"Cyril VELTER" <cyril.velter@metadys.com> writes:
> I'm converting an application to use the V3 protocol features in the 7.4
> libpq. As I need to make a design choice regarding the use of prepared
> statements, I'm wondering what ressources does a prepared statement use on
> the server ? If I need to create several hundred in each backend, is there a
> big memory overhead ?
It'd depend on the complexity of the query plans, but I'd think order of a
few KB per query.
> What's the time spent by the backend to find one
> prepared statement into a list of several hundreds of them ?
The prepared queries are indexed by a hash table, so the lookup time
should be fairly constant independent of their number.
regards, tom lane