Re: Kathleen Moriarty's Discuss on draft-ietf-httpbis-tunnel-protocol-04: (with DISCUSS and COMMENT)

[Martin Thomson <[e-mail]>]

Hi Kathleen,

I've responded separately to the secdir review.  There was a lot of
overlap between that and Stephen's review.

On 9 June 2015 at 13:59, Kathleen Moriarty
<[e-mail]> wrote:
> It seems to me that authentication relies on TLS.  Maybe stating this
> explicitly would address the concern?  Is there a reason this should be
> in the ALPN header(I'm not sure of that, just asking)?

We're not actually authenticating this stuff.  As I noted in my other
response, this is a promise that the client makes and one that the
proxy cannot enforce (because, TLS).  So the real uses for this header
field are: prioritization (move connections from slow and fat pipes to
fast and thin pipes, maybe), or early and cleaner denial.

The latter allows the proxy to quickly generate an HTTP status code
without having to do DPI or whatever other eldritch horrors they
currently are forced to do to recognize and deny things they don't
want.

The WebRTC case is interesting, because you can actually have some
assurance about the trustworthiness of the header field.  If you trust
the browsers, that is (though I'm not advocating that, browser people
are the most untrustworthy).