On Tue, Aug 4, 2009 at 11:56 PM, Robert Haas<robertmhaas@gmail.com> wrote:
> Beats me. It looks like the first few queries are pulling stuff into
> cache, and then after that it settles down, but I'm not sure why it
> takes 5 repetitions to do that. Is the plan changing?
Yeah, we're just guessing without the explain analyze output.
But as long as we're guessing, perhaps it's doing a sequential scan on
one of the tables and each query is reading in new parts of the table
until the whole table is in cache. Is this a machine with lots of RAM
but a small setting for shared_buffers?
--
greg
http://mit.edu/~gsstark/resume.pdf