Heikki Linnakangas wrote:
> However, it caught me by total surprise that the performance with 1
> buffer is so horrible. Using 2 buffers is enough to avoid whatever the
> issue is with just 1 buffer. I have no idea what's causing that. There
> must be some interaction that I don't understand.
Ok, I found the reason for that. I was using this query for the selects:
SELECT COUNT(*) FROM (SELECT 1 FROM stock_copytest LIMIT 10000000) AS a;
Stock_copytest is larger than RAM size, that's why I used the LIMIT to
make the result set memory resident. That had the side effect that
apparently the limit-node kept the single buffer pinned which defeated
the buffer ring completely. To avoid issues like that we apparently want
to use 2-4 buffers instead of just 1.
I'll review my test methodology and keep testing...
-- Heikki Linnakangas EnterpriseDB http://www.enterprisedb.com
