cancel
Showing results for 
Search instead for 
Did you mean: 

High % DB Time

Former Member
0 Kudos

Hi folks,

I'm seeing an increase in Dialog response times on a clients ECC system (Windows/Oracle)

Average % DB Time over a month was 60.57% (< 40% recommended)

DB Buffer quailty is consistantly between 98-99%

DD Cache Quality is 99%

No issues with with recursive calls vs user calls.

Hit ratios all good in ST02 (98%-100%)

And yet, still we see % DB time over 60%

I've identified a number of DB intensive jobs/reports but they're not regular and the high % DB Time is fairly consistant throughout the month.

Any suggestions would be greatly appreciated.

Thanks.

Accepted Solutions (0)

Answers (3)

Answers (3)

Former Member
0 Kudos

Hi,

What is the resource size that you have (CPU/Memory)? How they are allocated for each instance (DB, CI etc.)?

What is the Oracle version and the Patch set that you are using?

Regards,

Nick Loy

thiago_cavalheiro
Active Participant
0 Kudos

Hi Paul,

At first I would check the following three points:

- Database parameters and patches: bad parameter settings will cause your database's optimizer to choose bad execution plan, hence your application will need to wait more than it actually need. Use the script from note 1171650 to check that. Make sure the parameters are properly set and that you have a recent SBP.

- Hardware bottleneck: check for I/O and CPU mainly. Note 793113 might help with the IO.

- Statistics. Check notes 838725 and 588668.

Regards,

Thiago

Former Member
0 Kudos

Hi, thank you for the reply.

On the back of your suggest I have identified approximately 30 DB parameters that are not set to SAPs recommendation.  Unfortunately it could be up to a week before I get the "OK" to change them.

In the meantime, does anyone have anymore thoughts?

I've looked at so many stats in ST03n and everything is within acceptable times (processing time, CPU time, GUI time).  However, % DB Time during core business hours remains at an average of between 65%-75%.

I've tried to identify which transactions maybe causing this, but it's inconsistent from one week to the next.  No patterns.

Any further help would be appreciated.

thiago_cavalheiro
Active Participant
0 Kudos

Hi Paul,

The parameters are very critical for performance, and most of the general performance issues are caused by them.Since you are saying you cannot find any pattern then it only reinforces that. I recommend you to try an emergency downtime to apply them.

Other than that I can only recommend you to take a look at the hardware response times and check whether there is any room for tuning at this level, which I believe you are already doing.

Regards,

Thiago

former_member182657
Active Contributor
0 Kudos

Hi Paul,

Please share your hardware details and also suggest you to use the workload analysis in ST03N to pinpoint the transactions/programs taking too long at a particular time.

Regards,

Gaurav