Relationship to WREC and extensible proxies

["Mark Day" <markday@cisco.com>]

People who are subscribed to the WREC mailing list (for the IETF working
group on Web Replication and Caching) know that there has recently been a
discussion there about the appropriate division of labor between three
identifiable groups  who are interested in apparently-similar problems
within the IETF. Those three groups are: WREC itself, this (CDN peering)
group, and the group working on extensible proxies (see www.extproxy.org).

On Friday's design team conference call it was pointed out that many people
on this list are likely to be unaware of that discussion, and would probably
appreciate a summary.

As I understand it, WREC was chartered primarily to produce a taxonomy of
the problem area, although many people expected that would be only a first
step before "real work". If WREC is to do more, it must be rechartered. Mark
Nottingham recently proposed a list of items that could be taken on by a
rechartered WREC group:

>* content peering
>* enhanced coherence mechanisms (invalidation)
>* log summary formats
>* surrogate role clarification
>* semantic transparency issues in intermediates
>* coordination with content negotiation, other groups which affect
>   intermediates
>* proxy discovery (very important, in light of interception proxies)

There is a sense that content peering is squarely within the charter of WREC
and thus that it seems odd to propose a BOF on the subject.  Others feel
that narrowly-focused groups would be more effective than a single working
group that tries to take on all aspects of web replication and caching (for
example, there is no longer a single "email" or "directories" working group
within the IETF).  My opinion has been that I don't mind what organizational
structure content peering goes into, as long as it can be pursued
effectively there.

Patrik Fältström, one of the two Area Directors for the Applications area,
has asked the people involved in preparing the various BOF/rechartering
proposals:

> I would like to see a one-page paper (not much more is needed, or?)
> which talks about what different components exist in a content
> network.
>
> A list such as (but it should fit together):
>
>   - Distribution of data from origin server to copies within
> administrative
>     control of the owner of the data
>   - Distribution of data from origin server to copies outside of <above>
>   - Preloading / load on demand of caches
>   - Cache discovery
>   - Routing of client to closest site with data
>   - Switching of client to good server within site
>   - Billing / charging / statistics collection
>   - HTTP / HTML extensions for cache control
>   - Continuation of work on taxonomy
> :
> :
> :
>
> You are the one doing the list, and you then tell what on that list
> should be handled by each of the proposed wg's. I am definitly not
> saying that you have to create wg's for everything (I rather see not
> :-), but what pieces you are going to work on, and what have to wait.
>
> And, I want a list which you all in the "wrec" area agree on. Not one
> explanation from each one of you (which I have got so far :-). I.e.
> the stories have to fit together.

At this point, I think that Ian Cooper (for WREC), Hilarie Orman (for
extensible proxies), and I (for CDN peering) will work together to address
Patrik's request.  I expect that we will develop a joint proposal about the
"big picture" and the relationship among WGs which we will then share with
the various mailing lists.

And to recap quickly, you can find out more about extensible proxies at
www.extproxy.org and about WREC at
http://www.ietf.org/html.charters/wrec-charter.html.

--Mark

Mark Stuart Day
Senior Scientist
Cisco Systems
+1 (781) 663-8310
markday@cisco.com