Re: [Lurk] New documented submitted

[Phillip Hallam-Baker <[e-mail]>]

On Fri, Mar 18, 2016 at 3:03 PM, Salz, Rich <[e-mail]> wrote:
> https://datatracker.ietf.org/doc/draft-erb-lurk-rsalg/
>
>
>
> A PFS-preserving protocol for LURK.  If the WG is created, we will ask the
> group to adopt it.
>
>
>
> There is an IPR disclosure that is in the system. Our intent is royalty-free
> for anyone who cross-licenses their patents to us.

"   The protocol should not become a generic signing oracle, even if it
   is suboptimal with regard to network bandwidth utilization.  This is
   done by not simply signing values, but by computing the full
   signature hash at the KeyOwner."

I started off with that as a requirement. Then I deleted it.  The
reasoning is as follows:

First, what does the limitation get us? Is a TLS signature key to be
shared with any other purpose? I really hope not. That would be very
bad indeed. So what does this restriction achieve?

There is a measurable cost. It makes the protocol specific to TLS and
specific TLS key agreements to boot. It also breaks with existing
APIs. I won't be able to use your approach with IIS unless and until
they update the TLS implementation. If I use another Web server I have
to wait for implementation in OpenSSL or whatever or have to deal with
custom patches.

Many existing servers already have a socket for plugging in custom
crypto to support accelerator boards. I agree we should not limit
ourselves to that API but we definitely need to support it.