This uses a simple UDP-based protocol to map URLs to file names.
Clients then retrieve the local copy from the shared filestore
by a mechanism such as nfs. If the document isn't in the cache,
the client retrieves it directly (using Socks to get thru firewalls)
and then loads it into the shared filestore (via nfs) and registers
it with the cache server. This approach places a minimal burden
on the cache server and so caters well for large sites. I will be
demonstrating this as part of my HTML+ browser at the WWW'94 conference.
-- Best wishes,Dave Raggett
----------------------------------------------------------------------------- Hewlett Packard Laboratories, +44 272 228046 Bristol, England dsr@hplb.hpl.hp.com