Re: run time stats and x$ tables

Hi Ed,

These statistics are cummulative, so Tuesday's figures will include Monday's. Most of my APT scripts like this one just look at the cummulative statistics for the life of the instance, because as a consultant, that's often all I've got to work with.

However, there is nothing stopping you from using the same technique that the Oracle utlbstat.sql and utlestat.sql scripts do to compute the difference in the statistics over an interval.

Regards,
Steve Adams

http://www.ixora.com.au/

http://www.oreilly.com/catalog/orinternals/

http://www.christianity.com.au/

On Mon, 01 Nov 1999 14:23:52 GMT, Ed Stevens <Ed.Stevens_at_nmm.nissan-usa.com> wrote:

>I have begun tracking performance statistics on my Oracle databases.
>Right now I am focusing on cache miss ratio, using the following query:
>
>select
> p.bp_name buffer_pool,
> lpad(to_char(100 * sum(s.pread) / sum(s.dbbget + s.conget), '990.00')
>|| '%', 13) miss_rate
>from
> sys.x_$kcbwds s,
> sys.x_$kcbwbpd p
>where
> s.set_id >= p.bp_lo_sid and
> s.set_id <= p.bp_hi_sid and
> p.bp_size != 0
>group by
> p.bp_name
>/
>
>(X_$kcbwds and x_$kcbwbpd are views that map directly to the tables
>X$kcbwds and x$kcbwbpd).
>
>My basic question is – when are numbers in these ‘statistics gathering’
>tables updated, and when are they zeroed? At its roots, the above
>query is selecting PREAD, DBBGET, and CONGET from x$kcbwds, and
>bp_name from x$kcbwbpd. But I don’t really know for what time frame
>these numbers are derived. If I run this query every day for a week,
>and the database is never shut down, do Tuesday’s numbers also include
>Monday’s numbers?
>
>- Ed Stevens
Received on Mon Nov 01 1999 - 14:10:09 CST