> Is there any way of reading the whole database into memory? By changing
> the
> number of shared buffers the first query still takes long time and the
> subsequent similar queries are fast.
> If there is way of pinning the database in the memory all similar
> queries will take the same time. My database is around 200MB and could
> easily be pinned into RAM.
If you have enough RAM, your OS will cache the physical files as they
are used. I'm not sure how much of a speedup would be gained by PG
keeping the data in RAM and doing the caching itself, but my guess is that
it wouldn't be that great of a speedup over simply having the data in the
disk cache.
steve
