DSXchange

Final suggestion - go for a bigger data type, for example Decimal(38) or VarChar(255). If that doesn't work, log a support call. I'm not using PX at the moment, so I can't test my theories.

I’m in a real big hurry. There are times when you can't afford to be "in a real big hurry"; when you have to step back and analyse. Indeed, anyone I teach DataStage will begin with a paper-based design before even touching the keyboard, as I've found over many years that describing what you p...

And if your read-only job doesn't go faster, then the likely bottleneck is one of the database server itself, your query (it's not a complex ten-way join by any chance?) or the network connection between the two. If you're on a LAN with everyone else emailing photos to and fro, I'd guess that the ne...

Either you or I misunderstand the term Cartesian product, then. As I understand the term, it's all rows from table B for each row in table A.

It suggests that the index rebuild was not successful. Try LIST.INDEX DS_JOBS ALL LIST.INDEX DS_JOBOBJECTS ALL to determine whether any of their indices needs to be rebuilt. Make sure you're logged in with Administrator access, and that no-one else is using the Repository when rebuilding indexes. Th...

You didn't state how you were doing the join - presumably a Join stage. No matter what method you use, you're going to have to process a large number of rows (684132000 based on the figures you supplied). There's no reason that DataStage would not be able to process this volume, provided you have th...

You have provided data to a column marked as Date or Time or TimeStamp. The ODBC driver (or the BASIC SQL Client Interface, if your target is a UniVerse table or DataStage table) has detected that the data do not, in fact, match the required format. Send your output to a text file to verify that thi...

How do you want to create the hashed file? (a) from the Hashed File stage - can't be done, but you can RESIZE the hashed file afterwards (provided it's not already hit 2GB) (b) from the UV stage - append " 64BIT" to the CREATE TABLE statement on the Edit DDL tab (c) using a CREATE.FILE or CREATE TAB...

The NUM function returns 1 if its argument can be regarded as numeric, and 0 otherwise. That's all it does. NUM(0.25) should return 1. However, as manteena suggested, NUM("0.25 ") returns 0, since " " is not a numeric component in this context. Could it be that your data are stored with leading or t...

Do a search for 2147483647 - you might learn about handling of integers in DataStage, at least in the server environment (where BigInt is treated as Integer). I can not test whether the same occurs in PX, but it sounds like it is.

The reason that this works is that UV-UTF8 encodes Unicode code points in such a fashion that any single-byte characters (for example ASCII) are still stored as single-byte characters, except for those that use the same code points as dynamic array delimiters (248 through 255).

Many customers do not know what character maps they're using. For Chinese it's much more likely to be BIG5 or GB2312 (or the Windows equivalents). Unfortunately I don't have access to an NLS-enabled system at the moment, so can't add much more.

Ascential has published a benchmark that achieved 300GB/hour (projected).

Sssshh!

Within the script, you can capture the exit status through the shell variable $? - indeed, this is what QualityStage itself does. Exit status of 0 means success, a non-zero exit status indicates some form of error. Alas, I do not know of a reference that lists all possible exit status values from Qu...