Alex_Pastuhov
Legendary Poster
Re: Accumalation of outstanding requests on callobjects processes of enterprise servers
Yes, the plan should look good then, despite the delay.
Although, Oracle could be coming up with a different plan if it sees unusual contention for somethiing it cares about and that includes disks. Not very likely, though. And it could be as you suggested to do with the caches as well. So a different plan would not necessarily be conclusive.
You really need to get perfmon data for all disks and cross-match it against the times you know there were such problems...
Yes, the plan should look good then, despite the delay.
Although, Oracle could be coming up with a different plan if it sees unusual contention for somethiing it cares about and that includes disks. Not very likely, though. And it could be as you suggested to do with the caches as well. So a different plan would not necessarily be conclusive.
You really need to get perfmon data for all disks and cross-match it against the times you know there were such problems...