Performance issue running jobs in parallel

[somu_june]

Hi,

I have a three server jobs , the three jobs are reading from a file and witting to a hashfile and there is a transformer stage between input file and hash file in all the three jobs.

when I ran the three jobs in sequential order in job sequence each job is completed in 5 minutes and total time taken by a sequence job is 16 minutes to complete all the three jobs.

I designed a job sequence to run three jobs in parallel, after runing the three jobs in parallel, each job is taking 11 to 12 minutes and the sequence is taking 13 minutes to complete.

The jobs are running slowly when I run the jobs in parallel.
what I noticed is CPU elapsed time increased more than 2 times when the job ran in parallel with other jobs.

When ran in Sequential

587375 rows read from ext
587375 rows written to out
27.240 CPU seconds used, 287.000 seconds elapsed.


When ran in parallel with other jobs

587375 rows read from ext
587375 rows written to out
27.600 CPU seconds used, 664.000 seconds elapsed.

Below are the options selected in Hash file stage
Type 30 Dynamic
Selected option Allow stage write cache.
Clear file before writting

My question is do I need to increase the project level write catche size from default 128 when I run the jobs in parallel. How to achive same job run time as sequential job runtime


Thanks,
somarjau