Here's the second version of the I/O timings patch. Changes from the
previous version:
* Rebased against master.
* Added the missing pg_stat_statements upgrade script that I
accidentally left out from the previous version.
* Added a tool to test timing overhead under contrib/pg_test_timing
I hope that having a tool to measure the overhead and check the sanity
of clock sources is enough to answer the worries about the potential
performance hit. We could also check that the clock source is fast
enough on start-up/when the guc is changed, but that seems a bit too
much and leaves open the question about what is fast enough.
About issues with stats file bloat - if it really is a blocker, I can
easily rip out the per-table or even per-database stats fields. The
patch is plenty useful without them. It seemed like a useful tool for
overworked DBAs with limited amount of SSD space available to easily
figure out which tables and indexes would benefit most from fast
storage.
--
Ants Aasma