One document matched: draft-alvestrand-dispatch-rtcweb-datagram-00.xml
<?xml version="1.0" encoding="US-ASCII"?>
<!DOCTYPE rfc SYSTEM "rfc2629.dtd">
<?rfc toc="yes"?>
<?rfc tocompact="yes"?>
<?rfc tocdepth="3"?>
<?rfc tocindent="yes"?>
<?rfc symrefs="yes"?>
<?rfc sortrefs="yes"?>
<?rfc comments="yes"?>
<?rfc inline="yes"?>
<?rfc compact="yes"?>
<?rfc subcompact="no"?>
<rfc category="exp" docName="draft-alvestrand-dispatch-rtcweb-datagram-00"
ipr="trust200902">
<front>
<title abbrev="webm datagram">A Datagram Transport for the RTC-Web
profile</title>
<author fullname="Harald Tveit Alvestrand" initials="H."
surname="Alvestrand">
<organization>Google</organization>
<address>
<postal>
<street>Kungsbron 2</street>
<city>Stockholm</city>
<region></region>
<code>11122</code>
<country>Sweden</country>
</postal>
<email>harald@alvestrand.no</email>
</address>
</author>
<date day="10" month="November" year="2010" />
<abstract>
<t>This document describes a combination and profiling of existing IETF
protocols to provide a datagram service that is suitable as a generic
transport substrate for the RTC-Web family of real-time audio/video
applications.</t>
</abstract>
<note title="Requirements Language">
<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>
</note>
</front>
<middle>
<section title="Introduction">
<t>When transporting audio / video data between participants on the
current Internet, there are a number of obstacles to be faced.</t>
<t>Among them are NAT boxes, firewalls, connection interruptions, the
availability of multiple paths between participants, and capacity
issues.</t>
<t>This memo describes a combination of existing protocols that can be
used to achieve a seamless datagram transport service across this very
heterogenous environment.</t>
<t>An overview of the effort of which this is a part can be found in the
overview document, <xref target="overview"></xref>.</t>
</section>
<section title="Terminology">
<t>This draft uses a couple of commonly used terms in quite specific
ways. The reader is advised to study these definitions carefully.</t>
<t>(TODO: Agree on terminology to use)</t>
<t><list style="hanging">
<t hangText="Session">An association with two endpoints, between
which datagrams flow.</t>
<t hangText="Datagram">A sequence of octets, of a given length. In
this specification, a datagram does not carry addressing
information.</t>
<t hangText="Channel">One means of transporting a datagram over a
session. A session may have multiple channels at any time.</t>
<t hangText="Endpoint">One end of a session. This document does not
distinguish between an initiator and a responder endpoint.</t>
<t hangText="Control channel">A means of communication between the
endpoints of a session that does not require a transport to be
active. Typically, authentication, authorization and negotiation is
carried out over the control channel. The specification of the
control channel is out of scope for this specification.</t>
</list></t>
</section>
<section title="Service model">
<t>The basic model presented is a datagram model. On top of this one can
layer various services, such as pseudoTCP (REF), RTP<xref
target="RFC3550"></xref> or any other higher layer protocol that is
capable of running across a datagram service.</t>
<t>The addressing model departs from the traditional Internet model in
that end point addresses are not used for endpoint identification, only
for channel establisment; instead, an initial packet exchange, using ICE
<xref target="RFC5245"></xref>, is used to bind a channel to a
prenegotiated session.</t>
<t>The datagram service is not completely transparent; in particular, it
is not possible to carry a datagram where the two highest bits of the
first octet are zero and octet 5 to 8 contain the value 0x2112A442,
since these datagrams are reserved for use of the STUN protocol (RFC
5389 section 6).</t>
</section>
<section title="Channel types">
<t></t>
<section title="UDP channel">
<t>An UDP channel is negotiated using ICE. Each datagram is simply
carried as the content of an UDP packet.</t>
</section>
<section title="TCP channel">
<t>A TCP channel consists of a TCP connection, over which are sent
datagrams packaged according to (REF). The binding of a TCP channel is
done by executing an ICE negotiation over the first few packets passed
across the TCP channel.</t>
</section>
<section title="TLS channel">
<t>A TLS channel consists of a standard TLS negotiation, followed by
passing datagrams over the TLS record layer; the length fields of
(REF) are not used. A TLS channel is bound to its session by
