate a unique username to each client, or MAY choose to allocate the
same username to more than one client (for example, to all clients from
the same department or company). For each allocation, the server SHOULD
generate a new random nonce when the allocation is first attempted
following the randomness recommendations in <xref
target="RFC4086"></xref> and SHOULD expire the nonce at least once every
hour during the lifetime of the allocation.</t>
<t>All requests after the initial Allocate must use the same username as
that used to create the allocation, to prevent attackers from hijacking
the client's allocation. Specifically, if the server requires the use of
the Long-Term Credential mechanism, and if a non-Allocate request passes
authentication under this mechanism, and if the 5-tuple identifies an
existing allocation, but the request does not use the same username as
used to create the allocation, then the request MUST be rejected with a
441 (Wrong Credentials) error.</t>
<t>When a TURN message arrives at the server from the client, the server
uses the 5-tuple in the message to identify the associated allocation.
For all TURN messages (including ChannelData) EXCEPT an Allocate
request, if the 5-tuple does not identify an existing allocation, then
the message MUST either be rejected with a 437 Allocation Mismatch error
(if it is a request), or silently ignored (if it is an indication or a
ChannelData message). A client receiving a 437 error response to a
request other than Allocate MUST assume the allocation no longer
exists.</t>
<t>The client SHOULD include the SOFTWARE attribute in all Allocate and
Refresh requests and MAY include it in any other requests or
indications. The server SHOULD include the SOFTWARE attribute in all
Allocate and Refresh responses (either success or failure) and MAY
include it in other responses or indications. The client and the server
MAY include the FINGERPRINT attribute in any STUN-formatted messages
defined in this document.</t>
<t>TURN does not use the backwards-compatibility mechanism described in
<xref target="RFC5389"></xref>.</t>
<t>TURN as defined in this specification only supports IPv4. The
client's IP address, the server's IP address and all IP addresses
appearing in a relayed-transport-address MUST be IPv4 addresses.</t>
<t>By default, TURN runs on the same ports as STUN: 3478 for TURN over
UDP and TCP, and 5349 for TURN over TLS. However, TURN has its own set
of SRV service names: "turn" for UDP and TCP, and "turns" for TLS.
Either the SRV procedures or the ALTERNATE-SERVER procedures, both
described in <xref target="sec-create-allocation"></xref>, can be used
to run TURN on a different port.</t>
<t>To ensure interoperability, a TURN server MUST support the use of UDP
transport between the client and the server, and SHOULD support the use
of TCP and TLS transport.</t>
<t>When UDP transport is used between the client and the server, the
client will retransmit a request if it does not receive a response
within a certain timeout period. Because of this, the server may receive
two (or more) requests with the same 5-tuple and same transaction id.
STUN requires that the server recognize this case and treat the request
as idempotent (see <xref target="RFC5389"></xref>). Some implementations
may choose to meet this requirement by remembering all received requests
and the corresponding responses for 40 seconds. Other implementations
may choose to reprocess the request and arrange that such reprocessing
returns essentially the same response. To aid implementors who choose
the latter approach (the so-called "stateless stack approach"), this
specification includes some implementation notes on how this might be
done. Implementations are free to choose either approach or choose some
other approach that gives the same results.</t>
<t>When TCP transport is used between the client and the server, it is
possible that a bit error will cause a length field in a TURN packet to
become corrupted, causing the receiver to lose synchronization with the
incoming stream of TURN messages. A client or server which detects a
long sequence of invalid TURN messages over TCP transport SHOULD close
the corresponding TCP connection to help the other end detect this
situation more rapidly.</t>
<t>To mitigate either intentional or unintentional denial-of-service
attacks against the server by clients with valid usernames and
passwords, it is RECOMMENDED that the server impose limits on both the
number of allocations active at one time for a given username and on the
amount of bandwidth those allocations can use. The server should reject
new allocations that would exceed the limit on the allowed number of
allocations active at one time with a 486 (Allocation Quota Exceeded)
(see <xref target="sec-rcv-allocate"></xref>), and should discard
application data traffic that exceeds the bandwidth quota.</t>
</section>
<section anchor="sec-allocations" title="Allocations">
<t>All TURN operations revolve around allocations, and all TURN messages
are associated with an allocation. An allocation conceptually consists
of the following state data:<list style="symbols">
<t>the relayed transport address</t>
<t>The 5-tuple: (client's IP address, client's port, server IP
address, server port, transport protocol)</t>
<t>the authentication information</t>
<t>the time-to-expiry</t>
<t>A list of permissions</t>
<t>A list of channel to peer bindings</t>
</list>The relayed transport address is the transport address
allocated by the server for communicating with peers, while the 5-tuple
describes the communication path between the client and the server. On
the client, the 5-tuple uses the client's host transport address, while
on the server the 5-tuple uses the client's server-reflexive transport
address.</t>
<t>Both the relayed-transport-address and the 5-tuple MUST be unique
across all allocations, so either one can be used to uniquely identify
the allocation.</t>
<t>The authentication information (e.g., username, password, realm, and
nonce) are used to both verify subsequent requests and to compute the
message integrity of responses. The username, realm, and nonce values
are initially those used in the authenticated Allocate request that
creates the allocation, though the server can change the nonce value
during the lifetime of the allocation using a 438 (Stale Nonce) reply.
Note that rather than storing the password explicitly, it may be
desirable for security reasons for the server to store the key value
which is an MD5 hash over the username, realm and password (see <xref
target="RFC5389"></xref>).</t>
<t>The time-to-expiry is the time in seconds left until the allocation
expires. Each Allocate or Refresh transaction sets this timer, which
then ticks down towards 0. By default, each Allocate or Refresh
transaction resets this timer to the default lifetime value of 600
seconds (10 minutes), but the client can request a different value in
the Allocate and Refresh request. Allocations can only be refreshed
using the Refresh request; sending data to a peer does not refresh an
allocation. When an allocation expires, the state data associated with
the allocation can be freed.</t>
<t>The list of permissions is described in <xref
target="sec-permissions"></xref> and the list of channels is described
in <xref target="sec-channels"></xref>.</t>
</section>
<section anchor="sec-create-allocation" title="Creating an Allocation">
<t>An allocation on the server is created using an Allocate
transaction.</t>
<section anchor="sec-send-allocate" title="Sending an Allocate Request">
<t>The client forms an Allocate request as follows.</t>
<t>The client first picks a host transport address. It is RECOMMENDED
that the client pick a currently-unused transport address, typically
by allowing the underlying OS to pick a currently-unused port for a
new socket.</t>
<t>The client then picks a transport protocol to use between the
client and the server. The transport protocol MUST be one of UDP, TCP,
or TLS over TCP. Since this specification only allows UDP between the
server and the peers, it is RECOMMENDED that the client pick UDP
unless it has a reason to use a different transport. One reason to
pick a different transport would be that the client believes, either
through configuration or by experiment, that it is unable to contact
any TURN server using UDP. See <xref target="sec-transports"></xref>
for more discussion.</t>
<t>The client also picks a server transport address, which SHOULD be
done as follows. The client receives (perhaps through configuration) a
domain name for a TURN server. The client then uses the DNS procedures
described in <xref target="RFC5389"></xref>, but using an SRV service
name of "turn" (or "turns" for TURN over TLS) instead of "stun" (or
"stuns"). For example, to find servers in the example.com domain, the
client performs a lookup for '_turn._udp.example.com',
'_turn._tcp.example.com', and '_turns._tcp.example.com' if the client
wants to communicate with the server using UDP, TCP, or TLS over TCP,
respectively.</t>
<t>The client MUST include a REQUESTED-TRANSPORT attribute in the
request. This attribute specifies the transport protocol between the
server and the peers (note that this is NOT the transport protocol
that appears in the 5-tuple). In this specification, the
REQUESTED-TRANSPORT type is always UDP. This attribute is included to
allow future extensions specify other protocols.</t>
<t>If the client wishes the server to initialize the time-to-expiry
field of the allocation to some value other the default lifetime, then
it MAY include a LIFETIME attribute specifying its desired value. This
is just a request, and the server may elect to use a different value.
Note that the server will ignore requests to initialize the field to
less than the default value.</t>
<t>If the client wishes to later use the DONT-FRAGMENT attribute in
one or more Send indications on this allocation, then the client
SHOULD include the DONT-FRAGMENT attribute in the Allocate request.
This allows the client to test whether this attribute is supported by
the server.</t>
<t>If the client requires the port number of the relayed-transport
address be even, the client includes the EVEN-PORT attribute. If this
attribute is not included, then the port can be even or odd. By
setting the R bit in the EVEN-PORT attribute to 1, the client can
request that the server reserve the next highest port number (on the
same IP address) for a subsequent allocation. If the R bit is 0, no
such request is made.</t>
<t>The client MAY also include a RESERVATION-TOKEN attribute in the
request to ask the server to use a previously reserved port for the
allocation. If the RESERVATION-TOKEN attribute is included, then the
client MUST omit the EVEN-PORT attribute.</t>
<t>Once constructed, the client sends the Allocate request on the
5-tuple.</t>
</section>
<section anchor="sec-rcv-allocate" title="Receiving an Allocate Request">
<t>When the server receives an Allocate request, it performs the
following checks:<list style="numbers">
<t>The server SHOULD require that the request be authenticated
using the Long-Term Credential mechanism of <xref
target="RFC5389"></xref>.</t>
<t>The server checks if the 5-tuple is currently in use by an
existing allocation. If yes, the server rejects the request with a
437 (Allocation Mismatch) error.</t>
<t>The server checks if the request contain a REQUESTED-TRANSPORT
attribute. If the REQUESTED-TRANSPORT attribute is not included or
is malformed, the server rejects the request with a 400 (Bad
Request) error. Otherwise, if the attribute is included but
specifies a protocol other that UDP, the server rejects the
request with a 442 (Unsupported Transport Protocol) error.</t>
<t>The request may contain a DONT-FRAGMENT attribute. If it does,
but the server does not support sending UDP datagrams with the DF
bit set to 1 (see <xref target="sec-ip-header-fields"></xref>),
then the server treats the DONT-FRAGMENT attribute in the Allocate
request as an unknown comprehension-required attribute.</t>
<t>The server checks if the request contains a RESERVATION-TOKEN
attribute. If yes, and the request also contains a EVEN-PORT
attribute, then the server rejects the request with a 400 (Bad
Request) error. Otherwise it checks to see if the token is valid
(i.e., the token is in range and has not expired, and the
corresponding relayed transport address is still available). If
the token is not valid for some reason, the server rejects the
request with a 508 (Insufficient Port Capacity) error.</t>
<t>The server checks if the request contains an EVEN-PORT
attribute. If yes, then the server checks that it can satisfy the
request (i.e., can allocate a relayed-transport-address as
described below). If the server cannot satisfy the request, then
the server rejects the request with a 508 (Insufficient Port
Capacity) error.</t>
<t>At any point, the server MAY choose to reject the request with
a 486 (Allocation Quota Reached) error if it feels the client is
trying to exceed some locally-defined allocation quota. The server
is free to define this allocation quota any way it wishes, but
SHOULD define it based on the username used to authenticate the
request, and not on the client's transport address.</t>
<t>Also at any point, the server MAY choose to reject the request
with a 300 (Try Alternate) error if it wishes to redirect the
client to a different server. The use of this error code and
attribute follow the specification in <xref
target="RFC5389"></xref>.</t>
</list></t>
<t>If all the checks pass, the server creates the allocation. The
5-tuple is set to the 5-tuple from the Allocate request, while the
list of permissions and the list of channels are initially empty.</t>
<t>The server chooses a relayed-transport-address for the allocation
as follows:<list style="symbols">
<t>If the request contains a RESERVATION-TOKEN, the server uses
the previously-reserved transport address corresponding to the
included token (if it is still available). Note that the
reservation is a server-wide reservation and is not specific to a
particular allocation, since the Allocate request containing the
RESERVATION-TOKEN uses a different 5-tuple than the Allocate
request that made the reservation. The 5-tuple for the Allocate
request containing the RESERVATION-TOKEN attribute can be any
allowed 5-tuple; it can use a different client IP address and
port, a different transport protocol, and even different server IP
address and port (provided, of course, that the server IP address
and port is one that the server is listening for TURN requests
on).</t>
<t>If the request contains an EVEN-PORT attribute with the R bit
set to 0, then the server allocates a relayed-transport-address
with an even port number.</t>
<t>If the request contains an EVEN-PORT attribute with the R bit
set to 1, then the server looks for a pair of port numbers N and
N+1 on the same IP address, where N is even. Port N is used in the
current allocation, while the relayed transport address with port
N+1 is assigned a token and reserved for a future allocation. The
server MUST hold this reservation for at least 30 seconds, and MAY
choose to hold longer (e.g. until the allocation with port N
expires). The server then includes the token in a
RESERVATION-TOKEN attribute in the success response.</t>
<t>Otherwise, the server allocates any available
relayed-transport-address.</t>
</list></t>
<t>In all cases, the server SHOULD only allocate ports from the range
49152 – 65535 (the Dynamic and/or Private Port range <xref
target="Port-Numbers"></xref>), unless the TURN server application
knows, through some means not specified here, that other applications
running on the same host as the TURN server application will not be
impacted by allocating ports outside this range. This condition can
often be satisfied by running the TURN server application on a
dedicated machine and/or by arranging that any other applications on
the machine allocate ports before the TURN server application starts.
