cache-only backend

[matthew sporleder <msporleder@gmail.com>]

I was browsing through the backends for slapd and wondering if there
was anything like this:
Deploy a replica with no existing database,
Any searches not found in the local repository are searched from a set
of masters,
If found: cache response, return result,
Else if not found: return failure.

I can see this being beneficial as deploying (or fixing) replicas is
difficult with a large database, and could actually be very good at
keeping replicas up-to-date without a separate replication process
(depending on cache expire times, etc).  I can also see this holding
some big disadvantages for the master servers having to provide tons
of information to populate these replicas.

-- 
_Matt