Phoenix Kiula <phoenix.kiula@gmail.com> writes:
> My question: with that kind of volume and the underlying aggregation
> functions (by product id, dates, possibly IP addresses or at least
> countries of origin..) will PG ever be a good choice?
Well, only you're able to judge that for your own data and use cases.
Your query is sorting 10,000 records in half a second which is not great but
not terrible either. I think the only way you'll be able to speed that up is
by changing your index design so that Postgres can access the data you need
without sorting through all the irrelevant records.
I suspect others already suggested this, but you might look at partial
indexes. If your queries are very dynamic against relatively static data you
might look at building denormalized caches of the precalculated data.
--
Gregory Stark
EnterpriseDB http://www.enterprisedb.com
Ask me about EnterpriseDB's On-Demand Production Tuning
