On Mon, Nov 14, 2011 at 5:26 PM, Leyne, Sean <s...@broadviewsoftware.com>wrote:
> > > In your example we see 7149 physical reads and 214192 times this
>
> The ideal statistic that would help in knowing whether the cache needs to
> be increased would be "disk re-reads", but that would require that the
> engine track a list of the RDB$Key of every row accessed in a query to make
> the distinction between "read" (initial load) and "re-read" (re-load into
> cache due to LRU). That list would likely require more overhead than
> really its worth.
>
Tracking page numbers would be more effective, but even then the statistics
aren't completely clear. For example, Classic has to reread a page
everytime it is referenced after a change by another process, so the header
page and active TIP are read frequently even if there are unused pages in
the cache.
Cheers,
Ann
------------------------------------------------------------------------------
RSA(R) Conference 2012
Save $700 by Nov 18
Register now
http://p.sf.net/sfu/rsa-sfdev2dev1
Firebird-Devel mailing list, web interface at
https://lists.sourceforge.net/lists/listinfo/firebird-devel