In any case, the TURN server SHOULD NOT allocate ports in the range 0
- 1023 (the Well-Known Port range) to discourage clients from using
TURN to run standard services.</t>
<t><list>
<t>NOTE: The IETF is currently investigating the topic of
randomized port assignments to avoid certain types of attacks (see
<xref target="I-D.ietf-tsvwg-port-randomization"></xref>). It is
strongly recommended that a TURN implementor keep abreast of this
topic and, if appropriate, implement a randomized port assignment
algorithm. This is especially applicable to servers that choose to
pre-allocate a number of ports from the underlying OS and then
later assign them to allocations; for example, a server may choose
this technique to implement the EVEN-PORT attribute.</t>
</list></t>
<t>The server determines the initial value of the time-to-expiry field
as follows. If the request contains a LIFETIME attribute, then the
server computes MIN(client's proposed lifetime, server's maximum
allowed lifetime). If this computed lifetime is greater than the
default lifetime, then the server uses that value. Otherwise, the
server uses the default lifetime. It is RECOMMENDED that the server
use a maximum allowed lifetime value of no more than 3600 seconds (1
hour). Servers that implement allocation quotas or charge users for
allocations in some way may wish to use a smaller maximum allowed
lifetime (perhaps as small as the default lifetime) to more quickly
remove orphaned allocations (that is, allocations where the
corresponding client has crashed or terminated or the client
connection has been lost for some reason). Also note that the
time-to-expiry is recomputed with each successful Refresh request, and
thus the value computed here applies only until the first refresh.</t>
<t>Once the allocation is created, the server replies with a success
response. The success response contains:<list style="symbols">
<t>A XOR-RELAYED-ADDRESS attribute containing the relayed
transport address;</t>
<t>A LIFETIME attribute containing the current value of the
time-to-expiry timer;</t>
<t>A RESERVATION-TOKEN attribute (if a second relayed transport
address was reserved).</t>
<t>An XOR-MAPPED-ADDRESS attribute containing the client's IP
address and port (from the 5-tuple).</t>
</list></t>
<t><list>
<t>NOTE: The XOR-MAPPED-ADDRESS attribute is included in the
response as a convenience to the client. TURN itself does not make
use of this value, but clients running ICE can often need this
value and can thus avoid having to do an extra Binding transaction
with some STUN server to learn it.</t>
</list></t>
<t>The response (either success or error) is sent back to the client
on the 5-tuple.</t>
<t><list>
<t>NOTE: Implementations may implement the idempotency of the
Allocate request over UDP using the so-called "stateless stack
approach" as follows. To detect retransmissions when the original
request was successful in creating an allocation, the server can
store the transaction id that created the request with the
allocation data and compare it with incoming Allocate requests on
the same 5-tuple. Once such a request is detected, the server can
stop parsing the request and immediately generate a success
response. When building this response, the value of the LIFETIME
attribute can be taken from the time-to-expiry field in the
allocate state data, even though this value may differ slightly
from the LIFETIME value originally returned. In addition, the
server may need to store an indication of any reservation token
returned in the original response, so that this may be returned in
any retransmitted responses.</t>
<t>For the case where the original request was unsuccessful in
creating an allocation, the server may choose to do nothing
special. Note, however, that there is a rare case where the server
rejects the original request but accepts the retransmitted request
(because conditions have changed in the brief intervening time
period). If the client receives the first failure response, it
will ignore the second (success) response and believe that an
allocation was not created. An allocation created in this matter
will eventually timeout, since the client will not refresh it.
Furthermore, if the client later retries with the same 5-tuple but
different transaction id, it will receive a 437 (Allocation
Mismatch), which will cause it to retry with a different 5-tuple.
The server may use a smaller maximum lifetime value to minimize
the lifetime of allocations "orphaned" in this manner.</t>
</list></t>
</section>
<section title="Receiving an Allocate Success Response">
<t>If the client receives an Allocate success response, then it MUST
check that the mapped address and the relayed transport address are in
an address family that the client understands and is prepared to deal
with. This specification only covers the case where these two
addresses are IPv4 addresses. If these two addresses are not in an
address family that the client is prepared to deal with, then the
client MUST delete the allocation (<xref
target="sec-refreshing-allocation"></xref>) and MUST NOT attempt to
create another allocation on that server until it believes the
mismatch has been fixed.</t>
<t><list>
<t>The IETF is currently considering mechanisms for transitioning
between IPv4 and IPv6 that could result in a client originating an
Allocate request over IPv6, but the request would arrive at the
server over IPv4, or vica-versa. Hence the importance of this
check.</t>
</list></t>
<t>Otherwise, the client creates its own copy of the allocation data
structure to track what is happening on the server. In particular, the
client needs to remember the actual lifetime received back from the
server, rather than the value sent to the server in the request. The
client must also remember the 5-tuple used for the request and the
username and password it used to authenticate the request to ensure
that it reuses them for subsequent messages. The client also needs to
track the channels and permissions it establishes on the server.</t>
<t>The client will probably wish to send the relayed transport address
to peers (using some method not specified here) so the peers can
communicate with it. The client may also wish to use the
server-reflexive address it receives in the XOR-MAPPED-ADDRESS
attribute in its ICE processing.</t>
</section>
<section anchor="sec-allocate-error-response"
title="Receiving an Allocate Error Response">
<t>If the client receives an Allocate error response, then the
processing depends on the actual error code returned:<list
style="symbols">
<t>(Request timed out): There is either a problem with the server,
or a problem reaching the server with the chosen transport. The
client considers the current transaction as having failed but MAY
choose to retry the Allocate request using a different transport
(e.g., TCP instead of UDP).</t>
<t>300 (Try Alternate): The server would like the client to use
the server specified in the ALTERNATE-SERVER attribute instead.
The client considers the current transaction as having failed, but
SHOULD try the Allocate request with the alternate server before
trying any other servers (e.g., other servers discovered using the
SRV procedures). When trying the Allocate request with the
alternate server, the client follows the ALTERNATE-SERVER
procedures specified in <xref target="RFC5389"></xref>.</t>
<t>400 (Bad Request): The server believes the client's request is
malformed for some reason. The client considers the current
transaction as having failed. The client MAY notify the user or
operator and SHOULD NOT retry the request with this server until
it believes the problem has been fixed.</t>
<t>401 (Unauthorized): If the client has followed the procedures
of the Long-Term Credential mechanism and still gets this error,
then the server is not accepting the client's credentials. In this
case, the client considers the current transaction as having
failed and SHOULD notify the user or operator. The client SHOULD
NOT send any further requests to this server until it believes the
problem has been fixed.</t>
<t>403 (Forbidden): The request is valid, but the server is
refusing to perform it, likely due to administrative restrictions.
The client considers the current transaction as having failed. The
client MAY notify the user or operator and SHOULD NOT retry the
same request with this server until it believes the problem has
been fixed.</t>
<t>420 (Unknown Attribute): If the client included a DONT-FRAGMENT
attribute in the request and the server rejected the request with
a 420 error code and listed the DONT-FRAGMENT attribute in the
UNKNOWN-ATTRIBUTES attribute in the error response, then the
client now knows that the server does not support the
DONT-FRAGMENT attribute. The client considers the current
transaction as having failed but MAY choose to retry the Allocate
request without the DONT-FRAGMENT attribute.</t>
<t>437 (Allocation Mismatch): This indicates that the client has
picked a 5-tuple which the server sees as already in use. One way
this could happen is if an intervening NAT assigned a mapped
transport address that was used by another client which recently
crashed. The client considers the current transaction as having
failed. The client SHOULD pick another client transport address
and retry the Allocate request (using a different transaction id).
The client SHOULD try three different client transport addresses
before giving up on this server. Once the client gives up on the
server, it SHOULD NOT try to create another allocation on the
server for 2 minutes.</t>
<t>438 (Stale Nonce): See the procedures for the Long-Term
Credential mechanism <xref target="RFC5389"></xref>.</t>
<t>441 (Wrong Credentials): The client should not receive this
error in response to a Allocate request. The client MAY notify the
user or operator and SHOULD NOT retry the same request with this
server until it believes the problem has been fixed.</t>
<t>442 (Unsupported Transport Address): The client should not
receive this error in response to a request for a UDP allocation.
The client MAY notify the user or operator and SHOULD NOT
reattempt the request with this server until it believes the
problem has been fixed.</t>
<t>486 (Allocation Quota Reached): The server is currently unable
to create any more allocations with this username. The client
considers the current transaction as having failed. The client
SHOULD wait at least 1 minute before trying to create any more
allocations on the server.</t>
<t>508 (Insufficient Port Capacity): The server has no more
relayed transport addresses available, or has none with the
requested properties, or the one that was reserved is no longer
available. The client considers the current operation as having
failed. If the client is using either the EVEN-PORT or the
RESERVATION-TOKEN attribute, then the client MAY choose to remove
or modify this attribute and try again immediately. Otherwise, the
client SHOULD wait at least 1 minute before trying to create any
more allocations on this server.</t>
</list>An unknown error response MUST be handled as described in
<xref target="RFC5389"></xref>.</t>
</section>
</section>
<section anchor="sec-refreshing-allocation"
title="Refreshing an Allocation">
<t>A Refresh transaction can be used to either (a) refresh an existing
allocation and update its time-to-expiry, or (b) delete an existing
allocation.</t>
<t>If a client wishes to continue using an allocation, then the client
MUST refresh it before it expires. It is suggested that the client
refresh the allocation roughly 1 minute before it expires. If a client
no longer wishes to use an allocation, then it SHOULD explicitly delete
the allocation. A client MAY also refresh an allocation at any time for
other reasons.</t>
<section title="Sending a Refresh Request">
<t>If the client wishes to immediately delete an existing allocation,
it includes a LIFETIME attribute with a value of 0. All other forms of
the request refresh the allocation.</t>
<t>The Refresh transaction updates the time-to-expiry timer of an
allocation. If the client wishes the server to set the time-to-expiry
timer to something other than the default lifetime, it includes a
LIFETIME attribute with the requested value. The server then computes
a new time-to-expiry value in the same way as it does for an Allocate
transaction, with the exception that a requested lifetime of 0 causes
the server to immediately delete the allocation.</t>
</section>
<section anchor="sec-rcv-refresh" title="Receiving a Refresh Request">
<t>When the server receives a Refresh request, it processes as per
<xref target="sec-general-behavior"></xref> plus the specific rules
mentioned here.</t>
<t>The server computes a value called the "desired lifetime" as
follows: If the request contains a LIFETIME attribute and the
attribute value is 0, then the "desired lifetime" is 0. Otherwise, if
the request contains a LIFETIME attribute, then the server computes
MIN(client's requested lifetime, server's maximum allowed lifetime).
If this computed value is greater than the default lifetime, then the
"desired lifetime" is the computed value. Otherwise the "desired
lifetime" is the default lifetime.</t>
<t>Subsequent processing depends on the desired lifetime value:<list
style="symbols">
<t>If desired lifetime is 0, then the request succeeds and the
allocation is deleted.</t>
<t>If the desired lifetime is non-zero, then the request succeeds
and the allocation's time-to-expiry is set to the desired
lifetime</t>
</list>If the request succeeds, then server sends a success response
containing:<list style="symbols">
<t>A LIFETIME attribute containing the current value of the
time-to-expiry timer.</t>
</list></t>
<t></t>
<t><list>
<t>NOTE: A server need not do anything special to implement
idempotency of Refresh requests over UDP using the "stateless
stack approach". Retransmitted Refresh requests with a non-zero
desired lifetime will simply refresh the allocation. A
retransmitted Refresh request with a zero desired lifetime will
cause a 437 (Allocation Mismatch) response if the allocation has
already been deleted, but the client will treat this as equivalent
to a success response (see below).</t>
</list></t>
</section>
<section title="Receiving a Refresh Response">
<t>If the client receives a success response to its Refresh request
with a non-zero lifetime, it updates its copy of the allocation data
structure with the time-to-expiry value contained in the response.</t>
<t>If the client receives a 437 (Allocation Mismatch) error response
to a request to delete the allocation, then the allocation no longer
exists and it should consider its request as having effectively
succeeded.</t>
</section>
</section>
<section anchor="sec-permissions" title="Permissions">
<t>For each allocation, the server keeps a list of zero or more
permissions. Each permission consists of an IP address which uniquely
identifies the permission, and an associated time-to-expiry. The IP
address describes a set of peers that are allowed to send data to the
client, and the time-to-expiry is the number of seconds until the
permission expires.</t>
<t>By sending either CreatePermission requests or ChannelBind requests,
the client can cause the server to install or refresh a permission for a
given IP address. This causes one of two things to happen:<list
style="symbols">
<t>If no permission for that IP address exists, then a permission is
created with the given IP address and a time-to-expiry equal to
Permission Lifetime.</t>
<t>If a permission for that IP address already exists, then the
time-to-expiry for that permission is reset to Permission
Lifetime.</t>
</list>The Permission Lifetime MUST be 300 seconds (= 5 minutes).</t>
<t>Each permission’s time-to-expiry decreases down once per second
until it reaches 0, at which point the permission expires and is
deleted.</t>
<t>CreatePermission and ChannelBind requests may be freely intermixed on
a permission. A given permission may be installed or refreshed at one
point in time with a CreatePermission request, and then refreshed with a
ChannelBind request at a different point in time, or vice-versa.</t>
<t>When a UDP datagram arrives at the relayed transport address for the
allocation, the server extracts the source IP address from the IP
header. The server then compares this address with the IP address
associated with each permission in the list of permissions for the
allocation. If no match is found, relaying is not permitted, and the
server silently discards the UDP datagram. If an exact match is found,
then the permission check is considered to have succeeded and the server
continues to process the UDP datagram as specified elsewhere (<xref
target="sec-sending-data-indication"></xref>). Note that only addresses
are compared and port numbers are not considered.</t>
<t>The permissions for one allocation are totally unrelated to the
permissions for a different allocation. If an allocation expires, all
its permissions expire with it.</t>
<t><list>
<t>NOTE: Though TURN permissions expire after 5 minutes, many NATs
deployed at the time of publication expire their UDP bindings
considerably faster. Thus an application using TURN will probably
wish to send some sort of keep-alive traffic at a much faster rate.
