2010/3/28 Thom Brown <thombrown@gmail.com>:
> The problem here is that you appear to require an index update, trigger
> firing and constraint check for every single row. First thing I'd suggest
> is remove the indexes. Apply that after your import, otherwise it'll have
> to update the index for every single entry.
+1
> And the trigger won't help
> either. Import into a single table and split it out into further tables
> after if required.
note: partitioning could help if there were multiple physical volumes
/ spindles for data directory.
for maximizing performance, I would rather split the CSV input (with
awk/perl/whatever) before loading, to have one backend for each
partition loader.
> And finally the constraint should probably be applied
> after too, so cull any violating rows after importing.
+1
--
Filip Rembiałkowski
JID,mailto:filip.rembialkowski@gmail.com
http://filip.rembialkowski.net/
