DSXchange

I suggest you point UVSPOOL someplace that exists and you'll have read/write permission.

Skewing will most likely occur. If it's data range partitioned and most of your data falls into a recent data range, you'll find an unbalanced partition. You'll want to think about repartitioning using a more distrubuting method so that during transformation you're rebalanced. Going back into the da...

Ahhh yes, my goal to be minimally invasive, preserve the simplicity of design, but make it 20x faster. A simple switch to DIRECT path loading would do the trick.

I've inherited a job that uses this stage in Automatic mode. I suspect the stage is using a CONVENTIONAL load and would like to use a DIRECT path. Having it build the control file and load file, then on command line add the DIRECT=TRUE is one option, I was hoping for something less invasive. Is ther...

Almost 5 years from joining to your first post? That must be a record!

Communication is handled via ODBC or Oracle client connectivity. If you're on Windoze, ODBC is handled thru the Control Panel on the DS server. For Oracle client connectivity you'll configure the Oracle client on the DS server. To use Oracle OCI you'll use OCI/plugin stages, for ODBC, well, ODBC.

Smells like permissions have changed. If the job object files are owned by others or groups you may have some sticky bit situations or someone changed group level permissions.

Look into ssh as well when rethinking

Where's the bottleneck? What's the job design used? What are the source/target database technologies? Did you implement parallelism techniques within the database? What are your loading methods? What's your parallelism? What's your hardware utilization? What's the target table indexing strategies (m...

I get asked this question all the time. There's differences between porting/transporting data between two databases, populating a data warehouse, data import, or building data extraction processes. My usual answer is that we move rows of data, not columns. It's never 1:1 relationship between a sourc...

Hints are suggestions, not requirements. Is the query the same as in the job? What about the parallelism in the ORA stage?

One other circumstance back on 6.0 was when you tried to run a non-multi-instance enabled job as a multi-instance via job control. Since you didn't state this as the case, it's probably a server resources. You may wish to try the Microsoft solution and reboot the server to take care of any questions...

The SCD stage doesn't address most of my points. You only can use it according to its limitations, you must model and have data that conform to its design specifications.

Yeah, well, working 90 hour weeks for the last two months means exhaustion. This is my last week straddling two customer projects 1300 miles apart. Starting next week I'm hoping to have my usual presence again.