DSXchange

I interviewed the initial and chief architect in Lee Scheffler interview - The ghost of DataStage past with some history in it but not all the ownership details. Do a forum search for vmark and you might find some threads about it.

Don't forget DataStage for Unix System Services on the mainframe. This is a massively parallel product that really leverages the muscle of a mainframe and does not need to generate COBOL code. Might be cheaper than the 360 and you can leverage it for off mainframe processing as well.

If you want to use reference data more than once, or use different columns from the same table/file in more than one lookup, the copy stage is excellent for performance. The source data is converted into native parallel data just once and used in different places in the job.

Good tip, much better than mine since you have to sort the data to do a remove duplicates you might as well add the key change property when you sort.

To merge the two files you would need to run Change Capture AND Change Apply. Otherwise you would only end up with output from the primary file and nothing from the lookup file. You might be better off with the Merge stage that runs updates from the lookup file against the primary file in one step.

To address the original question - you should be able to upgrade your MetaStage to version 8. The MetaStage repository gets replaced by the Metadata Server. The Business Glossary 7 becomes a better browser based version 8. Some of the MetaStage client tools are missing until IBM release the Metadata...

It still makes no sense. Why is there only one lookup row? Does this row have those actual values it in or is this some type of pseudo code? It looks like you only join on about 10 columns and then use the pseudo code columns (such as NOT50 and 200-300) to build Stage Variable business rule code. Bu...

A transformer can use stage variables to remember values from the previous row. You do this by using one stage variable to compare and several to hold the comparison values. If you remember that Stage Variables are executed in the order they appear this makes it easy to compare adjacent rows of data...

I think you can use the message handler to not only demote a warning to information but to suppress the message altogether. As Ray says this is a high risk approach. Each warning message may be telling you of a row of data being dropped from your job or an insert statement that hasn't work. You want...

There are implicit metadata validation and conversions that the stages perform without any code and explicit validation and conversion that you can encode into a Modify or Transformer stage via functions. Generally if data comes from a database most of it can be left as implicit - an integer field o...

We don't have two cent coins in Australia any more Ray. Since they wiped out the 1 and 2 cent pieces you have to add your 5 cents worth.

It depends on your ratio of inserts to updates! Faster still is to split your input into two streams and send them to a bulk load for inserts and an update only for updates.

If you don't mind support both server and parallel jobs on your site the server one will be a low fuss method - less warning messages for sequential file data. Parallel job becomes better if you have a sort requirement or you have more than a couple stages between the input and output. 2 nodes and 4...

The sequence job exception handler is like a VB on error goto statement, it is there in case an unexpected error occurs somewhere in the sequence job code. A job failing is not really an unexpected error. It is something sequence jobs are designed to handle - you use triggers to decide what happens ...

Assuming that version 8 is not as stable (as is true of any version 0 of a product is) two incentives of accepting the risk and going with the new release would be the slowly changing dimension stage and the six new QualityStage stages. Both could save you a lot of dev time and make for more robust ...