Note in-line
--
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The Co-operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Optimising Oracle Seminar dates:
http://www.jlcomp.demon.co.uk/seminar.html
"Igor Racic" <igor.racicC_at_freesurf.fr_no_C> wrote in message
news:40980203$0$13079$636a15ce_at_news.free.fr...
>
> Hi, Jonathan
>
> Many times in your book, posts on newsgroups and articles
> on the web I ask myself "How he found that thing ?" (I hope I'm note the
> only one :-) )
>
Most of the consultancy work I get has to deal with
odd behaviour that needs a solution rather than a
quick fix. This means I get to see more anomalies than
most, and often have to investigate why they are appearing.
Then in my spare time I just dig around and experiment a lot.
> Just curious here, what would be the reason that you will trust more
> this statistic over another one ?
>
History and consistency.
consistent reads and db block gets have always
shown odd anomalies when observed closely.
v$buffer_pool_statistics appeared and contradicted
v$sysstat quite dramatically.
v$latch (cache buffers chains) frequently indicated
that the figures in v$sysstat were suspect.
sum(x$kcbsw) always seemed to be very close to the
value for 'session logical reads'.
Of course - this begs the question of what a session logical
read is; or perhaps defines it as a type of buffer access that
is recorded in x$kcbsw.
Received on Tue May 04 2004 - 19:10:12 CDT
