Leonardo F wrote:
> Is it mainly a planning problem or an execution time problem?
I have here a database with a table partitioned across 400 sub-tables.
I'm using a trigger-based solution with constraint exclusion. The thing
that takes the longest is planning queries. I made THAT problem just go
away for the most part by using cached queries (only works within the
same database connection, but thats no problem for me).
I also tried a rule-based partitioning, that indeed breaks down quickly
performance wise.
Also, the trigger is a ON INSERT AND UPDATE, and only on the main table.
If your main insert/update process knows into which partition to insert,
you can speed that up even more; while still beeing able to use the
automated partitioning for everything else.
LG
Rene
