On 28 July 2010 02:58, Howard Rogers <hjr@diznix.com> wrote:
> For what it's worth, I wrote up the performance comparison here:
> http://diznix.com/dizwell/archives/153
>
Thanks, very interesting results. I wonder, are the results being
sorted by the database? The performance degradation for large numbers
of results might be explained by it switching over from an internal to
an external sort, in which case tweaking work_mem might make a
difference.
Of course this is pure speculation without the EXPLAIN ANALYSE output.
Regards,
Dean