Pivot stage in parellel mode.

Madhu1981 · Post by **Madhu1981** » Thu Apr 20, 2006 9:09 am

Hi,

I am using pivot stage in my job. The pivot runs in sequential mode as default.

If i cahnge it to parallel, will my functinality affects or not. Shall i change it to parallel.

My Functionality is :

I am getting four fields from pivot input and writting one filed in pivot output.

Kindly help me.

kumar_s · Post by **kumar_s** » Thu Apr 20, 2006 9:20 am

As long as the partition is based on the all the key mentioned in the PIVOT the functionality will not be affected in parallel operation.

ArndW · Post by **ArndW** » Thu Apr 20, 2006 9:28 am

This is a very basic PX type question that we cannot answer for you. If the pivot can be done within the datasets in each processing node then you should be able to work in parallel. You should try it and see if it works and perhaps provide some feedback to the forum when you've finished.

I looked into the documentation on my PC with 7.5.2 and couldn't find the PIVOT stage documented for PX! I don't know if I might have deleted it by mistake.

DSguru2B · Post by **DSguru2B** » Thu Apr 20, 2006 10:19 am

Since pivot stage is completely positional, I dont think it would even give any edge on performance also. Have experience really bad performance with large amount of data when pivoting.

kumar_s · Post by **kumar_s** » Thu Apr 20, 2006 11:53 pm

Also depends on the number of column getting pivoted.

ray.wurlod · Post by **ray.wurlod** » Fri Apr 21, 2006 12:34 am

Not so. The only difference for execution mode is that, for parallel mode, you must partition (and, ideally, sort) on all pivot key columns. For sequential mode, since all rows are in the same partition, this is not necessary, through sorting is still beneficial.