DSXchange

Hi All,

We are interested in sorting the output of IBM's DB2 UDB High Performance Unload Utility. It appears that Ascential DataStage Version 7.5 has Sort Stage capabilities.

1. Has anyone used this feature for large (100GB+) volume sorting within Ascential DataStage?.

2. Have you found this feature sufficiently robust to handle your sorting needs, or do you use an external sort product such as CoSort or SyncSort?.

Any suggestion would be appreciated.

Bryceson

It's not meant for anything even remotely resembling a 'high' volume of records. We switched over to a command line UNIX sort and got at least a 10x speed improvement.

For your volumes you'll definitely need to go third party.

I concur with Craig. Using the builtin sort in DataStage for your volumes will .... hmmm, I wanted to use some witty simile but I can't think of an appropriate one - it is going to be very, very slow until it aborts.

Using CoSort or SyncSort will certainly speed things up. Have you tried to see your performance if your unload does an ORDER BY - depending on your partitioning and use of clustered indices you might get some excellent performance there without having to resort to an external sort.