3000 Rows - 7 seconds - very slow
On my PC (athlon 64 3000+ running Linux), selecting 3000 rows with 4
columns out of a 29 column table takes about 105 ms, including time to
transfer the results and convert them to native Python objects. It takes
about 85 ms on a test table with only those 4 columns.
There is definitely a problem somewhere on your system.
I'd suggest running this query in an infinite loop. Logically, it should
use 100% processor, with postgres using some percentage (30% here) and
your client using some other percentage (70% here). Is your processor used
to the max ?