Applications using ICE should follow the keep-alive guidelines of
ICE <xref target="I-D.ietf-mmusic-ice"></xref>, and applications not
using ICE are advised to do something similar.</t>
</list></t>
</section>
<section title="CreatePermission">
<t>TURN supports two ways for the client to install or refresh
permissions on the server. This section describes one way: the
CreatePermission request.</t>
<t>A CreatePermission request may be used in conjunction with either the
Send mechanism in <xref target="sec-sendanddata"></xref> or the Channel
mechanism in <xref target="sec-channels"></xref>.</t>
<section title="Forming a CreatePermission request">
<t>The client who wishes to install or refresh one or more permissions
can send a CreatePermission request to the server.</t>
<t>When forming a CreatePermission request, the client MUST include at
least one XOR-PEER-ADDRESS attribute, and MAY include more than one
such attribute. The IP address portion of each XOR-PEER-ADDRESS
attribute contains the IP address for which a permission should be
installed or refreshed. The port portion of each XOR-PEER-ADDRESS
attribute will be ignored and can be any arbitrary value. The various
XOR-PEER-ADDRESS attributes can appear in any order.</t>
</section>
<section title="Receiving a CreatePermission request">
<t>When the server receives the CreatePermission request, it processes
as per <xref target="sec-general-behavior"></xref> plus the specific
rules mentioned here.</t>
<t>The message is checked for validity. The CreatePermission request
MUST contain at least XOR-PEER-ADDRESS attribute and MAY contain
multiple such attributes. If no such attribute exists, or if any of
these attributes are invalid, then a 400 (Bad Request) error is
returned. If the request is valid, but the server is unable to satisfy
the request due to some capacity limit or similar, then a 508
(Insufficient Capacity) error is returned.</t>
<t>The server MAY impose restrictions on the IP address and port
values allowed in the XOR-PEER-ADDRESS attribute -- if a value is not
allowed, the server rejects the request with a 403 (Forbidden)
error.</t>
<t>If the message is valid and the server is capable of carrying out
the request, then the server installs or refreshes a permission for
the IP address contained in each XOR-PEER-ADDRESS attribute as
described in <xref target="sec-permissions"></xref>. The port portion
of each attribute is ignored and may be any arbitrary value.</t>
<t>The server then responds with a CreatePermission success response.
There are no mandatory attributes in the success response.</t>
<t><list>
<t>NOTE: A server need not do anything special to implement
idempotency of CreatePermission requests over UDP using the
"stateless stack approach". Retransmitted CreatePermission
requests will simply refresh the permissions.</t>
</list></t>
</section>
<section title="Receiving a CreatePermission response">
<t>If the client receives a valid CreatePermission success response,
then the client updates its data structures to indicate that the
permissions have been installed or refreshed.</t>
</section>
</section>
<section anchor="sec-sendanddata" title="Send and Data Methods">
<t>TURN supports two mechanisms for sending and receiving data from
peers. This section describes the use of the Send and Data mechanism,
while <xref target="sec-channels"></xref> describes the use of the
Channel mechanism.</t>
<section anchor="sec-forming-indication"
title="Forming a Send Indication">
<t>The client can use a Send indication to pass data to the server for
relaying to a peer. A client may use a Send indication even if a
channel is bound to that peer. However the client MUST ensure that
there is a permission installed for the IP address of the peer to
which the Send indication is being sent; this prevents a third party
from using a TURN server to send data to arbitrary destinations.</t>
<t>When forming a Send indication, the client MUST include a
XOR-PEER-ADDRESS attribute and a DATA attribute. The XOR-PEER-ADDRESS
attribute contains the transport address of the peer to which the data
is to be sent, and the DATA attribute contains the actual application
data to be sent to the peer.</t>
<t>The client MAY include a DONT-FRAGMENT attribute in the Send
indication if it wishes the server to set the DF bit on the UDP
datagram sent to the peer.</t>
</section>
<section title="Receiving a Send Indication">
<t>When the server receives a Send indication, it processes as per
<xref target="sec-general-behavior"></xref> plus the specific rules
mentioned here.</t>
<t>The message is first checked for validity. The Send indication MUST
contain both a XOR-PEER-ADDRESS attribute and a DATA attribute. If one
of these attributes is missing or invalid, then the message is
discarded. Note that the DATA attribute is allowed to contain zero
bytes of data.</t>
<t>The Send indication may also contain the DONT-FRAGMENT attribute.
If the server is unable to set the DF bit on outgoing UDP datagrams
when this attribute is present, then the server acts as if the
DONT-FRAGMENT attribute is an unknown comprehension-required attribute
(and thus the Send indication is discarded).</t>
<t>The server also checks that there is a permission installed for the
IP address contained in the XOR-PEER-ADDRESS attribute. If no such
permission exists, the message is discarded. Note that a Send
indication never causes the server to refresh the permission.</t>
<t>The server MAY impose restrictions on the IP address and port
values allowed in the XOR-PEER-ADDRESS attribute -- if a value is not
allowed, the server silently discards the Send indication.</t>
<t>If everything is OK, then the server forms a UDP datagram as
follows:<list style="symbols">
<t>the source transport address is the relayed transport address
of the allocation, where the allocation is determined by the
5-tuple on which the Send indication arrived;</t>
<t>the destination transport address is taken from the
XOR-PEER-ADDRESS attribute;</t>
<t>the data following the UDP header is the contents of the value
field of the DATA attribute.</t>
</list></t>
<t>The handling of the DONT-FRAGMENT attribute (if present), is
described in <xref target="sec-ip-header-fields"></xref>.</t>
<t>The resulting UDP datagram is then sent to the peer.</t>
</section>
<section anchor="sec-sending-data-indication"
title="Receiving a UDP Datagram">
<t>When the server receives a UDP datagram at a currently allocated
relayed transport address, the server looks up the allocation
associated with the relayed transport address. The server then checks
to see whether the set of permissions for the allocation allow the
relaying of the UDP datagram as described in <xref
target="sec-permissions"></xref>.</t>
<t>If relaying is permitted, then the server checks if there is a
channel bound to the peer that sent the UDP datagram (see <xref
target="sec-channels"></xref>). If a channel is bound, then processing
proceeds as described in <xref
target="sec-channel-relaying"></xref>.</t>
<t>If relaying is permitted but no channel is bound to the peer, then
the server forms and sends a Data indication. The Data indication MUST
contain both a XOR-PEER-ADDRESS and a DATA attribute. The DATA
attribute is set to the value of the ‘data octets’ field
from the datagram, and the XOR-PEER-ADDRESS attribute is set to the
source transport address of the received UDP datagram. The Data
indication is then sent on the 5-tuple associated with the
allocation.</t>
</section>
<section title="Receiving a Data Indication">
<t>When the client receives a Data indication, it checks that the Data
indication contains both a XOR-PEER-ADDRESS and a DATA attribute, and
discards the indication if it does not. The client SHOULD also check
that the XOR-PEER-ADDRESS attribute value contains an IP address with
which the client believes there is an active permission, and discard
the Data indication otherwise. Note that the DATA attribute is allowed
to contain zero bytes of data.</t>
<t><list>
<t>NOTE: The latter check protects the client against an attacker
who somehow manages to trick the server into installing
permissions not desired by the client.</t>
</list></t>
<t>If the Data indication passes the above checks, the client delivers
the data octets inside the DATA attribute to the application, along
with an indication that they were received from the peer whose
transport address is given by the XOR-PEER-ADDRESS attribute.</t>
</section>
</section>
<!-- Sending and Receiving Data -->
<section anchor="sec-channels" title="Channels">
<t>Channels provide a way for the client and server to send application
data using ChannelData messages, which have less overhead than Send and
Data indications.</t>
<t>The ChannelData message (see <xref
target="sec-channeldata-msg"></xref>) starts with a two-byte field that
carries the channel number. The values of this field are allocated as
follows:<list style="empty">
<t>0x0000 through 0x3FFF: These values can never be used for channel
numbers.</t>
<t>0x4000 through 0x7FFF: These values are the allowed channel
numbers (16,383 possible values)</t>
<t>0x8000 through 0xFFFF: These values are reserved for future
use.</t>
</list>Because of this division, ChannelData messages can be
distinguished from STUN-formatted messages (e.g., Allocate request, Send
indication, etc) by examining the first two bits of the message:<list>
<t>0b00: STUN-formatted message (since the first two bits of a
STUN-formatted message are always zero)</t>
<t>0b01: ChannelData message (since the channel number is the first
field in the ChannelData message and channel numbers fall in the
range 0x4000 - 0x7FFF)</t>
<t>0b10: Reserved</t>
<t>0b11: Reserved</t>
</list>The reserved values may be used in the future to extend the
range of channel numbers. Thus an implementation MUST NOT assume that a
TURN message always starts with a 0 bit.</t>
<t>Channel bindings are always initiated by the client. The client can
bind a channel to a peer at any time during the lifetime of the
allocation. The client may bind a channel to a peer before exchanging
data with it, or after exchanging data with it (using Send and Data
indications) for some time, or may choose never to bind a channel to it.
The client can also bind channels to some peers while not binding
channels to other peers.</t>
<t>Channel bindings are specific to an allocation, so that the use of a
channel number or peer transport address in a channel binding in one
allocation has no impact on their use in a different allocation. If an
allocation expires, all its channel bindings expire with it.</t>
<t>A channel binding consists of:<list style="symbols">
<t>A channel number;</t>
<t>A transport address (of the peer);</t>
<t>A time-to-expiry timer.</t>
</list>Within the context of an allocation, a channel binding is
uniquely identified either by the channel number or by the peer's
transport address. Thus the same channel cannot be bound to two
different transport addresses, nor can the same transport address be
bound to two different channels.</t>
<t>A channel binding lasts for 10 minutes unless refreshed. Refreshing
the binding (by the server receiving a ChannelBind request rebinding the
channel to the same peer) resets the time-to-expiry timer back to 10
minutes.</t>
<t>When the channel binding expires, the channel becomes unbound. Once
unbound, the channel number can be bound to a different transport
address, and the transport address can be bound to a different channel
number. To prevent race conditions, the client MUST wait 5 minutes after
the channel binding expires before attempting to bind the channel number
to a different transport address or the transport address to a different
channel number.</t>
<t>When binding a channel to a peer, the client SHOULD be prepared to
receive ChannelData messages on the channel from the server as soon as
it has sent the ChannelBind request. Over UDP, it is possible for the
client to receive ChannelData messages from the server before it
receives a ChannelBind success response.</t>
<t>In the other direction, the client MAY elect to send ChannelData
messages before receiving the ChannelBind success response. Doing so,
however, runs the risk of having the ChannelData messages dropped by the
server if the ChannelBind request does not succeed for some reason
(e.g., packet lost if the request is sent over UDP, or the server being
unable to fulfill the request). A client that wishes to be safe should
either queue the data, or use Send indications until the channel binding
is confirmed.</t>
<section title="Sending a ChannelBind Request">
<t>A channel binding is created or refreshed using a ChannelBind
transaction. A ChannelBind transaction also creates or refreshes a
permission towards the peer (see <xref
target="sec-permissions"></xref>).</t>
<t>To initiate the ChannelBind transaction, the client forms a
ChannelBind request. The channel to be bound is specified in a
CHANNEL-NUMBER attribute, and the peer's transport address is
specified in a XOR-PEER-ADDRESS attribute. <xref
target="sec-receiving-ChannelBind"></xref> describes the restrictions
on these attributes.</t>
<t>Rebinding a channel to the same transport address that it is
already bound to provides a way to refresh a channel binding and the
corresponding permission without sending data to the peer. Note
however, that permissions need to be refreshed more frequently than
channels.</t>
</section>
<section anchor="sec-receiving-ChannelBind"
title="Receiving a ChannelBind Request">
<t>When the server receives a ChannelBind request, it processes as per
<xref target="sec-general-behavior"></xref> plus the specific rules
mentioned here.</t>
<t>The server checks the following:<list style="symbols">
<t>The request contains both a CHANNEL-NUMBER and a
XOR-PEER-ADDRESS attribute;</t>
<t>The channel number is in the range 0x4000 through 0x7FFE
(inclusive);</t>
<t>The channel number is not currently bound to a different
transport address (same transport address is OK);</t>
<t>The transport address is not currently bound to a different
channel number.</t>
</list></t>
<t>If any of these tests fail, the server replies with a 400 (Bad
Request) error.</t>
<t>The server MAY impose restrictions on the IP address and port
values allowed in the XOR-PEER-ADDRESS attribute -- if a value is not
allowed, the server rejects the request with a 403 (Forbidden)
error.</t>
<t>If the request is valid, but the server is unable to fulfill the
request due to some capacity limit or similar, the server replies with
a 508 (Insufficient Capacity) error.</t>
<t>Otherwise, the server replies with a ChannelBind success response.
There are no required attributes in a successful ChannelBind
response.</t>
<t>If the server can satisfy the request, then the server creates or
refreshes the channel binding using the channel number in the
CHANNEL-NUMBER attribute and the transport address in the
XOR-PEER-ADDRESS attribute. The server also installs or refreshes a
permission for the IP address in the XOR-PEER-ADDRESS attribute as
described in <xref target="sec-permissions"></xref>.</t>
<t><list>
<t>NOTE: A server need not do anything special to implement
idempotency of ChannelBind requests over UDP using the "stateless
stack approach". Retransmitted ChannelBind requests will simply
refresh the channel binding and the corresponding permission.
Furthermore, the client must wait 5 minutes before binding a
previously bound channel number or peer address to a different
channel, eliminating the possibility that the transaction would
initially fail but succeed on a retransmission.</t>
</list></t>
</section>
<section title="Receiving a ChannelBind Response">
<t>When the client receives a ChannelBind success response, it updates
its data structures to record that the channel binding is now active.
