Showing results for 
Show  only  | Search instead for 
Did you mean: 

Slow Performance Warehouse Write Recommendation


I am facing the following issue. Checking the Deployment Health dashboards for Performance Warehouse Write, I am noticing that the performance warehouse is suffering from slow write times late at night and in the early hours of the morning.

I realize that this may be due to nightly tasks that are executed. Besides checking for the characteristics of the disk, what other recommendations can be provided to speed up writing measurements.

The client is running performance warehouse on Oracle 12g, on a partitioned tables. Any suggestions are welcome.


Hi Jose

I have come across a very similar situation and the issue was that the database was shared with many other systems and the box couldn't cope with the clean up job. It was taking up to 13 hours to run.

I suggested moving the DB on dedicated hardware.

The other thing to check is that there is sufficient resources on the box. For an XL we are talking about 4-6 cores just for the warehouse DB.

Consider, re-indexing the tables even if in your case, Oracle should take care of it rather well anyway.




Hey Jose,

I would also recommend to check if they have another
process running during the same time frame as our clean up task. I find that many clients
run their database backups at around the same time as the Performance Warehouse clean up task. That situation can very likely force table locking (or other types of locks) and
cause our task to go much much slower.

Good luck!



Hey Jose,

We had a same situation where the scheduled task was taking more than 2 hours and the data gap was started for that time, even a year before the same task was not taking more than 5-10 minutes but with the passage of time the database size was increasing which was not according to our initial sizing (large # of agents + BT etc...) then we decided to partition the database suggested by Dynatrace for the XL or above environment and now from that time again the scheduled clean up task is not taking more then 10 minutes.

We also cannot ignore the @Miguel M. point.