I'd like to add my vote for the web replication and caching working group
re-charter. The group has developed a taxonomy and list of issues and
research problems, as required when originally chartered. The important open
issues which seem relevant for future WREC work are:
* Browser location of and binding to an authorized, available proxy.
* Content distribution networks, surrogates, and content peering.
* Content consistency in autonomous proxy/replica servers.
The WPAD work addresses the first item; I'd like to see forward progress on
this. It is not a single-vendor solution that WREC has been asked to
rubber-stamp, I believe it is a step in the right direction. It helps to
alleviate the need for interception proxies by allowing browsers to locate
and bind to an intermediate. One issue is that the intermediate may not be
an authorized distributor for the content provider; that's where the
CDN/surrogate work comes in.
Content providers are delegating authority to serve their content to
distribution networks. There is a push for these networks to peer with each
other to distribute content. This requires cooperation at the protocol level
to ensure interoperability and clean semantics from content provider end to
end user. I think IETF is a good place to do this protocol architecture
work.
The third item relates to using invalidation to achieve content consistency
rather than validation. This can improve efficiency and response time. I do
not know of a better forum to discuss this than IETF. It could be outside
the WREC working group, but to me is well aligned with replication and
caching.
The discussion on WREC has been good to see. The working group was too quiet
for too long. I support re-chartering WREC to discuss these topics. I do not
think there is so much to discuss that we need to splinter into multiple
working group, nor do I feel the topics are too disparate.
I do not think WREC should attempt to discuss the following topics:
* Proxy implementation issues. Forward proxy server implementations are
mature. There are issues in the known problems draft that will hopefully be
worked on, but I do not believe WREC should devote further attention to them
at this time. Interception has been discussed enough IMHO.
* HTTP protocol level issues. There was an HTTP working group; that's the
right place for this work. The invalidation work, if it happens in WREC,
will need to work with an external group to evolve HTTP. If there is not
such a group, I think the work should be deprioritized in WREC. Protocol
compliance also does not belong in WREC.
Regards,
-- jad --
John Dilley
jad@akamai.com
This archive was generated by hypermail 2b29 : Thu Nov 18 2004 - 11:21:29 MST