DSXchange

You can't read something that doesn't exist. But you can use, say, a Column Generator stage to create the additional field (and re-map the other).

Your file is breaking the rules. If strings are delimited by double-quote, they can't contain double-quote. Get the provider of the file to get it right - either use a different delimiter character (or none), or double the internal double-quote.

Don't forget that, to be able to see the analysis results (and notes, if any) n DataStage, then DataStage will need to be using shared table definitions.

When you search for fixes/patches at Fix Central, the information about FP1 should show you what's included in it. Otherwise you can read the relase notes for Fix Pack 1 in the Information Center.

XMETA is the default name for the common metadata repository. IADB is the suggested name for the Information Analyzer database. You get your choice of DB2, Oracle or Microsoft SQL Server for the former, and pretty much anything accessible via ODBC for the latter.

It CAN be done with DataStage alone, but will require coding of routines and design of restartability (and, possibly, recovery) in DataStage jobs.

The only difficult part is writing the site-specific requirements as specifications for the ETL developer.

The value that represents null must have five characters if the data type is string[5].

There's no such thing, for a user, as the third row in a table. The mechanism by which rows are stored in a table, are entirely the province of the database server.

No it doesn't. Because there is also the possibility that a match was found but IDENTITY is null in data set A. The proper test for not found is that B.ID is null. However, none of that has anything at all to do with non equi joins.

Metadata Workbench may be able to resolve some, at least, of the data flows.

Not using a Join stage.

If the two sources are in the same database, then you can have DataStage push the join query into that database.

Some solution, but not a general solution, might be possible via the range lookup mechanism.

What's a "phase" in this context?

Try disabling generation of operational metadata (an option in the job run options dialog) as a test.

Empty parameter value should be invalid - and prevent a job from running - for all parameter types except String and Encrypted. You can check for job parameter values in a before-job subroutine and immediately abort the job (by setting ErrorCode) if any is invalid, if that is a requirement. If it wa...

Your understanding is pretty close to the mark, Phil. In fact, spot on at a logical level.