DSXchange

Before you do all the rebuilding servers (it won't help) and other suggestions (may help) but the main problem is the job design. Upsert requires equivalent to array size 1 and record count / commit frequency of 1 also, and I would guess it's overriding the 20000 value to do what is required: update...

You have a typo in your subject; it should be date(). Date conversion codes are documented in the DataStage BASIC Reference Guide and other places as well. The "DW" conversion code description is "Requests only the day number within the week (1 through 7, where Sunday is 7). If NLS lo...

The way I read your entries is that rhino1 and rhino2 are two different hostnames. You could connect by entering actual IP addresses in place of host name on the login. If you do that then you can be connected to both servers in two clients at once.

You can manually bounce only the JobMonApp process (search for tech note).

When I have seen it crash, I found java out of memory errors in some of the dump files in the java directory where the jobmoninit script lives.

Sounds like the job monitor process had crashed on your server. There are many topics and tech notes about that issue.

You could start here... pl/sql tutorial

It can depend on if the bottleneck is memory, CPU, disk, network. Tune the job design first. Also check any OS settings that can impose limits.

The DB2 parallel node to node stuff where one computer executes commands on the other... must be on like platforms (same OS) to set it up that way.

Array size of 1 can absolutely kill performance.

How good/average/bad the amount is depends upon the cost of living there. Ask your friend.

No need to use a Transformer loop. You could use "key break detection" (search the product guide for it) and the LastRowInGroup() function with stage variables. You could also back up to your Sort stage and set the keyChange column setting to True (it gives 1 or 0 if key changes) and work ...

Try a few things to learn a few things... Replace the final SQL Server target stage (is it bulk load or standard insert and what stage type is it?) with a sequential file. Run it on both servers again to gather the times. Turn off/disable the ODB temporarily on the test server or at the job level as...

You're welcome; glad I could help.

Yeah, well you made me feel guilty...

If you need an urgent response, contact the Support phone number. You're paying them after all.