Doesn't appear so, the only reference on the man page to anything like
service time is 'milliseconds per average seek', qualifying it with
'not at disks report seek times'. We use another server performance
monitoring tool here, Perceive (?), which shows that that 3 Oracle
related disks appear to be 85% busy during the batch load. The disk
you've noticed being hardest hit, shown as re17, contains the index and
rollback tablespaces. On the whole it's looking like there's no 'quick
win' and the only real solution is to go back to the batch process and
possibly redesign it.
Many thanks,
Rob
Received on Mon Mar 20 2006 - 10:18:23 CST