It also updates its data structures to record that the corresponding
permission has been installed or refreshed.</t>
<t>If the client receives a ChannelBind failure response that
indicates that the channel information is out-of-sync between the
client and the server (e.g., an unexpected 400 "Bad Request"
response), then it is RECOMMENDED that the client immediately delete
the allocation and start afresh with a new allocation.</t>
</section>
<section anchor="sec-channeldata-msg" title="The ChannelData Message">
<t>The ChannelData message is used to carry application data between
the client and the server. It has the following format:</t>
<figure>
<artwork><![CDATA[
0 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
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Channel Number | Length |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| |
/ Application Data /
/ /
| |
| +-------------------------------+
| |
+-------------------------------+]]></artwork>
</figure>
<t>The Channel Number field specifies the number of the channel on
which the data is traveling, and thus the address of the peer that is
sending or is to receive the data.</t>
<t>The Length field specifies the length in bytes of the application
data field (i.e., it does not include the size of the ChannelData
header). Note that 0 is a valid length.</t>
<t>The Application Data field carries the data the client is trying to
send to the peer, or that the peer is sending to the client.</t>
</section>
<section anchor="sec-sending-channeldata-msg"
title="Sending a ChannelData Message">
<t>Once a client has bound a channel to a peer, then when the client
has data to send to that peer it may use either a ChannelData message
or a Send indication; that is, the client is not obligated to use the
channel when it exists and may freely intermix the two message types
when sending data to the peer. The server, on the other hand, MUST use
the ChannelData message if a channel has been bound to the peer.</t>
<t>The fields of the ChannelData message are filled in as described in
<xref target="sec-channeldata-msg"></xref>.</t>
<t>Over stream transports, the ChannelData message MUST be padded to a
multiple of four bytes in order to ensure the alignment of subsequent
messages. The padding is not reflected in the length field of the
ChannelData message, so the actual size of a ChannelData message
(including padding) is (4 + Length) rounded up to the nearest multiple
of 4. Over UDP, the padding is not required but MAY be included.</t>
<t>The ChannelData message is then sent on the 5-tuple associated with
the allocation.</t>
</section>
<section title="Receiving a ChannelData Message">
<t>The receiver of the ChannelData message uses the first two bits to
distinguish it from STUN-formatted messages, as described above. If
the message uses a value in the reserved range (0x8000 through
0xFFFF), then the message is silently discarded.</t>
<t>If the ChannelData message is received in a UDP datagram, and if
the UDP datagram is too short to contain the claimed length of the
ChannelData message (i.e., the UDP header length field value is less
than the ChannelData header length field value + 4 + 8), then the
message is silently discarded.</t>
<t>If the ChannelData message is received over TCP or over TLS over
TCP, then the actual length of the ChannelData message is as described
in <xref target="sec-sending-channeldata-msg"></xref>.</t>
<t>If the ChannelData message is received on a channel which is not
bound to any peer, then the message is silently discarded.</t>
<t>On the client, it is RECOMMENDED that the client discard the
ChannelData message if the client believes there is no active
permission towards the peer. On the server, the receipt of a
ChannelData message MUST NOT refresh either the channel binding or the
permission towards the peer.</t>
<t>On the server, if no errors are detected, the server relays the
application data to the peer by forming a UDP datagram as
follows:<list style="symbols">
<t>the source transport address is the relayed transport address
of the allocation, where the allocation is determined by the
5-tuple on which the ChannelData message arrived;</t>
<t>the destination transport address is the transport address to
which the channel is bound;</t>
<t>the data following the UDP header is the contents of the data
field of the ChannelData message.</t>
</list>The resulting UDP datagram is then sent to the peer. Note
that if the Length field in the ChannelData message is 0, then there
will be no data in the UDP datagram, but the UDP datagram is still
formed and sent.</t>
</section>
<section anchor="sec-channel-relaying"
title="Relaying Data from the Peer">
<t>When the server receives a UDP datagram on the relayed transport
address associated with an allocation, the server processes it as
described in <xref target="sec-sending-data-indication"></xref>. If
that section indicates that a ChannelData message should be sent
(because there is a channel bound to the peer that sent to UDP
datagram), then the server forms and sends a ChannelData message as
described in <xref target="sec-sending-channeldata-msg"></xref>.</t>
</section>
</section>
<section anchor="sec-ip-header-fields" title="IP Header Fields">
<t>This section describes how the server sets various fields in the IP
header when relaying between the client and the peer or vica-versa. The
descriptions in this section apply: (a) when the server sends a UDP
datagram to the peer, or (b) when the server sends a Data indication or
ChannelData message to the client over UDP transport. The descriptions
in this section do not apply to TURN messages sent over TCP or TLS
transport from the server to the client.</t>
<t>The descriptions below have two parts: a preferred behavior and an
alternate behavior. The server SHOULD implement the preferred behavior,
but if that is not possible for a particular field, then it SHOULD
implement the alternative behavior.</t>
<t>Time to Live (TTL) field<list style="empty">
<t>Preferred Behavior: If the incoming value is 0, then the drop the
incoming packet. Otherwise set the outgoing Time to Live/Hop Count
to one less than the incoming value.</t>
<t>Alternate Behavior: Set the outgoing value to the default for
outgoing packets.</t>
</list><vspace blankLines="1" /></t>
<t>Diff-Serv Code Point (DSCP) field <xref target="RFC2474"></xref><list
style="empty">
<t>Preferred Behavior: Set the outgoing value to the incoming value,
unless the server includes a differentiated services classifier and
marker <xref target="RFC2474"></xref>.</t>
<t>Alternate Behavior: Set the outgoing value to a fixed value,
which by default is Best Effort unless configured otherwise.</t>
<t>In both cases, if the server is immediately adjacent to a
differentiated services classifier and marker, then DSCP MAY be set
to any arbitrary value in the direction towards the classifier.</t>
</list></t>
<t><vspace blankLines="1" /></t>
<t>Explicit Congestion Notification (ECN) field <xref
target="RFC3168"></xref><list style="empty">
<t>Preferred Behavior: Set the outgoing value to the incoming value,
UNLESS the server is doing Active Queue Management, the incoming ECN
field is ECT(1) (=0b01) or ECT(0) (=0b10), and the server wishes to
indicate that congestion has been experienced, in which case set the
outgoing value to CE (=0b11).</t>
<t>Alternate Behavior: Set the outgoing value to Not-ECT
(=0b00).</t>
</list></t>
<t><vspace blankLines="1" /></t>
<t>IPv4 Fragmentation fields<list>
<t>Preferred Behavior: <list>
<t>When the server sends a packet to a peer in response to a
Send indication containing the DONT-FRAGMENT attribute, then set
the DF bit in the outgoing IP header to 1. In all other cases
when sending an outgoing packet containing application data
(e.g., Data indication, ChannelData message, or DONT-FRAGMENT
attribute not included in the Send indication), copy the DF bit
from the DF bit of the incoming packet that contained the
application data.</t>
<t>Set the other fragmentation fields (Identification, MF,
Fragment Offset) as appropriate for a packet originating from
the server.</t>
</list></t>
<t>Alternate Behavior: As described in the Preferred Behavior,
except always assume the incoming DF bit is 0.</t>
<t>In both the Preferred and Alternate Behaviors, the resulting
packet may be too large for the outgoing link. If this is the case,
then the normal fragmentation rules apply <xref
target="RFC1122"></xref>.</t>
</list></t>
<t><vspace blankLines="1" /></t>
<t>IPv4 Options<list>
<t>Preferred Behavior: The outgoing packet is sent without any IPv4
options.</t>
<t>Alternate Behavior: Same as preferred.</t>
</list></t>
</section>
<section anchor="sec-stun-methods" title="New STUN Methods">
<t>This section lists the codepoints for the new STUN methods defined in
this specification. See elsewhere in this document for the semantics of
these new methods.</t>
<figure>
<preamble></preamble>
<artwork><![CDATA[ 0x003 : Allocate (only request/response semantics defined)
0x004 : Refresh (only request/response semantics defined)
0x006 : Send (only indication semantics defined)
0x007 : Data (only indication semantics defined)
0x008 : CreatePermission (only request/response semantics defined
0x009 : ChannelBind (only request/response semantics defined)
]]></artwork>
</figure>
</section>
<section anchor="sec-stun-attributes" title="New STUN Attributes">
<figure>
<preamble>This STUN extension defines the following new
attributes:</preamble>
<artwork><![CDATA[
0x000C: CHANNEL-NUMBER
0x000D: LIFETIME
0x0010: Reserved (was BANDWIDTH)
0x0012: XOR-PEER-ADDRESS
0x0013: DATA
0x0016: XOR-RELAYED-ADDRESS
0x0018: EVEN-PORT
0x0019: REQUESTED-TRANSPORT
0x001A: DONT-FRAGMENT
0x0021: Reserved (was TIMER-VAL)
0x0022: RESERVATION-TOKEN
]]></artwork>
<postamble></postamble>
</figure>
<t>Some of these attributes have lengths that are not multiples of 4. By
the rules of STUN, any attribute whose length is not a multiple of 4
bytes MUST be immediately followed by 1 to 3 padding bytes to ensure the
next attribute (if any) would start on a 4-byte boundary (see <xref
target="RFC5389"></xref>).</t>
<section anchor="channelnums" title="CHANNEL-NUMBER">
<t>The CHANNEL-NUMBER attribute contains the number of the channel.
The value portion of this attribute is 4 bytes long and consists of a
16-bit unsigned integer, followed by a two-octet RFFU (Reserved For
Future Use) field which MUST be set to 0 on transmission and MUST be
ignored on reception.</t>
<figure>
<artwork><![CDATA[
0 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
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Channel Number | RFFU = 0 |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
]]></artwork>
</figure>
</section>
<section title="LIFETIME">
<t>The LIFETIME attribute represents the duration for which the server
will maintain an allocation in the absence of a refresh. The value
portion of this attribute is 4-bytes long and consists of a 32-bit
unsigned integral value representing the number of seconds remaining
until expiration.</t>
</section>
<section title="XOR-PEER-ADDRESS">
<t>The XOR-PEER-ADDRESS specifies the address and port of the peer as
seen from the TURN server. (In other words, the peer's
server-reflexive transport address if the peer is behind a NAT). It is
encoded in the same way as XOR-MAPPED-ADDRESS <xref
target="RFC5389"></xref>.</t>
</section>
<section title="DATA">
<t>The DATA attribute is present in all Send and Data indications. The
value portion of this attribute is variable-length and consists of the
application data (that is, the data that would immediately follow the
UDP header if the data was been sent directly between the client and
the peer). If the length of this attribute is not a multiple of 4,
then padding must be added after this attribute.</t>
</section>
<section title="XOR-RELAYED-ADDRESS">
<t>The XOR-RELAYED-ADDRESS is present in Allocate responses. It
specifies the address and port that the server allocated to the
client. It is encoded in the same way as XOR-MAPPED-ADDRESS <xref
target="RFC5389"></xref>.</t>
</section>
<section title="EVEN-PORT">
<t>This attribute allows the client to request that the port in the
relayed-transport-address be even, and (optionally) that the server
reserve the next-higher port number. The value portion of this
attribute is 1 byte long. Its format is:</t>
<figure>
<preamble></preamble>
<artwork><![CDATA[ 0
0 1 2 3 4 5 6 7
+-+-+-+-+-+-+-+-+
|R| RFFU |
+-+-+-+-+-+-+-+-+]]></artwork>
</figure>
<t></t>
<t>The value contains a single 1-bit flag:<list style="hanging">
<t hangText="R:">If 1, the server is requested to reserve the next
higher port number (on the same IP address) for a subsequent
allocation. If 0, no such reservation is requested.</t>
</list>The other 7 bits of the attribute's value must be set to zero
on transmission and ignored on reception.</t>
<t>Since the length of this attribute is not a multiple of 4, padding
must immediately follow this attribute.</t>
</section>
<section anchor="sec-requested-transport" title="REQUESTED-TRANSPORT">
<t>This attribute is used by the client to request a specific
transport protocol for the allocated transport address. The value of
this attribute is 4 bytes with the following format:</t>
<figure>
<artwork><![CDATA[ 0 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
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Protocol | RFFU |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
]]></artwork>
</figure>
<t>The Protocol field specifies the desired protocol. The codepoints
used in this field are taken from those allowed in the Protocol field
in the IPv4 header and the NextHeader field in the IPv6 header <xref
target="Protocol-Numbers"></xref>. This specification only allows the
use of codepoint 17 (User Datagram Protocol).</t>
<t>The RFFU field MUST be set to zero on transmission and MUST be
ignored on reception. It is reserved for future uses.</t>
</section>
<section title="DONT-FRAGMENT">
<t>This attribute is used by the client to request that the server set
the DF (Don't Fragment) bit in the IP header when relaying the
application data onward to the peer. This attribute has no value part
and thus the attribute length field is 0.</t>
</section>
<section title="RESERVATION-TOKEN">
<t>The RESERVATION-TOKEN attribute contains a token that uniquely
identifies a relayed transport address being held in reserve by the
server. The server includes this attribute in a success response to
tell the client about the token, and the client includes this
attribute in a subsequent Allocate request to request the server use
that relayed transport address for the allocation.</t>
<t>The attribute value is 8 bytes and contains the token value.</t>
</section>
</section>
<section anchor="sec-stun-errors" title="New STUN Error Response Codes">
<t>This document defines the following new error response codes:<list
style="hanging">
<t hangText="403">(Forbidden): The request was valid, but cannot be
performed due to administrative or similar restrictions.</t>
<t hangText="437">(Allocation Mismatch): A request was received by
the server that requires an allocation to be in place, but there is
none, or a request was received which requires no allocation, but
there is one.</t>
<t hangText="441">(Wrong Credentials): The credentials in the
(non-Allocate) request, though otherwise acceptable to the server,
do not match those used to create the allocation.</t>
<t hangText="442">(Unsupported Transport Protocol): The Allocate
request asked the server to use a transport protocol between the
server and the peer that the server does not support. NOTE: This
does NOT refer to the transport protocol used in the 5-tuple.</t>
<t hangText="486">(Allocation Quota Reached): No more allocations
using this username can be created at the present time.</t>
<t hangText="508">(Insufficient Capacity): The server is unable to
carry out the request due to some capacity limit being reached. In
an Allocate response, this could be due to the server having no more
relayed transport addresses available right now, or having none with
the requested properties, or the one that corresponds to the
specified reservation token is not available.</t>
</list></t>
</section>
<section title="Detailed Example">
<t>This section gives a example of the use of TURN, showing in detail
the contents of the messages exchanged. The example uses the network
diagram shown in the Overview (<xref
target="fig-turn-model"></xref>).</t>
<t>For each message, the attributes included in the message and their
values are shown. For convenience, values are shown in a human-readable
format rather than showing the actual octets; for example
"XOR-RELAYED-ADDRESS=192.0.2.15:9000" shows that the XOR-RELAYED-ADDRESS
attribute is included with an address of 192.0.2.15 and a port of 9000,
here the address and port are shown before the xor-ing is done. For
attributes with string-like values (e.g. SOFTWARE="Example client,
version 1.03" and NONCE="adl7W7PeDU4hKE72jdaQvbAMcr6h39sm"), the value
of the attribute is shown in quotes for readability, but these quotes do
not appear in the actual value.</t>
<t></t>
<figure>
<artwork><![CDATA[TURN TURN Peer Peer
client server A B
| | | |
|--- Allocate request -------------->| | |
| Transaction-Id=0xA56250D3F17ABE679422DE85 | |
| SOFTWARE="Example client, version 1.03" | |
| LIFETIME=3600 (1 hour) | | |
| REQUESTED-TRANSPORT=17 (UDP) | | |
| DONT-FRAGMENT | | |
| | | |
|<-- Allocate error response --------| | |
| Transaction-Id=0xA56250D3F17ABE679422DE85 | |
| SOFTWARE="Example server, version 1.17" | |
| ERROR-CODE=401 (Unauthorized) | | |
| REALM="example.com" | | |
| NONCE="adl7W7PeDU4hKE72jdaQvbAMcr6h39sm" | |
| | | |
|--- Allocate request -------------->| | |
| Transaction-Id=0xC271E932AD7446A32C234492 | |
| SOFTWARE="Example client 1.03" | | |
| LIFETIME=3600 (1 hour) | | |
| REQUESTED-TRANSPORT=17 (UDP) | | |
| DONT-FRAGMENT | | |
| USERNAME="George" | | |
| REALM="example.com" | | |
| NONCE="adl7W7PeDU4hKE72jdaQvbAMcr6h39sm" | |
| MESSAGE-INTEGRITY=... | | |
| | | |
|<-- Allocate success response ------| | |
| Transaction-Id=0xC271E932AD7446A32C234492 | |
| SOFTWARE="Example server, version 1.17" | |
| LIFETIME=1200 (20 minutes) | | |
| XOR-RELAYED-ADDRESS=192.0.2.15:50000 | |
| XOR-MAPPED-ADDRESS=192.0.2.1:7000 | |
| MESSAGE-INTEGRITY=... | | |
]]></artwork>
<postamble></postamble>
</figure>
<t>The client begins by selecting a host transport address to use for
the TURN session; in this example the client has selected 10.1.1.2:49721
as shown in <xref target="fig-turn-model"></xref>. The client then sends
an Allocate request to the server at the server transport address. The
client randomly selects a 96-bit transaction id of
0xA56250D3F17ABE679422DE85 for this transaction; this is encoded in the
transaction id field in the fixed header. The client includes a SOFTWARE
attribute that gives information about the client's software; here the
value is "Example client, version 1.03" to indicate that this is version
1.03 of something called the Example client. The client includes the
LIFETIME attribute because it wishes the allocation to have a longer
lifetime than the default of 10 minutes; the value of this attribute is
3600 seconds, which corresponds to 1 hour. The client must always
include a REQUESTED-TRANSPORT attribute in an Allocate request and the
only value allowed by this specification is 17, which indicates UDP
transport between the server and the peers. The client also includes the
DONT-FRAGMENT attribute because it wishes to use the DONT-FRAGMENT
attribute later in Send indications; this attribute consists of only an
attribute header, there is no value part. We assume the client has not
recently interacted with the server, thus the client does not include
USERNAME, REALM, NONCE, or MESSAGE-INTEGRITY attribute. Finally, note
that the order of attributes in a message is arbitrary (except for the
MESSAGE-INTEGRITY and FINGERPRINT attributes) and the client could have
used a different order.</t>
<t>The server follows the recommended practice in this specification of
requiring all requests to be authenticated. Thus when the server
receives the initial Allocate request, it rejects the request because
the request does not contain the authentication attributes. Following
the procedures of the Long-Term Credential Mechanism of STUN <xref
target="RFC5389"></xref>, the server includes an ERROR-CODE attribute
with a value of 401 (Unauthorized), a REALM attribute that specifies the
authentication realm used by the server (in this case, the server's
domain "example.com"), and a nonce value in a NONCE attribute. The
server also includes a SOFTWARE attribute that gives information about
the server's software.</t>
<t>The client, upon receipt of the 401 error, re-attempts the Allocate
request, this time including the authentication attributes. The client
selects a new transaction id, and then populates the new Allocate
request with the same attributes as before. The client includes a
USERNAME attribute and uses the realm value received from the server to
help it determine which value to use; here the client is configured to
use the username "George" for the realm "example.com". The client also
includes the REALM and NONCE attributes, which are just copied from the
401 error response. Finally, the client includes a MESSAGE-INTEGRITY
attribute as the last attribute in the message, whose value is an
HMAC-SHA1 hash over the contents of the message (shown as just "..."
