Kevin Grittner wrote:
> Assuming this is real, and not just "in the noise" -- what conclusions would you
> draw from this?
Was trying to demonstrate the general ability of pg_stat_bgwriter
snapshots at points in time to directly measure the buffer activity
Hannu was theorizing about, not necessarily show a useful benchmark of
any sort with that. Watching pgbench create a database isn't all that
interesting unless you either a) increase the database scale such that
at least one timed checkpoint kicks in, or b) turn on archive_mode so
the whole WAL COPY optimization is defeated. More on this topic later,
just happened to have that little example script ready to demonstrate
the measurement concept.
--
Greg Smith 2ndQuadrant US Baltimore, MD
PostgreSQL Training, Services and Support
greg@2ndQuadrant.com www.2ndQuadrant.us