Performance improvement for server job

[somu_june]

Hi

I have a server job with two active stages transformers and the first transformer stage had a complex logic like calling routines and 80 stage variables where in second transfromer is having less logic it had 4 stage variables and writing to a sequential file. The first transformer has 5 outputs and writing to a 4 sequential files and second transformer stage. The job is running with 750,000 records and ending in 06:50 minutes and the %CPU for first transformer it is showing 87% and for second transformer it is showing only 4% . I tried changing the inter process row buffer from 128 to 1024 KB and the job runtime decreased by only couple of seconds. The first transformer while processing it is processing only couple of 100 rows/second in couple of links. Is there any parameters that I need to fine tune the job to increase the performance of server job. Please let me know.


Thanks,
Soma

[asorrell]

Increasing the buffering isn't going to help. Your main transformer is taking a long time to process and is slowing down your "pipe" of data. Making the pipe bigger isn't going to help.

Speed up the first transformer by reducing the number of stage variables and routines called if possible.

[arunkumarmm]

What is your source? Where does your transformer gets the data from? Are you using look-up?