DSXchange

Free certification exams at the IoD conference in Vegas this year: both DataStage and QualityStage on offer, plus all the other IBM IM products. I'm thinking of taking the QS exam to see if I can guess my way through.

It's a problem with fixed width complex flat files that you end up having to put some type of conversion or validation on almost every field - whether it's trimming or null (blank) checking. You could split the job up into multiple jobs to make it easier to maintain.

That looks like a product defect - and the Surrogate Key stage has undergone some changes in version 8. I would export that job and send it through to the IBM-Ascential support team for debugging.

I would go with the connector stage. It's got the best re-use value - you configure the stage once and save it as a connection for use it in many jobs. You can use the Connector object from the table definition import menu. It has some handy job parameter features such as letting you create new job ...

Not sure how to fix your error message problem but the documentation on custom stages is in the Advanced Parallel Job Developers Guide.

You might also have a mismatch between the key or index used by DataStage and that set against the table. With an upsert the Oracle stage needs to do a huge number of checks to see if the row is already there and if the table is configured incorrectly these will run very slowly. Get an Oracle DBA to...

Your DataStage CPU licenses are expensive while you Metadata Server license comes free (with DataStage). So put the DataStage server on it's own grunt machine so you get as much CPU and RAM as it can take while running and put the Metadata domain and repository on the smaller machine. This will give...

I think if you are upgrading a server you have to go via 7.5. An easier upgrade is to export all projects, remove all existing DataStage software and install version 8 on a clean(ish) machine. DataStage 8 *should* be able to import version 6 jobs (though I'm not positive of this) and if that doesn't...

My definition of plug-in is that it requires software from outside of DataStage to work - such as a database client or a separate processing engine. Most of the new DataStage 8 plug-ins will be for parallel jobs - such as the data quality and federation server plugins. I would call CFF a standard st...

You might find it easier to use custom routines in a parallel transformation rather than a custom stage. I put a link to IBM's custom parallel routine tutorial in my blog post The Top 7 Online DataStage Tutorials.

Upgrade to version 8 and you get a slowly changing dimension stage (for parallel jobs only).

Yes. Local Admin means you should allocate suite and DataStage roles to your local operating system users and groups. Under Information Server Directory you get two users created for you (suite admin and WAS admin) and you have to create the other users and groups through the IS admin tool. There is...

I've been able to run the Metadata Server on the bundled DB2 database and on the free SQL Server Express. It also runs on Oracle. The database server does not need to be on the same machine as the metadata server though you want fast connectivity between the two. You could put the DataStage Server, ...

Does this field ever change? Seems like the type of field that gets locked in once the source record is created and is not in fact a slowly changing field. You might be able to switch the field over to the type that doesn't get compared.

The Information Server will install onto XP even though it's not an officially supported platform. Make sure you have the Visual C++ express installed first. Recommend you install once layer at a time - repository followed by domain. Don't do an all in one install as a minor error towards the end re...