Hi,
In a job design to achieve a particular transformation ,which approach will be better form performance tuning point of view ,using 3 TRansformer & 20 stage variable (source ,target and datavolume are same ).
Which job will take lesser time.Its not an interview question. I am trying to do performance tuning.
Hi,
Sorry I made a mistake in the post,
my question is: implementing a job logic using 3 Transformer back to back or 20 stage variable in single transformer (source ,target and datavolume are same ).
What else are the Transformer stages doing? Do you have inter-process row buffering enabled? When you watch this job in the Monitor, how much CPU is each Transformer stage using? Are they operating in one process or in separate processes?
IBM Software Services Group
Any contribution to this forum is my own opinion and does not necessarily reflect any position that IBM may hold.
Roopanwita wrote:Hey! Thanks for your reply!
How to see CPU usage for a particular stage..
In director, right click on the job and choose Monitor. Read through the columns that appear - if you don't find %CP right click on the columns and choose that. Alternatively, you can set the APT_PM_PLAYER_TIMING env variable to see the CPU usage for each operator.
It took me fifteen years to discover I had no talent for ETL, but I couldn't give it up because by that time I was too famous.
Would combine operators merge the three Transformers into one? I would guess that a single Transformer would work best but the easiest way to tell is to design the job both ways and run a test. The one time I would use two or more Transformers in a row is if the first Transformer has a constraint that removes a lot of rows to lighten the load on the subsequent Transformers.
