uvconfig - DS Tuning ?

[pramod_au]

Hi,

I am trying to understand how to select optimum values in uvconfig file so that different projects with different architecture can co-exist in the same server.

Overview:
One of project extensively uses multiple instance jobs while the other projects process huge volume of data (approx. 20 million records) per day. Currently some of jobs enter to unrecoverable state as the DS Engine is unable to provide resources to run the job. Hence it enters to job state code -14

How do I choose optimum values for the following parameters in uvconfig file so that all the jobs in both the projects run successfully? What are these parameters signify?

Current values:

MAXRLOCK 74
SYNCALOC 0
MFILES 50
RLTABSZ 75
MAXRLOCK 74
GSEMNUM 97

Thanks
Pramod

[ArndW]

Pramod,

I am surprised that the parameters you have chosen, since they are not the usual suspects when it comes to job tuning.

If you use a lot of hashed files then you should increase your UVCONFIG values for MFILES and T30FILES (double or triple them). This will increase the rotating file pool and potentially save some DataStage internal opening-and-closing of file units.

Since you have only stated the error (-14) and not the exact cause it would probably be counterproductive to change your UVCONFIG settings. It is going to be a better bet to look at your jobs and sequences and see if you can spread their runs out over time and perhaps look at your big (or long-running) jobs and see if they have been developed with performance in mind.

What sort of hardware and software are you running? 20Million records might be big for a single processor Windows box but hardly enough to cause a hiccup on a big 32-way P-Series running AIX.

[ray.wurlod]

You also need to change a mindset; DataStage Engine does not provide any resources - it demands them. It is the operating system that must provide the resources. If you are getting code -14 errors either the demand level is too great or the supply level is too small. Basic economics.