above); this HMAC-SHA1 computation includes a password value, thus an
attacker cannot compute the message integrity value without somehow
knowing the secret password.</t>
<t>The server, upon receipt of the authenticated Allocate request,
checks that everything is OK, then creates an allocation. The server
replies with an Allocate success response. The server includes a
LIFETIME attribute giving the lifetime of the allocation; here, the
server has reduced the client's requested 1 hour lifetime to just 20
minutes, because this particular server doesn't allow lifetimes longer
than 20 minutes. The server includes an XOR-RELAYED-ADDRESS attribute
whose value is the relayed transport address of the allocation. The
server includes an XOR-MAPPED-ADDRESS attribute whose value is the
server-reflexive address of the client; this value is not used otherwise
in TURN but is returned as a convenience to the client. The server
includes a MESSAGE-INTEGRITY attribute to authenticate the response and
to insure its integrity; note that the response does not contain the
USERNAME, REALM, and NONCE attributes. The server also includes a
SOFTWARE attribute.</t>
<t></t>
<figure>
<artwork><![CDATA[TURN TURN Peer Peer
client server A B
|--- CreatePermission request ------>| | |
| Transaction-Id=0xE5913A8F460956CA277D3319 | |
| XOR-PEER-ADDRESS=192.0.2.150:0 | | |
| USERNAME="George" | | |
| REALM="example.com" | | |
| NONCE="adl7W7PeDU4hKE72jdaQvbAMcr6h39sm" | |
| MESSAGE-INTEGRITY=... | | |
| | | |
|<-- CreatePermission success resp.--| | |
| Transaction-Id=0xE5913A8F460956CA277D3319 | |
| MESSAGE-INTEGRITY=... | | |
]]></artwork>
<postamble></postamble>
</figure>
<t>The client then creates a permission towards peer A in preparation
for sending it some application data. This is done through a
CreatePermission request. The XOR-PEER-ADDRESS attribute contains the IP
address for which a permission is established (the IP address of peer
A); note that the port number in the attribute is ignored when used in a
CreatePermission request, and here it has been set to 0; also note how
the client uses Peer A's server-reflexive IP address and not its
(private) host address. The client uses the same username, realm, and
nonce values as in the previous request on the allocation. Though it is
allowed to do so, the client has chosen not to include a SOFTWARE
attribute in this request.</t>
<t>The server receives the CreatePermission request, creates the
corresponding permission, and then replies with a CreatePermission
success response. Like the client, the server chooses not to include the
SOFTWARE attribute in its reply. Again, note how success responses
contain a MESSAGE-INTEGRITY attribute (assuming the server uses the
Long-Term Credential Mechanism), but no USERNAME, REALM, and NONCE
attributes.</t>
<t></t>
<figure>
<artwork><![CDATA[TURN TURN Peer Peer
client server A B
|--- Send indication --------------->| | |
| Transaction-Id=0x1278E9ACA2711637EF7D3328 | |
| XOR-PEER-ADDRESS=192.0.2.150:32102 | |
| DONT-FRAGMENT | | |
| DATA=... | | |
| |-- UDP dgm ->| |
| | data=... | |
| | | |
| |<- UDP dgm --| |
| | data=... | |
|<-- Data indication ----------------| | |
| Transaction-Id=0x8231AE8F9242DA9FF287FEFF | |
| XOR-PEER-ADDRSSS=192.0.2.150:32102 | |
| DATA=... | | |
]]></artwork>
<postamble></postamble>
</figure>
<t>The client now sends application data to Peer A using a Send
indication. Peer A's server-reflexive transport address is specified in
the XOR-PEER-ADDRESS attribute, and the application data (shown here as
just "...") is specified in the DATA attribute. The client is doing a
form of path MTU discovery at the application layer and thus specifies
(by including the DONT-FRAGMENT attribute) that the server should set
the DF bit in the UDP datagram send to the peer. Indications cannot be
authenticated using the Long-Term Credential Mechanism of STUN, so no
MESSAGE-INTEGRITY attribute is included in the message. An application
wishing to ensure that its data is not altered or forged must
integrity-protect its data at the application level.</t>
<t>Upon receipt of the Send indication, the server extracts the
application data and sends it in a UDP datagram to Peer A, with the
relayed-transport-address as the source transport address of the
datagram, and with the DF bit set as requested. Note that, had the
client not previously established a permission for Peer A's
server-reflexive IP address, then the server would have silently
discarded the Send indication instead.</t>
<t>Peer A then replies with its own UDP datagram containing application
data. The datagram is sent to the relayed-transport-address on the
server. When this arrives, the server creates a Data indication
containing the source of the UDP datagram in the XOR-PEER-ADDRESS
attribute, and the data from the UDP datagram in the DATA attribute. The
resulting Data indication is then sent to the client.</t>
<t></t>
<figure>
<artwork><![CDATA[TURN TURN Peer Peer
client server A B
|--- ChannelBind request ----------->| | |
| Transaction-Id=0x6490D3BC175AFF3D84513212 | |
| CHANNEL-NUMBER=0x4000 | | |
| XOR-PEER-ADDRESS=192.0.2.210:49191 | |
| USERNAME="George" | | |
| REALM="example.com" | | |
| NONCE="adl7W7PeDU4hKE72jdaQvbAMcr6h39sm" | |
| MESSAGE-INTEGRITY=... | | |
| | | |
|<-- ChannelBind success response ---| | |
| Transaction-Id=0x6490D3BC175AFF3D84513212 | |
| MESSAGE-INTEGRITY=... | | |
]]></artwork>
<postamble></postamble>
</figure>
<t>The client now binds a channel to Peer B, specifying a free channel
number (0x4000) in the CHANNEL-NUMBER attribute, and Peer B's transport
address in the XOR-PEER-ADDRESS attribute. As before, the client re-uses
the username, realm, and nonce from its last request in the message.</t>
<t>Upon receipt of the request, the server binds the channel number to
the peer, installs a permission for Peer B's IP address, and then
replies with ChannelBind success response.</t>
<t></t>
<figure>
<artwork><![CDATA[TURN TURN Peer Peer
client server A B
|--- ChannelData ------------------->| | |
| Channel-number=0x4000 |--- UDP datagram --------->|
| Data=... | Data=... |
| | | |
| |<-- UDP datagram ----------|
| | Data=... | |
|<-- ChannelData --------------------| | |
| Channel-number=0x4000 | | |
| Data=... | | |
]]></artwork>
<postamble></postamble>
</figure>
<t>The client now sends a ChannelData message to the server with data
destined for Peer B. The ChannelData message is not a STUN message, and
thus has no transaction id. Instead, it has only three fields: a channel
number, data, and data length; here the channel number field is 0x4000
(the channel the client just bound to Peer B). When the server receives
the ChannelData message, it checks that the channel is currently bound
(which it is) and then sends the data onward to Peer B in a UDP
datagram, using the relayed-transport-address as the source transport
address and 192.0.2.210:49191 (the value of the XOR-PEER-ADDRESS
attribute in the ChannelBind request) as the destination transport
address.</t>
<t>Later, Peer B sends a UDP datagram back to the
relayed-transport-address. This causes the server to send a ChannelData
message to the client containing the data from the UDP datagram. The
server knows which client to send the ChannelData message to because of
the relayed-transport-address the UDP datagram arrived at, and knows to
use channel 0x4000 because this is the channel bound to
192.0.2.210:49191. Note that if there had not been any channel number
bound to that address, the server would have used a Data indication
instead.</t>
<t></t>
<figure>
<artwork><![CDATA[TURN TURN Peer Peer
client server A B
|--- Refresh request --------------->| | |
| Transaction-Id=0x0864B3C27ADE9354B4312414 | |
| SOFTWARE="Example client 1.03" | | |
| USERNAME="George" | | |
| REALM="example.com" | | |
| NONCE="adl7W7PeDU4hKE72jdaQvbAMcr6h39sm" | |
| MESSAGE-INTEGRITY=... | | |
| | | |
|<-- Refresh error response ---------| | |
| Transaction-Id=0x0864B3C27ADE9354B4312414 | |
| SOFTWARE="Example server, version 1.17" | |
| ERROR-CODE=438 (Stale Nonce) | | |
| REALM="example.com" | | |
| NONCE="npSw1Xw239bBwGYhjNWgz2yH47sxB2j" | |
| | | |
|--- Refresh request --------------->| | |
| Transaction-Id=0x427BD3E625A85FC731DC4191 | |
| SOFTWARE="Example client 1.03" | | |
| USERNAME="George" | | |
| REALM="example.com" | | |
| NONCE="npSw1Xw239bBwGYhjNWgz2yH47sxB2j" | |
| MESSAGE-INTEGRITY=... | | |
| | | |
|<-- Refresh success response -------| | |
| Transaction-Id=0x427BD3E625A85FC731DC4191 | |
| SOFTWARE="Example server, version 1.17" | |
| LIFETIME=600 (10 minutes) | | |
]]></artwork>
<postamble></postamble>
</figure>
<t>Sometime before the 20 minute lifetime is up, the client refreshes
the allocation. This is done using a Refresh request. As before, the
client includes the latest username, realm, and nonce values in the
request. The client also includes the SOFTWARE attribute, following the
recommended practice of always including this attribute in Allocate and
Refresh messages. When the server receives the Refresh request, it
notices that the nonce value has expired, and so replies with 438 (Stale
Nonce) error given a new nonce value. The client then reattempts the
request, this time with the new nonce value. This second attempt is
accepted, and the server replies with a success response. Note that the
client did not include a LIFETIME attribute in the request, so the
server refreshes the allocation for the default lifetime of 10 minutes
(as can be seen by the LIFETIME attribute in the success response).</t>
<t></t>
</section>
<section anchor="sec-security" title="Security Considerations">
<t>This section considers attacks that are possible in a TURN
deployment, and discusses how they are mitigated by mechanisms in the
protocol or recommended practices in the implementation.</t>
<t>Note: Most of the attacks on TURN are mitigated by the server
requiring requests be authenticated using the Long-Term Credential
mechanism of STUN. Thus it is strongly recommended that servers demand
that requests be authenticated. However, in certain deployments, the use
of this mechanism may be unnecessary. An example might be a deployment
where access to the TURN server is available only through a network
where their are fairly tight controls over what devices can connect to
the network (and by whom) and what software these devices can use.
