DSXchange

7.5 worked because it did not check rigorously enough. And now you're relying on that "bug".

I recommend a web search (Google, etc.) for MD5.

It's essentially a cryptographic hash function. It does not guarantee uniqueness, and therefore is a really poor choice for a surrogate key.

Resist stupid requirements!

How many nodes are you running in parallel? What partitioning algorithm are you using?

What (precise) error message is being logged?

So long as the trimmed string is the same, a lookup using VarChar data types of different defined lengths should still work OK.

SURA wrote:I am not sure why IBM is not providing the native connectivity for SQL Server.

They can't provide it on all platforms.

There's something in your licensing agreement where you promise not to use the software for Evil purposes.

Really.

Check it out.

Link colours show the status of the links, not of the job. If the link finishes successfully, its colour will be the configured "success" colour.

The most important thing is the partitioning, but you have already mentioned that you have hash partitioned using the Join key column as the partitioning column. So that should be alright, provided you've done the same on both inputs. Next most important is that the inputs are sorted identically on ...

I think the problem is that this message, which is generated by the database server, is not appearing in the log, but gsbrown would like to see it there. I don't have an answer for this one, but might suggest you investigate environment variables that control what is emitted by the database. These w...

Then how come the Lookup stage can find the matches?!!

Can you please be more specfic about what Join keys you used, and how these were partitioned and sorted? Similarly, can you be specific about the keys used for Lookup, and how the Lookup inputs were partitioned?

... assuming powerful hardware. I note that the OP is on Windows. Nuff said.

Look at the Mapping tab on the output links. Or specify Auto-Map Columns on the stage (t without opening it).

From the outset. Those and Parameter Sets were the two aspects of functionality that went into both server and parallel (but not mainframe) job types.