DSXchange

Get faster hardware.

There's nothing else.

Use a Job sequence that runs the first job, reads the file (perhaps into user variables using a "reader" routine) then allocates parameter values to the second job.

We do not attempt to diagnose errors from APT_CombinedOperatorController. You have been posting here long enough to know that, and what to do about it.

You seem to have "network unavailable" or "node unavailable" issues. Involve your network administrator. Test connectivity first with ping then with rsh.

Don't know, because you have not posted the input and output record schema nor the command being executed. Why not post the generated OSH? It will all be in that.

Try it. You will see how to do these things. Job parameters are available.

The JCL Randall provided (suitably adapted to your site) will produce one file that you can transfer using FTP.

Welcome aboard. You are going to need to test the Netezza ODBC driver using the Data Direct test utilities. Details are in the various readme files in the branded_odbc folder.

Unless the records are large, 40~50 million records is not huge.

If you are on SMP hardware, don't worry about re-partitioning - it will be done through shared memory.

The "excat equal" function? You can always write your own functions (parallel routines). You will NOT get exactly the same function as Iconv/Oconv in parallel jobs unless you use a BASIC Transformer stage. What you can do is substitute those functions that are available in parallel Transformer stage...

Your problem is that the shell in which the job runs is closed when the job finishes. So, while you could generate a new value for an environment variable (in an after-job subroutine), this would be destroyed when the environment (that is, the shell) is closed. No shell can update an environment var...

Don't know about "any". Certainly "some". Read about it in the Orchestrate Operators manual - for some reason the NOWARN Specification is not mentioned in the Parallel Job Developer's Guide.