On Sat, 26 Mar 2005, Tom Lane wrote:
> Oleg Bartunov <oleg@sai.msu.su> writes:
>> I'm looking for blocks *actually* read from disk, since IO is the most
>> important factor.
>
> Well, you won't find that out from Postgres, since it has no idea
> whether a read() request was satisfied from kernel disk cache or had
> to actually go to disk.
so, what's the usefulness of pg_stat* ?
From my experiments I feel heap_blks_read is the table blocks read from disk,
well, sort of, and heap_blks_hit - from buffer. At least, I observed when I
repeat the same query several times (heap_blks_read+heap_blks_hit) doesn't
changed while more blocks come from heap_blks_hit.
>
> You could turn on log_statement_stats to get some possibly-more-reliable
> numbers from the kernel via getrusage(), but this will only tell you
> about totals across the whole backend process, not per-relation ...
>
Hmm, it's impossible to do researching :( We have so many parameters and
almost no reliable stats. How do you believe you did a good choice ?
It's common in db worlds to have IO statistics (timings are not important)
to research algorithms and implementation.
> regards, tom lane
>
Regards, Oleg
_____________________________________________________________
Oleg Bartunov, sci.researcher, hostmaster of AstroNet,
Sternberg Astronomical Institute, Moscow University (Russia)
Internet: oleg@sai.msu.su, http://www.sai.msu.su/~megera/
phone: +007(095)939-16-83, +007(095)939-23-83
