DSXchange

Ascential regard the structure of the repository tables as intellectual property and have not published them. The commands that you can enter at the Command window in the Administrator client are, for the most part, UniVerse commands. Manuals for these can be downloaded from IBM's web site. However,...

It's up to you to manage allocation of unique identifiers! Any attempt to use the same job identifier string is the same as attempting to start a job that is already running.

The CoSort stage does not do aggregation.

Description of the Derivation field is quite clear in the documentation that accompanies the CoSort stage software. (Clicking on F1 isn't quite as useful as you might think - it opens the generic "plug in stage" help.)

A major disadvantage of distributed files, from DataStage point of view, is that they do not benefit from memory caching, which means that your I/O will potentially be 1000 times slower or worse. (Of course, unless you have a very large system, you're not going to cache 3.6GB hashed files anyway.) A...

Can someone tell me how you can find the last day of a month? I have been trying to use the ConvertMonth function for testing purposes by using ConvertMonth("2004/12","L") but I keep getting an error message indicating invalid Month.TAG. Can you explain what the Month.TAG is? I don't think I have t...

Because the record layout is different, you're going to need 100 separate DataStage jobs. Your only other possibility is to create one generic job, that reads one "column" per row, and the world's most horrible parsing to handle the 100 different record layouts. Using the KISS principle, do the 100 ...

Pre-load the lookup tables to memory. You can't do this with ODBC; you will need to use hashed files. Pre-allocate disk for the hashed files (use MINIMUM.MODULUS option, to get a value use the Hashed File Calculator on your DataStage CD). Loading the hashed files won't take long (minutes). Plus you ...

However, why not do this with a DataStage job? Read the 13 columns and output one, containing line feeds and repeats of columns 1 and 2. This technique has been discussed in the past, do a search on the Forum. It's exceedingly fast. He's doing the reverse, 1 row becomes 11. So, a splitter style tra...

Also, isn't there an issue with using GOSUB syntax within a Routine? I was under the impression that was a no-no as you tended to exit the routine when the first 'return' was encountered. Or that the presence of the multiple 'return' statements irritated the compiler... It is a known glitch in the ...

The list you get via "import hashed file definitions" intentionally filters out those hashed files used for DataStage repository, so you're safe to proceed on that basis. All of the repository hashed files have names beginning with "DS_" or "RT_", or are "VOC" or "VOCLIB". If you make sure that your...

You've written an infinite loop containing ReadSeq statements. That is, there is no exit from this loop. Try Loop While ReadSeq rfileline from rfilevar * statements Repeat However, why not do this with a DataStage job? Read the 13 columns and output one, containing line feeds and repeats of columns ...

Generating the DAT file with a Sequential File stage will be much faster. Since you're not automatically invoking the bulk loader (sqlldr), you can have your CTL file pre-written - maybe adapt the one that DataStage has produced for you. Or you can have a job design that uses an ORABULK stage - to w...

What is the relationship Datastage and IBM IBM gave Ascential a lot of money for their database. Ascential went, "Woohoo! We love you! Use our DataStage toys!" IBM get richer off Ascential's toys. -T.J. P.S. I'm sure I missed a few details here and there. Informix purchased Ardent for US$800M. IBM ...

There is a small performance overhead in using any additional software. Therefore it follows that there will be a small overhead in loading the BASIC engine. That's where the main overhead will be - actually loading the engine. A small extra overhead in process management, but PX is doing lots of th...

A job is not permitted to attach itself (see help on DSAttachJob) and therefore can not re-start itself. To effect conditional re-start, the best solution (imho, of course) is to have detection of the file handled by a job control routine (or job sequence); it then triggers the actual job, which del...