Tightly-run corporate networks can arguably fall into this category.</t>
<section title="Outsider Attacks">
<t>Outsider attacks are ones where the attacker has no credentials in
the system, and is attempting to disrupt the service seen by the
client or the server.</t>
<section title="Obtaining Unauthorized Allocations">
<t>An attacker might wish to obtain allocations on a TURN server for
any number of nefarious purposes. A TURN server provides a mechanism
for sending and receiving packets while cloaking the actual IP
address of the client. This makes TURN servers an attractive target
for attackers who wish to use it to mask their true identity.</t>
<t>An attacker might also wish to simply utilize the services of a
TURN server without paying for them. Since TURN services require
resources from the provider, it is anticipated that their usage will
come with a cost.</t>
<t>These attacks are prevented using the digest authentication
mechanism which allows the TURN server to determine the identity of
the requestor and whether the requestor is allowed to obtain the
allocation.</t>
</section>
<section title="Offline Dictionary Attacks">
<t>The digest authentication mechanism used by TURN is subject to
offline dictionary attacks. An attacker that is capable of
eavesdropping on a message exchange between a client and server can
determine the password by trying a number of candidate passwords and
seeing if one of them is correct. This attack works when the
passwords are low entropy, such as a word from the dictionary. This
attack can be mitigated by using strong passwords with large
entropy. In situations where even stronger mitigation is required,
TLS transport between the client and the server can be used.</t>
</section>
<section title="Faked Refreshes and Permissions">
<t>An attacker might wish to attack an active allocation by sending
it a Refresh request with an immediate expiration, in order to
delete it and disrupt service to the client. This is prevented by
authentication of refreshes. Similarly, an attacker wishing to send
CreatePermission requests to create permissions to undesirable
destinations is prevented from doing so through authentication. The
motivations for such an attack are described in <xref
target="sec-firewall"></xref>.</t>
</section>
<section title="Fake Data">
<t>An attacker might wish to send data to the client or the peer, as
if they came from the peer or client respectively. To do that, the
attacker can send the client a faked Data Indication or ChannelData
message, or send the TURN server a faked Send Indication or
ChannelData message.</t>
<t>Indeed, since indications and ChannelData messages are not
authenticated, this attack is not prevented by TURN. However, this
attack is generally present in IP-based communications and is not
substantially worsened by TURN. Consider an normal, non-TURN IP
session between hosts A and B. An attacker can send packets to B as
if they came from A by sending packets towards A with a spoofed IP
address of B. This attack requires the attacker to know the IP
addresses of A and B. With TURN, an attacker wishing to send packets
towards a client using a Data indication needs to know its IP
address (and port), the IP address and port of the TURN server, and
the IP address and port of the peer (for inclusion in the
XOR-PEER-ADDRESS attribute). To send a fake ChannelData message to a
client, an attacker needs to know the IP address and port of the
client, the IP address and port of the TURN server, and the channel
number. This particular combination is mildly more guessable than in
the non-TURN case.</t>
<t>These attacks are more properly mitigated by application layer
authentication techniques. In the case of real time traffic, usage
of SRTP <xref target="RFC3711"></xref> prevents these attacks.</t>
<t>In some situations, the TURN server may be situated in the
network such that it is able to send to hosts that the client cannot
directly send to. This can happen, for example, if the server is
located behind a firewall that allows packets from outside the
firewall to be delivered to the server, but not to other hosts
behind the firewall. In these situations, an attacker could send the
server a Send indication with an XOR-PEER-ADDRESS attribute
containing the transport address of one of the other hosts behind
the firewall. If the server was to allow relaying of traffic to
arbitrary peers, then this would provide a way for the attacker to
attack arbitrary hosts behind the firewall.</t>
<t>To mitigate this attack, TURN requires that the client establish
a permission to a host before sending it data. Thus an attacker can
only attack hosts that the client is already communicating with,
unless the attacker is able to create authenticated requests.
Furthermore, the server administrator may configure the server to
restrict the range of IP addresses and ports that it will relay data
to. To provide even greater security, the server administrator can
require that the client use TLS for all communication between the
client and the server.</t>
</section>
<section title="Impersonating a Server">
<t>When a client learns a relayed address from a TURN server, it
uses that relayed address in application protocols to receive
traffic. Therefore, an attacker wishing to intercept or redirect
that traffic might try to impersonate a TURN server and provide the
client with a faked relayed address.</t>
<t>This attack is prevented through the digest authentication
mechanism, which provides message integrity for responses in
addition to verifying that they came from the server. Furthermore,
an attacker cannot replay old server responses as the transaction ID
in the STUN header prevents this. Replay attacks are further
thwarted through frequent changes to the nonce value.</t>
</section>
<section title="Eavesdropping Traffic">
<t>TURN concerns itself primarily with authentication and message
integrity. Confidentiality is only a secondary concern, as TURN
control messages do not include information that is particularly
sensitive. The primary protocol content of the messages is the IP
address of the peer. If it is important to prevent an eavesdropper
on a TURN connection from learning this, TURN can be run over
TLS.</t>
<t>Confidentiality for the application data relayed by TURN is best
provided by the application protocol itself, since running TURN over
TLS does not protect application data between the server and the
peer. If confidentiality of application data is important, then the
application should encrypt or otherwise protect its data. For
example, for real time media, confidentiality can be provided by
using SRTP.</t>
</section>
<section title="TURN loop attack">
<t>An attacker might attempt to cause data packets to loop
indefinitely between two TURN servers. The attack goes as follows.
First, the attacker sends an Allocate request to server A, using the
source address of server B. Server A will send its response to
server B, and for the attack to succeed, the attacker must have the
ability to either view or guess the contents of this response, so
that the attacker can learn the allocated relayed-transport-address.
The attacker then sends an Allocate request to server B, using the
source address of server A. Again, the attacker must be able to view
or guess the contents of the response, so it can send learn the
allocated relayed-transport-address. Using the same spoofed source
address technique, the attacker then binds a channel number on
server A to the relayed-transport-address on server B, and similarly
binds the same channel number on server B to the
relayed-transport-address on server A. Finally, the attacker sends a
ChannelData message to server A.</t>
<t>The result is a data packet that loops from the
relayed-transport-address on server A to the
relayed-transport-address on server B, then from server B's
transport address to server A's transport address, and then around
the loop again.</t>
<t>This attack is mitigated as follows. By requiring all requests to
be authenticated and/or by randomizing the port number allocated for
the relayed-transport-address, the server forces the attacker to
either intercept or view responses sent to a third party (in this
case, the other server) so that the attacker can authenticate the
requests and learn the relayed-transport-address. Without one of
these two measures, an attacker can guess the contents of the
responses without needing to see them, which makes the attack much
easier to perform. Furthermore, by requiring authenticated requests,
the server forces the attacker to have credentials acceptable to the
server, which turns this from an outsider attack into an insider
attack and allows the attack to be traced back to the client
initiating it.</t>
<t>The attack can be further mitigated by imposing a per-username
limit on the bandwidth used to relay data by allocations owned by
that username, to limit the impact of this attack on other
allocations. More mitigation can be achieved by decrementing the TTL
when relaying data packets (if the underlying OS allows this).</t>
</section>
</section>
<section anchor="sec-firewall" title="Firewall Considerations">
<t>A key aspect of TURN's security considerations is that it should
not weaken the protections afforded by firewalls deployed between a
client and a TURN server. It is anticipated that TURN servers will
often be present on the public Internet, and clients may often be
inside enterprise networks with corporate firewalls. If TURN servers
provide a 'backdoor' for reaching into the enterprise, TURN will be
blocked by these firewalls.</t>
<t>TURN servers therefore emulate the behavior of NAT devices which
implement address-dependent filtering <xref target="RFC4787"></xref>,
a property common in many firewalls as well. When a NAT or firewall
implements this behavior, packets from an outside IP address are only
allowed to be sent to an internal IP address and port if the internal
IP address and port had recently sent a packet to that outside IP
address. TURN servers introduce the concept of permissions, which
provide exactly this same behavior on the TURN server. An attacker
cannot send a packet to a TURN server and expect it to be relayed
towards the client, unless the client has tried to contact the
attacker first.</t>
<t>It is important to note that some firewalls have policies which are
even more restrictive than address-dependent filtering. Firewalls can
also be configured with address and port dependent filtering, or can
be configured to disallow inbound traffic entirely. In these cases, if
a client is allowed to connect the TURN server, communications to the
client will be less restrictive than what the firewall would normally
allow.</t>
<section title="Faked Permissions">
<t>In firewalls and NAT devices, permissions are granted implicitly
through the traversal of a packet from the inside of the network
towards the outside peer. Thus, a permission cannot, by definition,
be created by any entity except one inside the firewall or NAT. With
TURN, this restriction no longer holds. Since the TURN server sits
outside the firewall, at attacker outside the firewall can now send
a message to the TURN server and try to create a permission for
itself.</t>
<t>This attack is prevented because all messages which create
permissions (i.e., ChannelBind and CreatePermission) are
authenticated.</t>
</section>
<section title="Blacklisted IP Addresses">
<t>Many firewalls can be configured with blacklists which prevent a
client behind the firewall from sending packets to, or receiving
packets from, ranges of blacklisted IP addresses. This is
accomplished by inspecting the source and destination addresses of
packets entering and exiting the firewall, respectively.</t>
<t>If a client connects to a TURN server, it will be able to bypass
such blacklisting policies and communicate with IP addresses which
the firewall would otherwise restrict. This is a problem for other
protocols that provide tunneling functions, such as VPNs. It is
possible to build TURN-aware firewalls which inspect TURN messages,
and check the IP address of the correspondent. TURN messages to
offending destinations can then be rejected. TURN is designed so
that this inspection can be done statelessly.</t>
</section>
<section title="Running Servers on Well-Known Ports">
<t>A malicious client behind a firewall might try to connect to a
TURN server and obtain an allocation which it then uses to run a
server. For example, a client might try to run a DNS server or FTP
server.</t>
<t>This is not possible in TURN. A TURN server will never accept
traffic from a peer for which the client has not installed a
permission. Thus, peers cannot just connect to the allocated port in
order to obtain the service.</t>
</section>
</section>
<section title="Insider Attacks">
<t>In insider attacks, a client has legitimate credentials but defies
the trust relationship that goes with those credentials. These attacks
cannot be prevented by cryptographic means but need to be considered
in the design of the protocol.</t>
<section title="DoS Against TURN Server">
<t>A client wishing to disrupt service to other clients might obtain
an allocation and then flood it with traffic, in an attempt to swamp
the server and prevent it from servicing other legitimate clients.
This is mitigated by the recommendation that the server limit the
amount of bandwidth it will relay for a given username. This won't
prevent a client from sending a large amount of traffic, but it
allows the server to immediately discard traffic in excess.</t>
<t>Since each allocation uses a port number on the IP address of the
TURN server, the number of allocations on a server is finite. An
attacker might attempt to consume all of them by requesting a large
number of allocations. This is prevented by the recommendation that
the server impose a limit of the number of allocations active at a
time for a given username.</t>
</section>
<section title="Anonymous Relaying of Malicious Traffic">
<t>TURN servers provide a degree of anonymization. A client can send
data to correspondent peers without revealing their own IP
addresses. TURN servers may therefore become attractive vehicles for
attackers to launch attacks against targets without fear of
detection. Indeed, it is possible for a client to chain together
multiple TURN servers, such that any number of relays can be used
before a target receives a packet.</t>
<t>Administrators who are worried about this attack can maintain
logs which capture the actual source IP and port of the client, and
perhaps even every permission that client installs. This will allow
for forensic tracing to determine the original source, should it be
discovered that an attack is being relayed through a TURN
server.</t>
</section>
<section title="Manipulating other Allocations">
<t>An attacker might attempt to disrupt service to other users of
the TURN server by sending Refresh requests or CreatePermission
requests which (through source address spoofing) appear to be coming
from another user of the TURN server. TURN prevents this by
requiring that the credentials used in CreatePermission, Refresh,
and ChannelBind messages match those used to create the initial
allocation. Thus, the fake requests from the attacker will be
rejected.</t>
</section>
</section>
<section title="Other Considerations">
<t>Any relay addresses learned through an Allocate request will not
operate properly with IPSec Authentication Header (AH) <xref
target="RFC4302"></xref> in transport or tunnel mode. However,
tunnel-mode IPSec ESP <xref target="RFC4303"></xref> should still
operate.</t>
</section>
</section>
<section title="IANA Considerations">
<t>Since TURN is an extension to STUN <xref target="RFC5389"></xref>,
the methods, attributes and error codes defined in this specification
are new methods, attributes, and error codes for STUN. This section
requests IANA to add these new protocol elements to the IANA registry of
STUN protocol elements.</t>
<t>The codepoints for the new STUN methods defined in this specification
are listed in <xref target="sec-stun-methods"></xref>.</t>
<t>The codepoints for the new STUN attributes defined in this
specification are listed in <xref
target="sec-stun-attributes"></xref>.</t>
<t>The codepoints for the new STUN error codes defined in this
specification are listed in <xref target="sec-stun-errors"></xref>.</t>
<t>IANA is requested to allocate the SRV service name of "turn" for TURN
over UDP or TCP, and the service name of "turns" for TURN over TLS.</t>
<t>IANA is requested to create a registry for TURN channel numbers,
initially populated as follows:<list>
<t>0x0000 through 0x3FFF: Not available for use, since they conflict
with the STUN header.</t>
<t>0x4000 through 0x7FFF: A TURN implementation is free to use
channel numbers in this range.</t>
<t>0x8000 through 0xFFFF: Reserved.</t>
</list>Any change to this registry must be made through an IETF
Standards Action.</t>
</section>
<section title="IAB Considerations">
<t>The IAB has studied the problem of "Unilateral Self Address Fixing",
which is the general process by which a client attempts to determine its
address in another realm on the other side of a NAT through a
collaborative protocol reflection mechanism <xref
target="RFC3424"></xref>. The TURN extension is an example of a protocol
that performs this type of function. The IAB has mandated that any
protocols developed for this purpose document a specific set of
considerations. These considerations and the responses for TURN are
documented in this section.</t>
<t>Consideration 1: Precise definition of a specific, limited-scope
problem that is to be solved with the UNSAF proposal. A short term fix
should not be generalized to solve other problems. Such generalizations
lead to the prolonged dependence on and usage of the supposed short term
fix -- meaning that it is no longer accurate to call it "short
term".</t>
<t>Response: TURN is a protocol for communication between a relay (=
TURN server) and its client. The protocol allows a client that is behind
a NAT to obtain and use a public IP address on the relay. As a
convenience to the client, TURN also allows the client to determine its
server-reflexive transport address.</t>
<t>Consideration 2: Description of an exit strategy/transition plan. The
better short term fixes are the ones that will naturally see less and
less use as the appropriate technology is deployed.</t>
<t>Response: TURN will no longer be needed once there are no longer any
NATs. Unfortunately, as of the date of publication of this document, it
no longer seems very likely that NATs will go away any time soon.
