DSXchange

Where did you see physmem and virtmem?

I'm not sure if the ODBC interface supports BigInt as a datatype. Are values in the BigInt range (i.e. greater than what an Int can handle, for example 2147483649 ) correctly loaded?

Yes, look into the CFF (Complex Flat File) stage.

And what happens when you turn off buffering?

The FTP stage is not suited for such compressed files. I've "cheated" in the past and defined a record length of 1 and passed compressed binary data through FTP that way, but it is neither efficient nor recommended and it is preferable to do an external FTP call.

Try turning off buffering on the hashed file. What operations do you perform in the transform stage (anything tricky or complex)?

You specify the name of the database during installation. You can check your installation log file (and response file, if you created one) to see what you did.

Do you get better output when you turn off operator combination? (APT_DISABLE_COMBINATION set to "true").

Unless you supply some more information and error messages, my guess is that the system is out of memory. Is that possible?

As far as I know these particular formats for MF Cobol files are not supported. Perhaps one of the IBM insiders here might know if it is on the list for future enhancements.

The first parameter to all of these routines is the JobHandle, which is the "file unit" to which the job name has been opened using DSAttachJob(). In this case you would also want to add a call to DSDetachJob() so that you don't have too many file units open.

Kim - you got it the first time around. Dictionary type entries: Col 1 - "D"ata or "I"-Type Col 2 - Position (for D) or Formula (for I) Col 3 - Conversion. i.e. MD2, for Masked Decimal 2 places comma separated Col 4 - Display column header Col 5 - Display format Col 6 - "S&q...

Normally the ODBC connector, as a passive stage, doesn't consume CPU. Is there a transform stage before that? If one process uses 20-30% CPU then you should be able to run 4 concurrent processes, if you attempt to run more your performance will go down. Either increase CPU and/or memory (does your s...

This is a tricky issue, which I think leads back to how the JobHandle is returned on resetting the job. I would recommend adding a line JobHandle=DSPrepareJob(JobHandle) as Craig suggested to replace your if-then-else construct. Note that DSPrepareJob() returns a different JobHandle when it resets t...

Dump your score and see how many processes your job runs. Look at the monitor to see %age CPU per stage. Try running 1 process on 1 node and see what your CPU utilization comes out to.