On 01/29/2014 11:54 AM, Robert Haas wrote:
> I agree. I find it somewhat unlikely that pg_stat_statements is
> fragile enough that these few extra counters are going to make much of
> a difference. At the same time, I find min and max a dubious value
> proposition. It seems highly likely to me that stddev will pay its
> way, but I'm much less certain about the others.
What I really want is percentiles, but I'm pretty sure we already shot
that down. ;-)
I could use min/max -- think of performance test runs. However, I agree
that they're less valuable than stddev.
--
Josh Berkus
PostgreSQL Experts Inc.
http://pgexperts.com