Need help to interpret CPU usage

dataservices performance
1

Hi,

I need help to interpret CPU usage when running a job…

I selected the option “Run as a separate process” in the Match Transform, when execute the job I can see 2 PID created for the data flow, however from the Resource Monitor I can only see 1 of the process has CPU usage, the second one has no CPU activity at all.

Does that mean the 2nd process is waiting for something? if so how would the “Run as a separate process” helps the performance?

My second query is which system is controlling the usage of CPU - the OS or the al_jobserver? when I run the job first time, it uses 50% of server CPU, then we added and double the CPU on server, when run the same job again, its CPU usage drops to 25% … Also there’s nothing is running on the server.

Is there a setting on DS that we can use to adjust the CPU usage?

Our environment is DS 4.1 (14.1.1.338) on windows 2008

Thanks

chauh :australia: (BOB member since 2010-12-23)

2

The run as a separate process doesn’t sound like it is helping you any. I rarely ever use that feature because it simply hasn’t been necessary. If you have a particular step in a Dataflow that uses gobs of memory then maybe you should consider writing it in a different manner.

My understanding of the run as a separate process is that it was put in place for situations where the step in the Dataflow needed to use lots of memory. It’s not there for CPU optimization.

I vaguely remember a discussion where either run as a separate process or the DOP setting wasn’t used in the Match transform. I think it will be single threaded always due to the nature of having to match rows to other rows.

eganjp :us: (BOB member since 2007-09-12)