However, the need for TURN will also decrease as the number of NATs with
the mapping property of Endpoint-Independent Mapping <xref
target="RFC4787"></xref> increases.</t>
<t>Consideration 3: Discussion of specific issues that may render
systems more "brittle". For example, approaches that involve using data
at multiple network layers create more dependencies, increase debugging
challenges, and make it harder to transition.</t>
<t>Response: TURN is "brittle" in that it requires the NAT bindings
between the client and the server to be maintained unchanged for the
lifetime of the allocation. This is typically done using keep-alives. If
this is not done, then the client will lose its allocation and can no
longer exchange data with its peers.</t>
<t>Consideration 4: Identify requirements for longer term, sound
technical solutions; contribute to the process of finding the right
longer term solution.</t>
<t>Response: The need for TURN will be reduced once NATs implement the
recommendations for NAT UDP behavior documented in <xref
target="RFC4787"></xref>. Applications are also strongly urged to use
ICE <xref target="I-D.ietf-mmusic-ice"></xref> to communicate with
peers; though ICE uses TURN, it does so only as a last resort, and uses
it in a controlled manner.</t>
<t>Consideration 5: Discussion of the impact of the noted practical
issues with existing deployed NATs and experience reports.</t>
<t>Response: Some NATs deployed today exhibit a mapping behavior other
than Endpoint-Independent mapping. These NATs are difficult to work
with, as they make it difficult or impossible for protocols like ICE to
use server-reflexive transport addresses on those NATs. A client behind
such a NAT is often forced to use a relay protocol like TURN because
"UDP hole punching" techniques <xref target="RFC5128"></xref> do not
work.</t>
</section>
<section title="Open Issues">
<t>Note to RFC Editor: Please remove this section prior to publication
of this document as an RFC.</t>
<t>This section lists the known issues in this version of the
specification.</t>
<t>(No known issues at this time).</t>
</section>
<section title="Changes from Previous Versions">
<t>Note to RFC Editor: Please remove this section prior to publication
of this document as an RFC.</t>
<t>This section lists the technical and major editorial changes between
the various versions of this specification. Minor editorial changes are
not described.</t>
<section title="Changes from -15 to -16">
<t><list style="symbols">
<t>Removed much of the text around the usage of ALTERNATE-SERVER.
Previously, this document modified the specification of
ALTERNATE-SERVER in the draft version of RFC 5389, but those
modifications made it into the final version of RFC 5389 and so
the extra text is no longer needed in this document. </t>
<t>Clarified the text in <xref target="sec-permissions"></xref>
and <xref target="sec-sending-data-indication"></xref> around the
checking of permissions when relaying a UDP datagram.</t>
<t>Added text specifying the requirements for support the various
transport methods.</t>
</list></t>
</section>
<section title="Changes from -14 to -15">
<t><list style="symbols">
<t>Added text saying that TURN servers and client MUST implement
the Long-Term Credential Mechanism. Added text strongly
recommending that servers require that all requests be
authenticated. Noted a few cases where not using the Long-Term
Credential Mechanism might be acceptable.</t>
<t>Added text to section <xref
target="sec-allocate-error-response"></xref> saying that unknown
error responses must be handled as per <xref
target="RFC5389"></xref>.</t>
<t>Added text clarifying the exact length of each attribute and
reminding the reader that certain attributes must be immediately
followed by 1 to 3 padding bytes.</t>
<t>Added a sentence to the acknowledgment section thanking Marc
Petit-Huginen for his efforts in implementing many previous
versions of the specification.</t>
<t>Fixed a number of minor document errors.</t>
</list></t>
</section>
<section title="Changes from -13 to -14">
<t><list style="symbols">
<t>Reworded the text in <xref target="sec-rcv-allocate"></xref>
and <xref target="sec-rcv-refresh"></xref> to more clearly
describe how the allocation lifetime is computed in the case where
a client requests a lifetime that is greater than both the default
lifetime and the server's maximum allowed lifetime.</t>
<t>In <xref target="sec-permissions"></xref>, changed the term
"default permission lifetime" to "Permission Lifetime" to make it
clearer that the lifetime of a permission is not configurable.</t>
<t>In <xref target="sec-rcv-allocate"></xref>, swapped the steps
that check the RESERVATION-TOKEN and EVEN-PORT attributes to
correctly handle the case where an Allocate request contains both
attributes. The new text correctly returns 400 Bad Request in this
case.</t>
<t>Added text in the Overview section to describe why various
timer values were chosen.</t>
<t>Added a sentence to the IAB consideration section saying that
the disappearance of NATs in the near-term seems unlikely.</t>
<t>The former "Other Features" section of the Overview has been
replaced with a series of sections describing various secondary
features of TURN, and the text describing and motivating these
secondary features has been expanded. As a part of this rewrite,
there is now a section that describes how to avoid IP
fragmentation when using TURN.</t>
<t>Added some additional text in the Overview to explain how a
client would select between UDP, TCP, and TLS transport.</t>
<t>Fixed various minor typos.</t>
</list></t>
</section>
<section title="Changes from -12 to -13">
<t><list style="symbols">
<t>Added a new error code: 403 (Forbidden).</t>
<t>When processing a CreatePermission or ChannelBind request
containing a XOR-PEER-ADDRESS attribute, the server is allow to
reject certain IP address and port combinations for administrative
or other reasons by returning a 403 (Forbidden) error.</t>
<t>Added a request to IANA to establish a registery for channel
numbers.</t>
<t>Clarified the usage of the nonce value: a new random nonce
SHOULD be selected for each Allocate attempt, and the nonce SHOULD
be expired at least once an hour. Referenced <xref
target="RFC4086"></xref> for guidelines on selecting the nonce
value.</t>
<t>Made a number of minor editoral changes.</t>
</list></t>
</section>
<section title="Changes from -11 to -12">
<t><list style="symbols">
<t>Changed the port numbers used in the examples for the client,
the peers, and the relayed-transport-address to put them in the
Dynamic port range. They were previously in the Registered port
range, which was arguably unrealistic.</t>
<t>Noted that the XOR-MAPPED-ADDRESS attribute is defined in RFC
5389.</t>
<t>Used the codepoint names (Not-ECT, ECT(0), ECT(1), and CE) when
talking about the ECN field.</t>
<t>Updated the Introduction to note that the client must not only
communicate its relayed-transport-address to the peers, but also
learn the peers' server-reflexive transport addresses. As a
result, removed the suggestion that the client could use a webpage
to communicate with its peers.</t>
<t>Added a description of the "TURN Loop attack" and its
mitigation to the Security Considerations section.</t>
<t>Fixed some errors in the examples in the Overview section. They
had not been updated to be consistent with the change introduced
in version -11 that a permission must be created before a client
can send data to a peer.</t>
<t>In the Additional Features subsection of the Overview, reworded
the discussion of what end-to-end features are preserved by TURN.
The previous text said that a number of features did not work, but
as of version -11, these features _may_ work. At the same time,
added a sentence noting that any Path MTU Discovery mechanism
using the DONT-FRAGMENT attribute will not receive ICMP messages
and will thus have to use techniques like those described in <xref
target="RFC4821"></xref>.</t>
<t>Added the recommendation that, when TCP transport is used
between the client and the server, both ends should close the
connection if they notice a long sequence of invalid TURN
messages. A likely cause of this is an undetected bit error
corrupting a length field somewhere.</t>
<t>Reworded the paragraph explaining that channel bindings are
per-allocation to further stress this point.</t>
<t>In the discussion on setting the fragmentation fields, added a
sentence saying that the client or server should follow the normal
rules for fragmentation as described in <xref
target="RFC1122"></xref>.</t>
</list></t>
</section>
<section title="Changes from -10 to -11">
<t><list style="symbols">
<t>Clarified that, when the client is redirected to an alternate
server, the client uses the same transport protocol to the
alternate server as it did to the original server.</t>
<t>Clarified the information that the server needs to store to
authenticate requests and to compute the message-integrity on
responses. Noted that the server need not store the password
explicitly, but can instead store the key value, which may be
desirable for security reasons.</t>
<t>Clarified that TURN runs on the same ports as TURN by default,
but noted that a server can use a different port because TURN has
its own SRV service names. Strengthened the language for using the
SRV procedures from "typically" to "SHOULD". Also added a sentence
in the IANA considerations section requesting that IANA reserve
the service names for TURN; previously they were described in the
text but not mentioned in the IANA considerations section.</t>
<t>Added a detailed example, complete with attributes and their
values, of the use of TURN.</t>
<t>Reduced the range of channel numbers. Channel numbers now range
from 0x4000 through 0x7FFF. Values in the range 0x8000 through
0xFFFF are now reserved.</t>
<t>Rewrote the IAB Considerations section to directly address the
considerations listed in <xref target="RFC3424"></xref>.</t>
<t>Generalized the 508 error code so it can be used for any sort
of capacity-related problem. This error code was previously
allowed only in Allocate responses, but is now also allowed in
CreatePermission and ChannelBind responses to indicate that the
server is unable to carry out the request due to some capacity
problem.</t>
<t>Changed the syntax of the CreatePermission request to allow
multiple XOR-PEER-ADDRESS attributes to appear in the message, so
that multiple permissions can be created or refreshed at the same
time.</t>
<t>Added the restriction that the server must already have a
permission installed for the IP address in the XOR-PEER-ADDRESS
attribute of a Send indication, otherwise the Send indication is
ignored by the server.</t>
<t>Put back the preferred behaviors into <xref
target="sec-ip-header-fields"></xref>, reversing the change made
in version -10.</t>
<t>Explicitly allow the server to restrict the range of IP
addresses and ports it is willing to relay data too.</t>
</list></t>
</section>
<section title="Changes from -09 to -10">
<t><list style="symbols">
<t>Changed the recommendation for using the SOFTWARE attribute.
Previously its use was recommended in all requests and responses;
now it is only recommended in Allocate and Refresh requests and
responses, though it may appear elsewhere. Also, version -09
incorrectly referred to this attribute as "SOFTWARE-TYPE".</t>
<t>Changed the name of the PEER-ADDRESS and RELAYED-ADDRESS
attributes to XOR-PEER-ADDRESS and XOR-RELAYED-ADDRESS
respectively for consistency with other specifications.</t>
<t>Removed the concept of a "preserving" allocation. All
allocations are now non-preserving. This simplifies the base
specification and allows it to advance more rapidly; see the
discussion in the BEHAVE meeting of 29 July 2008. The concept of a
preserving allocation will be advanced as an extension to TURN. As
part of this change, the P bit in the REQUESTED-PROPS attribute,
the ICMP attribute, and ICMP message relaying was removed.
