DSXchange

If you haven't made any parameter changes, then there's no need to recompile the Sequence job that runs it. If that is what you are asking.

Two typical solutions - run on a single node or switch to hash partitioning over the 'appropriate' columns. Have you tried either?

You changed what "it"? And where exactly did you make this change?

Have you checked your job's log to see if there were warnings logged during the write to the sequential file?

Confusing as the Merge stage only compares information in two files so not sure how your queries are involved. What exactly are you doing here?

Sounds like you may need debug information in your proc. Is it still being called? Throwing an error that you don't see?

Double-check the parameters you are passing the job from the Sequence, that's really the only difference here that I can think of.

Ah... Phil and his rusty six inch nail. I was just about to link them to your old post on the subject.

Talk to your DBA.

If anyone could, they would have by now I would think. Time to turn to your official support provider if you haven't already.

You need to convert it only if it is not that value. So first SetNull() on a full string match else do the conversion to a Timestamp.

1. Exactly, check the INTERRIMSTATUS which is what the job will finish with, provided your after-job routine doesn't blow up of course.

2. In addition to the missing SEEK issue, the fact that the first thing you are doing is writing an EOF to the file effectively truncates it. FYI.

I've always found Chuck Smith's work quite helpful - for example, the sixth tool down. I would routinely load the results into a table to allow querying.

Have you coded anything? Determined if you actually have a 'peformance' problem yet? Worry about indexes later when you have the ability to properly make a case.

www.ittoolbox.com auto-redirects, hence my comment.