DSXchange

Using multiple Transformer stages and inter-process row buffering can give you some performance gains, particularly if you have more than one processor.

Yes, put the"-" characters into the Fmt mask.

You marked the original post as version 8.

In earlier versions you need the manual called dsdskche.pdf

The warning comes whenever the cache fills with the total volume of hashed files requesting to be loaded into it.

In most UNIX implementations renice can only be used to lower an ordinary user's priority. So it's exactly the same story.

There's no point telling you where the checkpoints themselves are stored. Manipulate them with the functions you see in the job control code generated by compiling the job sequence.

I disagree with Craig, I find it quite readable.

Does this mean you're using paid time to look at DSXchange? If Catbert were your HR manager that would be deemed to be theft! I think ten is plenty. Each time I come in to DataStage (only once or twice per day, in unpaid time) there's 80-200 new posts to look at, and I wouldn't want a recent list th...

There's no such thing as the pid. A parallel job will have one conductor process, one section leader process per partition, and N player processes per partition, as specified in the score. I do hope you were not thinking about killing processes? Search the forum for many experts' opinions on how bad...

Why? It's not necessary. All you need to do is lock the record containing the flag and this will happen as part of normal transaction processing.

Also you didn't mention which database - techiques vary for different databases.

Theoretically do-able, and you've pretty much got a handle on the issues. You would need to calculate the frequencies on the individual partitions (so do separate investigations, or none), and I'm not certain that QualityStage job naming handles invocation IDs (which would be a show stopper).

TPump is a Teradata utility.

Scan through the stage types available to talk to Teradata and you will find a stage there that can talk via TPump. Maybe even two stages.

I'm not sure you can use a job parameter as a column name (though you can supply the entire SQL statement as a job parameter!). It may be easier, though, just to create views that have consistent column-naming conventions.

Of course.

You could write your own stage (a Build stage).
You could write a parallel routine, and invoke it from a Transformer stage.

Just to name two.

Try truncating the fractional seconds.

Close, but no cigar. Welcome aboard.

You have to sum up the size of all hashed files in all concurrently-running jobs.

Read Chapter 25 of the version 8 Server Job Developer's Guide to learn all there is to learn about hashed file caching.

If the exit status ( $? ) is 0, the file was traced and is 32-bit.

If the exit status is non zero there will be output.