Further, in <xref target="sec-ip-header-fields"></xref>, the
preferred behaviors were removed, leaving the alternate behaviors
as the specified behaviors.</t>
<t>Replaced the REQUESTED-PROPS attribute with the EVEN-PORT
attribute. The new attribute lacks the feature of the old
attribute of being an alternate way to specify new allocation
properties. As a consequence, the only way to specify a new
allocation property is to define a new attribute.</t>
<t>Added text recommending that the client check that the IP
address in XOR-PEER-ADDRESS attribute in a received Data
indication is one with which the client believes there is an
active permission. Similarly, it is recommended that the client
check that a permission exist when receiving a ChannelData
message.</t>
<t>Added text recommending that the client delete the allocation
if it receives a ChannelBind failure response on an unbound
channel.</t>
<t>Added the CreatePermission request/response transaction which
adds or updates permissions, and removed the ability for Send
indications and ChannelBind messages to install or update
permissions. The net effect is that only authenticate-able
messages (i.e., CreatePermission requests and ChannelBind
requests) can install or refresh permissions; unauthenticate-able
Send indications and ChannelData messages do not.</t>
<t>Removed all support for IPv6. All IPv6 support, including ways
of relaying between IPv4 and IPv6, will now be covered in <xref
target="I-D.ietf-behave-turn-ipv6"></xref>.</t>
<t>Reserved attribute code point 0x0021. This was previously used
for the TIMER-VAL attribute, which was removed when the
SetActiveDestination feature was removed.</t>
<t>Added the DONT-FRAGMENT attribute which allows the client to
request that the server set the DF bit when sending the UDP
datagram to the peer. This attribute may appear in both Allocate
requests and Send indications.</t>
<t>Changed how the ALTERNATE-SERVER attribute is used. The
attribute can no longer be used with any error code, but must be
used with 300 (Try Alternative). It can now appear in
unauthenticated responses, however there are restrictions around
how the subsequent Allocate request is authenticated.</t>
<t>Reworked the details of how idempotency of requests is handled,
making it clear that the stack can either remember all
transactions for 40 seconds, or can handle this using the
so-called "stateless stack approach". Made some changes to the
semantics of the Allocate, Refresh, and ChannelBind requests as a
consequence.</t>
<t>Added the requirement that a client cannot re-use previously
bound channel number or transport address until 5 minutes after
the channel binding expires. This avoids various race
conditions.</t>
<t>Removed the requirement that an allocation cannot be re-used
within 2 minutes of having been deleted. This requirement was put
in place to prevent mis-delivered packets but is no longer seen as
having any real value.</t>
<t>Added a recommendation that the server impose quotas on both
the number of allocations and the amount of bandwidth a given
username can use at one time. These quotas help protect against
denial-of-service attacks.</t>
<t>Completely rewrote the security considerations section.</t>
<t>Made quite a few changes to the descriptive text in both the
Overview and the normative text to try to further clarify
concepts.</t>
</list></t>
</section>
<section title="Changes from -08 to -09">
<t><list style="symbols">
<t>Added text to properly define the ICMP attribute. This
attribute was introduced in TURN-08, but not fully defined due to
an oversight. Clarified that the attribute can appear in a Data
indication, but not a Send indication. Added text to the section
on receiving a Data indication that points out that this attribute
may be present.</t>
<t>Changed the wording around the handling of the DSCP field to
allow the server to set the DSCP to an arbitrary value if the next
hop is a Diff-Serv classifier and marker.</t>
<t>When the server generates a 508 response due to an unsupported
flag in the REQUESTED-PROPS attribute, the server now includes the
REQUESTED-PROPS attribute in the response with all the flags it
supports set to 1. This allows the client to see if the server
does not understand one of its flags. Similarly, the client is now
allowed to immediately retry the request if it modifies the
included REQUESTED-PROPS attribute.</t>
<t>Clarified that the REQUESTED-PROPS attribute can be used in
conjunction with the RESERVATION-TOKEN attribute as long as both
the E and R bits are 0. The spec previously contradicted itself on
this point.</t>
<t>Clarified that when the server receives a ChannelData message
with a length field of 0, it sends a UDP Datagram to the peer that
contains no application data.</t>
<t>Rewrote some text around relaying incoming UDP Datagrams to
avoid duplication of text in the Data indication and Channel
sections.</t>
<t>Added a note that points out that the on-going work on
randomizing port allocations <xref
target="I-D.ietf-tsvwg-port-randomization"></xref> may be
applicable to TURN.</t>
<t>Clarified that the Allocate request containing a
RESERVATION-TOKEN attribute can use any 5-tuple, and that 5-tuple
need not have any specific relationship to the 5-tuple of the
Allocate request that created the reservation.</t>
<t>Added a note that discusses retransmitted Allocate requests
over UDP where the first request receives a failure response, but
the second receives a success response. The server may elect to
remember transmitted failure responses to avoid this
situation.</t>
<t>Added text about the usage of the SOFTWARE-TYPE attribute
(formerly known as the SERVER attribute) in TURN messages.</t>
<t>Rewrote the text in the Overview that motivates why TURN
supports TCP and TLS between the client and the server. The
previous text had been identified by various readers as inadequate
and misleading.</t>
<t>Rewrote the section how a server handles a Refresh request to
clarify processing in various error conditions. The new text makes
it clear that it is OK to delete a non-existent allocation. It
also clarifies how to handle retransmissions of Refresh requests
over UDP.</t>
<t>Renamed the "RELAY-ADDRESS" attribute to "RELAYED-ADDRESS",
since the text consistently uses the term "relayed transport
address" for the concept and ICE uses the term "relayed
candidate".</t>
<t>Changed the codepoint assigned to the error code "Wrong
Credentials" from 438 to 441 to avoid a conflict with the "Stale
Nonce" error code of STUN.</t>
<t>Changed the text to consistently use non-capitalized "request",
"response" and "indication", except in headings, error code names,
etc.</t>
<t>Added a note mentioning that TURN packets can be demuxed from
other packets arriving on the same socket by looking at the
5-tuple of the arriving packet.</t>
<t>Clarified that there are no required attributes is a
ChannelBind success response.</t>
</list></t>
</section>
<section title="Changes from -07 to -08">
<t><list style="symbols">
<t>Removed the BANDWIDTH attribute and all associated text
(including error code 507 "Insufficient Bandwidth Capacity"), as
the requirements for this feature were not clear and it was felt
the feature could be easily added later.</t>
<t>Changed the format of the REQUESTED-PROPS attribute from a
one-byte field to a set of bit flags. Changed the semantics of the
unused portion of the value from RFFU to "MUST be 0" to give a
more desirable behavior when new flags are defined.</t>
<t>Introduced the concept of Preserving vs. Non-Preserving
allocations. As a result, completely revamped the rules for how to
set the fields in the IP header, and added rules for relaying ICMP
messages when the allocation is Preserving.</t>
</list></t>
</section>
<section title="Changes from -06 to -07">
<t><list style="symbols">
<t>Rewrote the General Behavior section, making various changes in
the process.</t>
<t>Changed the usage of authentication from MUST to SHOULD.</t>
<t>Changed the requirement that subsequent requests use the same
username and password from MUST to SHOULD to allow for the
possibility of changing the credentials using some unspecified
mechanism.</t>
<t>Introduced a 438 (Wrong Credentials) error which is used when a
non-Allocate request authenticates but does not use the same
username and password as the Allocate request. Having a separate
error code for this case avoids the client being confused over
what the error actually is.</t>
<t>The server must now prevent the relayed transport address and
the 5-tuple from being reused in different allocations for 2
minutes after the allocation expires.</t>
<t>Changed the usage of FINGERPRINT from MUST NOT to MAY, to allow
for the possible multiplexing of TURN with some other
protocol.</t>
<t>Rewrote much of the section on Allocations, splitting it into
three new sections (one on allocations in general, one on creating
an allocation, and one on refreshing an allocation).</t>
<t>Replaced the mechanism for requesting relayed transport
addresses with specific properties. The new mechanism is less
powerful: a client can request an even port, or a pair of ports,
but cannot request a single odd port or a specific port as was
possible under the old mechanism. Nor can the client request a
specific IP address.</t>
<t>Changed the rules for handling ALTERNATE-SERVER, removing the
requirement that the referring server have "positive knowledge"
about the state of the alternate server. The new rules instead
rely on text in STUN to prevent referral loops.</t>
<t>Changed the rules for allocation lifetimes. Allocations
lifetimes are now a minimum of 10 minutes; the client can ask for
longer values, but requests for shorter values are ignored. The
text now recommends that the client refresh an allocation one
minute before it expires.</t>
<t>Put in temporary procedures for handling the BANDWIDTH
attribute, modelled on the LIFETIME attribute. These procedures
are mostly placeholders and likely to change in the next
revision.</t>
<t>Added a detailed description of how a client reacts to the
various errors it can receive in reply to an Allocate request.
This replaces the various descriptions that were previously
scattered throughout the document, which were inconsistent and
sometimes contradictory.</t>
<t>Added a new section that gives the normative rules for
permissions.</t>
<t>Changed the rules around permission lifetimes. The text used to
recommend a value of one minute; it MUST now be 5 minutes.</t>
<t>Removed the errors "Channel Missing or Invalid", "Peer Address
Missing or Invalid" and "Lifetime Malformed or Invalid" and used
400 "Bad Request" instead.</t>
<t>Rewrote portions of the section on Send and Data indications
and the section on Channels to try to make the client vs. server
behavior clearer.</t>
<t>Channel bindings now expire after 10 minutes, and must be
refreshed to keep them alive.</t>
<t>Binding a channel now installs or refreshes a permission for
the IP address of corresponding peer.</t>
<t>Changed the wording describing the situation when the client
sends a ChannelData message before receiving the ChannelBind
success response. -06 said that client SHOULD NOT do this; -07 now
says that a client MAY, but describes the consequences of doing
it.</t>
<t>Added a section discussing the setting of fields in the IP
header.</t>
<t>Replaced the REQUESTED-PORT-PROPS attribute with the
REQUESTED-PROPS attribute that has a different format and
semantics, but reuses the same code point.</t>
<t>Replaced the REQUESTED-IP attribute with the RESERVATION-TOKEN
attribute, which has a different format and semantics, but reuses
the same code point.</t>
<t>Removed error codes 443 and 444, and replaced them with 508
(Insufficient Port Capacity). Also changed the error text for code
507 from "Insufficient Capacity" to "Insufficient Bandwidth
Capacity".</t>
</list></t>
</section>
<section title="Changes from -05 to -06">
<t><list style="symbols">
<t>Changed the mechanism for allocating channels to the one
proposed by Eric Rescorla at the Dec 2007 IETF meeting.</t>
<t>Removed the framing mechanism (which was used to frame all
messages) and replaced it with the ChannelData message. As part of
this change, noted that the demux of ChannelData messages from
TURN messages can be done using the first two bits of the
message.</t>
<t>Rewrote the sections on transmitted and receiving data as a
result of the above to changes, splitting it into a section on
Send and Data indications and a separate section on channels.</t>
<t>Clarified the handling of Allocate request messages. In
particular, subsequent Allocate request messages over UDP with the
same transaction id are not an error but a retransmission.</t>
<t>Restricted the range of ports available for allocation to the
Dynamic and/or Private Port range, and noted when ports outside
this range can be used.</t>
<t>Changed the format of the REQUESTED-TRANSPORT attribute. The
previous version used 00 for UDP and 01 for TCP; the new version
uses protocol numbers from the IANA protocol number registry. The
format of the attribute also changed.</t>
<t>Made a large number of changes to the non-normative portion of
the document to reflect technical changes and improve the
presentation.</t>
<t>Added the Issues section.</t>
</list></t>
</section>
<section title="Changes from -04 to -05">
<t><list style="symbols">
<t>Removed the ability to allocate addresses for TCP relaying.
This is now covered in a separate document. However, communication
between the client and the server can still run over TCP or
TLS/TCP. This resulted in the removal of the Connect method and
the TIMER-VAL and CONNECT-STAT attributes.</t>
<t>Added the concept of channels. All communication between the
client and the server flows on a channel. Channels are numbered
0..65535. Channel 0 is used for TURN messages, while the remaining
channels are used for sending unencapsulated data to/from a remote
peer. This concept adds a new Channel Confirmation method and a
new CHANNEL-NUMBER attribute. The new attribute is also used in
the Send and Data methods.</t>
<t>The framing mechanism formally used just for stream-oriented
transports is now also used for UDP, and the former Type and
Reserved fields in the header have been replaced by a Channel
Number field. The length field is zero when running over UDP.</t>
<t>TURN now runs on its own port, rather than using the STUN port.
The use of channels requires this.</t>
<t>Removed the SetActiveDestination concept. This has been
replaced by the concept of channels.</t>
<t>Changed the allocation refresh mechanism. The new mechanism
uses a new Refresh method, rather than repeating the Allocation
transaction.</t>
<t>Changed the syntax of SRV requests for secure transport. The
new syntax is "_turns._tcp" rather than the old "_turn._tls". This
change mirrors the corresponding change in STUN SRV syntax.</t>
<t>Renamed the old REMOTE-ADDRESS attribute to PEER-ADDRESS, and
changed it to use the XOR-MAPPED-ADDRESS format.</t>
<t>Changed the RELAY-ADDRESS attribute to use the
XOR-MAPPED-ADDRESS format (instead of the MAPPED-ADDRESS
format)).</t>
<t>Renamed the 437 error code from "No Binding" to "Allocation
Mismatch".</t>
<t>Added a discussion of what happens if a client's public binding
on its outermost NAT changes.</t>
<t>The document now consistently uses the term "peer" as the name
of a remote endpoint with which the client wishes to
communicate.</t>
<t>Rewrote much of the document to describe the new concepts. At
the same time, tried to make the presentation clearer and less
repetitive.</t>
</list></t>
</section>
</section>
<section title="Acknowledgements">
<t>The authors would like to thank the various participants in the
BEHAVE working group for their many comments on this draft. Marc
Petit-Huguenin, Remi Denis-Courmont, Jason Fischl, Derek MacDonald,
Scott Godin, Cullen Jennings, Lars Eggert, Magnus Westerlund, Benny
Prijono, and Eric Rescorla have been particularly helpful, with Eric
suggesting the channel allocation mechanism, Cullen suggesting the
REQUESTED-PORT-PROPS mechanism, and Marc spending many hours
implementing the preliminary versions to look for problems. Christian
Huitema was an early contributor to this document and was a co-author on
the first few drafts. Finally, the authors would like to thank Dan Wing
for both his contributions to the text and his huge help in restarting
progress on this draft after work had stalled.</t>
</section>
</middle>
<back>
<references title="Normative References">
<?rfc include="reference.RFC.5389"?>
<?rfc include="reference.RFC.2119"?>
<?rfc include='reference.RFC.2474'?>
<?rfc include='reference.RFC.3168'?>
<?rfc include='reference.RFC.1122'?>
</references>
<references title="Informative References">
<?rfc include='reference.RFC.1191'?>
<?rfc include='reference.RFC.0791'?>
<?rfc include='reference.RFC.1918'?>
<?rfc include="reference.RFC.3424"?>
<?rfc include='reference.RFC.4787'?>
<?rfc include="reference.I-D.ietf-mmusic-ice"?>
<?rfc include="reference.I-D.ietf-behave-turn-tcp"?>
<?rfc include="reference.I-D.ietf-behave-turn-ipv6"?>
<?rfc include='reference.I-D.ietf-tsvwg-port-randomization'?>
<?rfc include='reference.RFC.5128'?>
<?rfc include='reference.RFC.1928'?>
<?rfc include='reference.RFC.3550'?>
<?rfc include='reference.RFC.3711'?>
<?rfc include='reference.RFC.4302'?>
<?rfc include='reference.RFC.4303'?>
<?rfc include='reference.RFC.4821'?>
<?rfc include='reference.RFC.3261'?>
<?rfc include='reference.I-D.rosenberg-mmusic-ice-nonsip'?>
<?rfc include='reference.RFC.4086'?>
<reference anchor="Frag-Harmful">
<front>
<title>Fragmentation Considered Harmful</title>
<author fullname="" surname="Kent">
<organization />
</author>
<author surname="Mogul">
<organization />
</author>
</front>
<annotation>Proc. SIGCOMM ‘87, vol. 17, No. 5, October
1987</annotation>
</reference>
<reference anchor="Port-Numbers"
target="http://www.iana.org/assignments/port-numbers">
<front>
<title>IANA Port Numbers Registry</title>
</front>
</reference>
<reference anchor="Protocol-Numbers"
target="http://www.iana.org/assignments/protocol-numbers">
<front>
<title>IANA Protocol Numbers Registry</title>
<author>
<organization></organization>
</author>
<date year="2005" />
</front>
</reference>
</references>
</back>
</rfc>| PAFTECH AB 2003-2026 | 2026-04-22 23:52:05 |