RE: Notes from WEBI WG Meeting

From: Lisa Dusseault (ldusseault@xythos.com)
Date: Wed Mar 27 2002 - 10:24:08 MST


I'd like to follow up on the list some points that concerned me in the
meeting. Thanks for the notes David which serve as a reminder...

> - We should explicitly allow notification to non-subscribing clients.

On the surface, this seems like a very bad idea to me, with no redeeming
benefits. Isn't sending notification to non-subscribing clients much like
spam? How does the non-subscribing client know what to do with the
notifications? How does the sender know what non-subscribing clients are
appropriate to send the notifications to -- how does it even know content
is being cached somewhere without a subscription? Don't the notifications
have to contain much more information if they are being sent to a client
without any "context" for the notification (since they haven't subscribed)?

I would even go so far, in the requirements document, as to say that the
protocol should never involve sending notifications to clients that have not
requested (subscribed to) them. I believe this is a simplifying
requirement, not a complicating one.

> - The protocol is designed to be used in the CDN and
> enterprise space with a
> hierarchy of intermediaries.
> - It should scale to a few hindered intermediaries fanning out at each
> level. The organization of the hierarchy is outside the scope of the
> protocol.

I would have assumed it would be a requirement of RUP that:
 - it be suitable for ad-hoc subscription to content expiration
notifications
 - it scale to 10,000s of intermediaries (this is in the requirements doc -
#4 in section 3)

If a resource update protocol is not suitable for ad-hoc subscription, and
the organization of the subscription hierarchy must be done by
administrators or otherwise outside the scope of the protocol, then I see
less of a need and a use for a standard in this area.

I believe however that this requirement is implicit in use case 5.4 in
http://search.ietf.org/internet-drafts/draft-ietf-webi-rup-reqs-03.txt.
Perhaps we need to make the language a little more explicit?

Lisa



This archive was generated by hypermail 2b29 : Thu Nov 18 2004 - 11:23:01 MST