Josh Berkus wrote:
> 1) Query caching is not a single problem, but rather several different
> problems requiring several different solutions.
>
> 2) Of these several different solutions, any particular query result caching
> implementation (but particularly MySQL's) is rather limited in its
> applicability, partly due to the tradeoffs required. Per your explanation,
> Oracle has improved this by offering a number of configurable options.
>
> 3) Certain other caching problems would be solved in part by the ability to
> construct "in-memory" tables which would be non-durable and protected from
> cache-flushing. This is what I'm interested in chatting about.
Just my 2 cents on this whole issue. I would lean towards having result
caching in pgpool versus the main backend. I want every ounce of memory
on a database server devoted to the database. Caching results would
double the effect of cache flushing ... ie, now both the results and the
pages used to build the results are in memory pushing out other stuff to
disk that may be just as important.
If it was in pgpool or something similar, I could devote a separate
machine just for caching results leaving the db server untouched.