DSXchange

Hi,

After each job completes (sequence, parallel) I would like to extract the time each stage took, so to identify the resource hogs. Right now i have to manually open up the job logs and jot the time down.

My question is, which is the best way to get that info automatically?

Do I have to parse the Job log? And if I have to, where/how to I get that job log? I hope it's in Ascii.


Thanks

Enable DSODB. Configure it to capture the level of detail that you desire. The information you need will be sitting in a database table for you.

You can also use some of the reporting environment variables such as $APT_PM_PLAYER_TIMING to capture that detail into the job log...there is another reporting environment variable to capture it to a file that I generally don't bother with.

You can print the job log to a file and grep what you need out of it, which is what I typically end up doing in the absence of having it in DSODB.

Mike

Use the Performance Analysis tool right there in DataStage Designer (or, as Mike suggested, use the database behind the Operations Console). Or use Monitor view in DataStage Director client, where you can get figures per stage per node quite easily.

The job log is not in ASCII. It's a database table in the local (project) repository. The name of the table is RT_LOGnnn, where nnn is the internal job number. Job names can be decoded to job numbers via the DS_JOBS table, while the structure of the log tables can be had via the dssh command LIST DICT RT_LOG.

I was out last few days. I will get the admin to help with the above tips. Thanks for the replies.

Also see this developerworks article.