Dave Edmondson <david(at)jlc(dot)net> writes:
> yes. I ran VACUUM ANALYZE after creating the indicies. (Actually, I VACUUM
> the database twice a day.) The data table literally has 145972 rows, and
> 145971 will match conf_id 4...
Hm. In that case the seqscan on data looks pretty reasonable ... not
sure if you can improve on this much, except by restructuring the tables.
How many rows does the query actually produce, anyway? It might be that
most of the time is going into sorting and delivering the result rows.
regards, tom lane