> However there's a lot more scope for improving a query along these
> lines, like adding indexes, or CLUSTERing on an index. It depends
> what other queries you are wanting to run.
> I don't know how much update/insert activity there will be on your
> database. However, if you were to add an index on the URL on both
> tables, then CLUSTER both tables on those indexes, and ANALYSE, then
> this query should run as a merge join, and be pretty quick.
> However, this is always going to be a long-running query, because it
> accesses at least one whole table scan of a large table.
There are already indexes on the url columns. I didn't cluster yet but
this is a pretty good idea, thanks. There will be no updates or
inserts. It's static data for research purposes.
In response to
pgsql-performance by date
|Next:||From: Marko Kreen||Date: 2008-06-30 13:16:26|
|Subject: Re: Federated Postgresql architecture ?|
|Previous:||From: Matthew Wakeling||Date: 2008-06-30 12:52:08|
|Subject: Re: Planner should use index on a LIKE 'foo%' query|