<insert process description>.</t>
</section>
<section title="DTLS channel">
<t>A DTLS channel is created by executing a DTLS connection
negotiation, followed by datagram exchange, where the datagrams are
protected by DTLS mechanisms. The DTLS channel is bound to its session
by <insert process>.</t>
<t></t>
</section>
<section title="Channels with relay">
<t>If there is no possibility of setting up a direct connection, a
relay must be used. The specification from TURN <xref
target="RFC5766"></xref>is used.</t>
</section>
</section>
<section title="Channel setup, teardown and usage">
<t>The service model envisioned here is that all datagrams arriving on a
session are considered equally valid. The session gives no guarantees
against duplication, loss or reordering; such concerns are left to the
higher protocol layers.</t>
<t>The expected normal usage is that two endpoints will exchange
addressing information that can be used for a series of potential
channels, that the endpoints will probe for working channels using ICE
(RFC 5245), and use the "best" candidate, while using the STUN probing
facilities to keep some number of "second best" candidates alive if the
"best" candidate stops working.</t>
<t>A data-sending endpoint may unilaterally decide to start or stop
using an established channel at any time. No negotiation is
necessary.</t>
<t>A receiving endpoint will learn that a channel has been removed by
not seeing any more STUN keepalive messages on that channel within
<timeout>.</t>
<t>A session is considered closed when all channels that have been
successfully established have timed out.</t>
</section>
<section title="An URI scheme for datagram channels">
<t>This URI scheme is mainly included in order to make it easy for APIs
that normally use URIs as what they use to refer to objects.</t>
<t>The DGSESSION URI scheme specifies the information required for a
session; it consists of two parts:</t>
<t><list style="symbols">
<t>An absolute reference, which includes the user name and password
used to establish the connection.</t>
<t>A series of addressing hints, which include the data necessary to
establish a channel.</t>
</list><TODO: Fill out an URI registration template for the
scheme></t>
<t>Example:</t>
<t>dgsession:username:password?ipv4:12.34.56:udp:12345&ipv6:2002::dead:beef:tcp:80&ipv4:12.34.56.78:tls:443</t>
<t>The sequence of addressing hints is an indication of the preference
of the URL constructor for the sequence in which to try these
candidates; the most preferred address is the one to the left.</t>
<t>Note that a DGSESSION URI is a capability; anyone with the URI will
be able to connect to the entity. They should therefore be handled in
the same way as (short-term) passwords, and never passed in the
clear.</t>
</section>
<section anchor="IANA" title="IANA Considerations">
<t anchor="uridef">This document registers the URI scheme from section
<xref target="uridef"></xref>.</t>
<t>Note to RFC Editor: this section may be removed on publication as an
RFC.</t>
</section>
<section anchor="Security" title="Security Considerations">
<t>As with all layered protocols, it is a matter for the application to
decide which level security should be provided at. For instance, an RTP
session protected using SRTP <ref> can be considered to not need
any further safeguards against interception, modification or replay, so
can be passed "in the clear" across any channel type here. For data
without such protection, adequate measures need to be taken; in
particular, it is trivially easy for someone with the ability to snoop
and insert packets to insert fake packets into an established UDP
channel.</t>
<t>The main defense against denial-of-service attacks is the fact that
the ICE mechanisms were designed for low cost refusal of unauthorized
connections.</t>
</section>
<section anchor="Acknowledgements" title="Acknowledgements">
<t></t>
</section>
</middle>
<back>
<references title="Normative References">
<?rfc include="reference.RFC.2119"?>
<?rfc include='reference.RFC.3550'?>
<?rfc include='reference.RFC.5245'?>
<?rfc include='reference.RFC.5766'?>
</references>
<references title="Informative References">
<reference anchor="overview">
<front>
<title>Overview: Real Time Protocols for Brower-based
Applications</title>
<author fullname="Harald" initials="H" surname="Alvestrand">
<organization>Google</organization>
</author>
<date day="9" month="November" year="2010" />
</front>
</reference>
<?rfc ?>
</references>
</back>
</rfc>
| PAFTECH AB 2003-2026 | 2026-04-23 08:24:34 |