I think that the soft state approach handles most of the problems,
but, what's the actual error scenario? If a roaming user stays in
touch with his forward proxy, he's not got any problem. If he gets
automagically switched to a new FP (how did that happen?) lots
of things break - there's no well-defined proxy switchover method.
That's a good problem for some group to work on ...
Hilarie
"Tomlinson, Gary" <gary(_dot_)tomlinson(_at_)cacheflow(_dot_)com> 08/13/01
11:49AM >>>
On Friday, August 10, 2001 @2:42 AMietf-openproxy(_at_)imc(_dot_)org -0800 Ian
Cooper wrote:
I was hoping to avoid raising the points ad nauseam on the list. I've
mentioned the issues here before, and Ted addressed them in another way
with respect to the overlay network aspects on Tuesday.
If I'm a roaming user, what happens to my end-to-end connection? (It can
be *very* different, especially with some of the use cases identified.)
If my OPES intermediary goes away what happens?
I agree with Ted's comments. OPES does need to articulate its impact upon
the end to end model in such cases. I see this articulation being
predicated
upon the error handling model chosen by the group. IMHO, what happens is
going to be dictated by the currently provisioned policy. The planned use
cases should help illuminate this for us.
Gary