hashed file through parallel jobs

tostay2003 · Post by **tostay2003** » Wed May 21, 2008 7:25 am

Hi,

I know that we can't use hashed file through parallel jobs, but can we use UV database to access via parallel job (which contacts datasets etc).

Thanks

OddJob · Post by **OddJob** » Wed May 21, 2008 7:37 am

When you create a Parallel job check to se if you have a Universe database control.

I tried this and didn't see one, and there don't appear to be any ODBC drivers either.

ArndW · Post by **ArndW** » Wed May 21, 2008 7:42 am

If you have UniVerse installed then you can access UniVerse hashed files from DataStage PX through ODBC. The DataStage engine doesn't have this functionality enabled, but I suspect that UniVerse can still access DataStage hashed files.

ray.wurlod · Post by **ray.wurlod** » Wed May 21, 2008 3:48 pm

ArndW wrote:I suspect that UniVerse can still access DataStage hashed files.

Only if there are no I-types and no indexes involved. Otherwise the magic numbers do not match and access is impossible.

ray.wurlod · Post by **ray.wurlod** » Wed May 21, 2008 3:52 pm

The biggest issue is that the UniVerse database is not distributed, therefore can not be accessed from any processing node that is not also the conductor node.

If you restrict the stage to execute only on the machine where UniVerse is installed, then there are several ways that you could access hashed files, including through a command line interface (External... stage) or your own routine that uses the InterCall or ICI API.

I suspect you could even use an ODBC Enterprise stage connecting to the localuv data source.

But why do you want to use hashed files in a parallel job anyway?