I've been thinking about the mail I sent yesterday as well as some
discussions within the architecture team. I've also been thinking about
Josh's proposed requirement that we specify enough detail about
technical trust establishment to get better interoperability than SAML.

We've been talking about the importance of proxy behavior throughout the
process.  Some of that will be local. For example how a proxy near the
RP knows that a particular machine is allowed to claim a hostname is a
local matter.

However, there are significant elements that have real protocol impacts
if we're going to have interoperability.

first, all of this proxy behavior is inherently optional: today's
proxies don't do it.  So, as I discussed yesterday we need mechanisms
for knowing what proxies have done and what they have not done.

Second, when things are decomposed like the host check living in an
organization and the realm check living closer to an IDP, then we need
to explain how those checks fit together to meet our security
guarantees.  Also, it is quite obvious that intermediates have important
roles to play. The value that federations bring to the ecosystem is
managing and negotiatingpolicies and agreements.  Some of that feeds
into protocol requirements for exchanging what policies are in play, and
for allowing the federation to filter (or provide filters) on the
behavior of actors.

Also, as Josh and I hope to explain in Prague, we believe that a new
technical trust mechanism is required for some common ABFAB deployments.

All this together suggests to me that we have a lot to think about in
terms of the AAA fabric that makes these federations possible. There's a
lot of discussion starting to filter into the architecture document.
However, I'm now convinced that it goes beyond that.
Protocol elements are required. 

I'm not ready with specific proposals at the moment.  What I do think is
important is creating some high-bandwidth discussions of the issue to
get more than just Josh and I into the right mental space.

This is a heads up that I'd like to have these sorts of discussions and
a call for interest.

--Sam
_______________________________________________
abfab mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/abfab

Reply via email to