Re: UPDATE STATISTICS LOW
In article <5gbtol$pfp@cssun.mathcs.emory.edu>, Jeff Craig <settler@hobb
es.dbintellect.com> writes
>We are following the recommended UPDATE STATISTICS
>procedures as outlined in the SERVERS_7.2 release notes
>on our 7.20.UC2 database server that runs on a multi-cpu
>Silicon Graphics system, and the UPDATE STATISTICS LOW
>on columns of multi-column indexes runs a L-O-N-G time
>on a big table (227 million rows). It takes about 6 hours
>to run for a two-column index. Since our table is growing,
>and since we have more than one index on the table, the
>UPDATE STATISTICS LOW will eventually take twelve or more>hours to run, exceeding our batch window. Have others on
>the list experienced comparable performance? Does anyone
>have any recommendations on ways to reduce the run time?
>
>-------------------------------------------------------------------
>Jeff Craig | dbINTELLECT Technologies, Inc.
>Senior Consultant | Golden, CO
> | (303) 275-6954 ext. 8008
> | email: settler@dbintellect.com
> | fax: (303) 275-2134
>-------------------------------------------------------------------
have you tried running sar -d 3 3, sar -u 3 3 and vmstat 3 3 to see if
it is a disk,cpu or memory bottleneck?
Is the index fragmented across several disks? What does onstat -g ath
show whilst the update statistics is running.
--
David Williams