> I agree that conflict avoidance is essential. Perhaps I was a little
> over-zealous when I wrote the above statement - the terminology is
> confusing I think - but one of the things Bill's notes of the telephone
> conference didn't expand on was the discussion we had about
> "infrastructure". We did make a change to the third of the BCP-type I-Ds
> proposed so that it was more infrastructure-oriented but I now see that
> perhaps we need to ammend the charter description to match this too,
> specifically that it is about replication and caching in general rather
> than just caching. i.e. we are not just looking at traditional proxy-caches
> but all of the so-called "intermediaries". Would that satisfy your concerns
> any better?
It would help. But basically I want to see an architecture which
supports all of:
direct client access to supplier-provided replicas
(on referral to those replicas via a URI resolution mechanism)
client access via intermediary to supplier-provided replicas
client access via intermediary to cached replicas
(and probably also supplier ability to "shoot down" cached but
obsolete replicas.)
I don't agree with the assumption that an intermediary should be
required. Intermediaries significantly decrease correctness,
transparency, and reliability, and add a huge amount of complexity,
for what may well turn out to be, in the long term, minimal gain.
I want an architecture that can do without intermediaries, or
minimize their use, when it makes sense to do so.
Keith
This archive was generated by hypermail 2b29 : Thu Nov 18 2004 - 11:21:25 MST