I am totally agree with Mark.
Ratio of page reads/page fetches for all queries has a little sense.
For example, the single badly designed query with many page fetches can
"improve" this parameter a lot, without any real improvement in the
performance, of course.
Regards,
Alexey Kovyazin
IBSurgeon www.ib-aid.com
On 04.05.2018 12:29, Mark Rotteveel [email protected]
[firebird-support] wrote:
On 3-5-2018 15:52, Javier Cintron [email protected] [firebird-support]
wrote:
>
>
> Thank you very much!
You shouldn't focus purely on this metric, greater page caches may lead
to more pages returned from memory, but also increase memory consumption
and overhead from things like cache eviction/coordination in the case of
Classic / SuperClassic (where the cache is per connection).
Mark
> On Thursday, May 3, 2018, 3:30:32 AM CDT, Dimitry Sibiryakov
> [email protected] [firebird-support] <[email protected]>
> wrote:
>
>
> 03.05.2018 7:56, Javier Cintron [email protected] [firebird-support]
> wrote:
> > Firebird ver 2.5.7.27050 64 bits
> > Windows Server 2012R2
> >
> > ¿Is there a way to calculate the effectiveness of incrementing the
> page cache buffer ?
>
> select cast(mon$page_reads as double precision)/mon$page_fetches from
> mon$io_stats
> where mon$stat_group=0
> If the value is getting lower - you are on the right way.
>
> --
> WBR, SD.
>
>
>
>
--
Mark Rotteveel
