Re: db buffer cache

Jonathan-

This is correct. It is a return of a current mode buffer rather than a clone buffer. Significant in that the call that bumps this statistic never clones a buffer. The work saving can be quite significant, when you think about all the work that can go into cloning a buffer and pinning associated undo blocks, reading trans table to find the undo, potential undo of the trans table, corresponding additional latch gets, etc.

• ricky

Jonathan Lewis wrote:
>
> Sorry about the delay in replying to this one.
>
> I haven't yet seen anything documenting this statistic,
> but when I was down in Australia with Steve Adams
> we kicked the idea around for a bit and came to a
> tentative conclusion that it was a consistent get where
> the CURRENT block was known to be the required block.
>
> (Which means that slightly less work is done - even
> though the relevant cache buffers chains latch is still
> hit).
>
> --
> Regards
>
> Jonathan Lewis
> http://www.jlcomp.demon.co.uk
>
> Next Seminars
> UK Sept, Nov
> USA x 2 November
>
> http://www.jlcomp.demon.co.uk/seminar.html
>
> srivenu wrote in message <1a68177.0208092309.30252050_at_posting.google.com>...
> >Thanks for the reply Jonathan
> >What are consistent_gets examinations ?
> >What is examined ?
> >regards
> >Srivenu
Received on Fri Aug 30 2002 - 08:15:10 CDT