Re: RBS waits

ashok_jayaraman_at_yahoo.com (Jayaraman Ashok) wrote in message news:<7ca2852.0208070503.28e88693_at_posting.google.com>...
> Hi
> Te following is the status of my database. It has been running for 3
> days now. Though i have 55 rollback segments , i always seem to have
> contention for it. transactions_per_rollback_segment = 10, and the
> rollback segments have initial 5m next 5m minextents 20. There are
> about 100 users connected at any point of time and around 30 - 40
> concurrent transactions. Where is the problem?
>
> SQL> select * from v$waitstat;
>
> CLASS COUNT TIME
> ------------------ ---------- ----------
> data block 18024355 24968386
> sort block 0 0
> save undo block 0 0
> segment header 93 665
> save undo header 0 0
> free list 0 0
> extent map 0 0
> bitmap block 0 0
> bitmap index block 0 0
> unused 0 0
> system undo header 0 0
>
> CLASS COUNT TIME
> ------------------ ---------- ----------
> system undo block 0 0
> undo header 1827 121484
> undo block 13504 43563
>
> Please help.
>
> Ashok

A non-zero wait count in v$waitstat for undo header and blocks is only a problem if the numbers are high in relation to the total IO you perform in the same measuring period. I believe that the Performance and Tuning manual actually has some comments to the effect that waits should be < 1% of gets (consistent gets + db_blocks in prior releases) plus waits to gets in v$rollstat should also be under 1%.

You might want to check what the manual for your release says.

HTH -- Mark D Powell -- Received on Wed Aug 07 2002 - 15:45:37 CDT