Job Performance

[pklcnu]

Dear Experts

I have a job designed where it reads the data from Oracle table and had few look ups in the job after that the job writes the out put to a dataset.

There are few million records that needs to be processed, during first few minutes after the job starts it shows like it is processing around 15,000 records/sec and after some time the number of records processed goes on reducing and by the time job completes the number of records processed is around 4000 rows/sec. And the whole job takes around 45 minutes to complete.

I was wondering why in the beginning it is 15,000 rows/sec and at the end only 4000 rows/sec. What should I do to keep the number of rows processed is high similar to the start of the job ?

Thanks

[ray.wurlod]

The same reason that file download rates vary when using Windows - the average is taken over the entire period so any delay will increase the time while not increasing the count as fast, so the average seems to decrease.

[chulett]

Exactly, it's an average.