Re: [TLS] Bakeoffs

[Nikos Mavrogiannopoulos <[e-mail]>]

On Wed, 2014-04-16 at 21:48 -0400, Patrick McManus wrote:

> I am very supportive of the TLS 1.3 charter because I earnestly
> believe it is largely the latency issues associated with current TLS
> standards that make it difficult for https to compete with http
> performance. The result is more plaintext and that sucks for users. I
> hope addressing that will be the focus of 1.3. There is some urgency
> in addressing that bottleneck - and that urgency isn't especially
> compatible with a ground up rework imo .

> Someone upthread hit the nail on the head for me when they said that
> what is being discussed isn't really a bake off of competing
> solutions, but really a CFP of competing ideas. There seems to be
> general agreement that incremental improvements and standardization of
> existing technologies are the areas the IETF is able to execute well
> in.

The issue is that we don't have any solutions, no other examples of
other IETF cryptographic protocols with smaller round-trip to copy from
(there is PKINIT Kerberos but that comes with issues), and
there have not been many ideas presented to reduce latency in TLS.
Thus asking the working group to produce a _new_ cryptographic protocol
in a short time frame, without any input from experts e.g., from
academia, on the topic, is not a good design practice (see WEP). 

Note that IETF typically takes an existing protocol that solves a
particular problem into action. In that case, judging from Eric's
presentation on the topic, it seems that this working group is asked to
invent a new one.

regards,
Nikos