[Spud] Unified UDP transport and tunnel protocol?

Tom Herbert <tom@herbertland.com> Tue, 31 March 2015 03:49 UTC

MIME-Version: 1.0
Date: Mon, 30 Mar 2015 20:48:59 -0700
Message-ID: <CALx6S34uEX0+bSq8sAXKJ=Gg1AgYEJ6XfMXQA90QN8_Qf=g3gA@mail.gmail.com>
From: Tom Herbert <tom@herbertland.com>
To: spud@ietf.org
Content-Type: text/plain; charset=UTF-8
Archived-At: <http://mailarchive.ietf.org/arch/msg/spud/KKfXD1JR_a3TDb5_W3sxN_Kn5uE>
Subject: [Spud] Unified UDP transport and tunnel protocol?
Precedence: list

Hello,

As I mentioned at the SPUD BOF last week, it would be nice if the SPUD
effort and other efforts for UDP encapsulation (such as work being
done in NVO3) were somehow coordinated. Looking at the prototype
protocol for SPUD I do believe that there are some opportunities here.
In particular, I think that the features of SPUD (session control,
magic number, DTLS, and some coded optional data) are also features we
want in UDP encapsulation. There might also be features being defined
on the encapsulation side that may be of interest to SPUD.

We have proposed Generic UDP Encapsulation (GUE presented in int-area
last week) to provide a simple and extensible UDP encapsulation
protocol. The base protocol as well as several extensions have been
defined. We are planning to define a session identifier option and I
think it makes sense to define that in light of the SPUD model.  The
magic number and DTLS can also supported. GUE is designed for
tunneling (encapsulation of IP protocol), but could be defined as a
more general "substrate" protocol to conceivably provide a common
header for both UDP encapsulation and UDP transports.

If you are interested, the GUE draft is draft-herbert-gue-03. I hope
have a draft on the session identifier option for GUE soon.

Thanks,
Tom

[Spud] Unified UDP transport and tunnel protocol? Tom Herbert