On 9 Aug 2000, Kai Henningsen wrote:
> It occurs to me that here would be a place for a dedicated cache. That is,
> Exim would forward the query to a caching process, which would do the
> actual queries if it couldn't satisfy them from the cache.
Benefit: a large cache.
Disadvantage: a single bottleneck for all Exim processes.
However: name servers do this kind of thing, so it can be efficient.
> Before starting to think about this more seriously, do people (preferrably
> people familiar with Exim's internals) think this might be useful, and
> what would people think is a reasonable interface to such a cache?
Maybe a new lookup type called "cache", whose data defines the query.
This could then be entirely general. I suppose if the data were exactly
as for the normal lookup, Exim could fall back to doing the query itself
in cases of cache failure.
--
Philip Hazel University of Cambridge Computing Service,
ph10@??? Cambridge, England. Phone: +44 1223 334714.
