Web Replication and Caching, second BOF
Chair: John Martin
Reported by Ingrid Melve
This was the second meeting which was intended to determine the usefulness of chartering a working group on web caching and replication. The meeting started with a brief overview of the charter followed by a more in-depth coverage of the areas of interest which had been identified on the mailing list, specifically, client to proxy-cache communication, draft-melve-clientcache-com-00.txt inter-cache communication, draft-melve-intercache-comproto-00.txt cache to network-device communication and requirements for a caching / replication infrastructure.
There was consensus that an initial emphasis should be placed on documenting what was currently available, though not necessarily labelling this as "Best Current Practice". Warning about doing both documentation of current practice and protocol development at the same time, there is a need for understanding the field before developing new protocols (if you do not know what you are talking about, you should not try to implement it)
There are several proprietary protocols in use, among these are WCCP which is used for network element to proxy communication and WPAD for client configuration. Public documentation as (at least) informational RFCs would be helpful before new work is done. The proponents of these protocols agreed to document these and push them through the IETF process in parallel with the initial WREC work.
Web replication is not yet widely deployed, whereas web caching proxies are, particularly outside the US. Comments were made about documenting how the current caching proxy infrastructure relates to overall design principles of the Internet, and how a caching and replication infrastructure is currently built.
The consensus in the meeting was that a working group in this area should (first) define the design goals or research issues before embarking on specific protocol development. In addition, the initial work should include a document which provided both a taxonomy of existing caching / replication developments and terminologies which would be used as a reference for later work an that the current drafts (draft-melve-*) could be merged to form the basis for this. Any existing protocols would in parallel be submitted by their various proponents as informational RFCs after an editing review by the working group. It was further agreed that the defined goals and milestones would need to be re-worked and re-ordered and that the time line should be shortened to end at the Summer 1999 IETF meeting. After this time, the charter would be re-visited to look at providing a framework for future caching / replication development within the IETF.