In order to try and move in some kind of positive direction I've compiled
the following list of possible things to work on, as identified in emails
and in the WREC known problems document. Apologies if it's a bit rough
around the edges, but the idea is to try and get some discussion going
among us.
Having spoken to a few folks, Mark Nottingham and I are coming around to
the idea that splitting some of the CDNP work into a separate group (as
proposed) is probably a good thing. However, we still think there are some
things that WREC should be working on (the items identified for WREC below
are those Mark and I believe we can make progress on and which need to be
delivered in a reasonable timeframe).
I've tried to identify some of the issues I think CDNP might be interested
in looking at, though I'm probably ignoring some other obvious
things. (This is really an attempt to ask "can you take these on?" while
also working from Patrik's small list.)
There are also some other issues that are interesting and which a group may
wish to visit down the road, or else a group might possibly form to address
them. If there are other bullets that folks think need adding, please do
so. Similarly, if you think I've got something in the wrong place please
feel free to edit.
WREC
====
* Proxy/surrogate discovery protocol for user agents
- initial intent is probably for "user agents" to mean browsers;
- should also consider some aspects of wider proxy/surrogate discovery
or registration that will probably be interesting to CDNP and extproxy
(e.g. proxies/surrogates being able to discover other proxies or surrogates)
- obviously of interest to the folks doing the extensible proxy work
* Cache coherence/invalidation protocol
- including delta encoding and preloading within the administrative
control of the data owner
- will probably touch on expiry semantics in passing
- will be of interest to the CDNP folks too
(Yes, that's it for now.)
I'm not going to suggest milestones at this point, but we feel that aiming
to have these completed within a 12-14 month window shouldn't be too
unreasonable. (And we definitely want to try and keep things focused and
achievable.)
CDNP
====
* Distribution of data from origin servers outside of the administrative
control of the data owner
(I'd specifically exclude the scenario where the distribution occurs within
a single administrative domain... do we really need to standardize that?)
* Routing request to closest site (service) with the requested data
- should probably consider caching proxies in the model
- (obviously has some aspects of competitive advantage as was discussed
from the WREC KP draft in Pittsburgh)
- definition of the "closest" site (see WREC KP draft on known failings
of ICP in heterogeneous network)
- may overlap with, or provide us with, (or show us that it's
unnecessary to have) some replacement for ICP in the caching world
Billing (and more)
==================
There seems to be a case for billing/accounting to be covered in all three
of WREC, CDNP and extensible proxies. I believe that the extproxy folks
have commented that this may be appropriate work for another group (AAA?),
and it definitely seems that it's beyond any one of those being discussed here.
If we therefore open the scope to the wider issue of authentication and
authorization I guess we can add
* Exchange format for log summaries (essentially some form of accounting
records; this is the problem identified in the WREC known problems document)
* User session management
* Authentication models for multi-sited services/TLS interaction
(A pet peeve of mine is that there's an awfully large amount of non-private
information that has to get sent via https and could be cached. Can that
issue be addressed, or is it simply a browser implementation issue?)
Remainder
=========
The remaining items I have on the list are:
* Outstanding HTTP header issues
- "Vary header underspecified"
- registration of new HTTP headers
* Various end-to-end issues and concerns (includes content negotiation)
most likely something that needs to be considered for a world where
surrogates and extproxies live in the network; a few folks are interested
in this; may be a very short lived WG in its own right
* Some kind of documentation to cover those recurring items on "how to make
servers and content cache friendly"
not really an IETF WG matter; individual submission/W3C or elsewhere
* Re-visit the taxonomy
something that WREC and/or CDNP will need to revisit in the future...
perhaps the WREC and CDNP terminology sections need rolling into one?
- definition of a surrogate ;-)
This archive was generated by hypermail 2b29 : Thu Nov 18 2004 - 11:21:29 MST