DSXchange

Just to be clear - you purchase either a Windoze or a Unix version of DataStage. The Clients are always on Windoze, the Server is platform selectable. Everything resides on the Server side. If you write your jobs to be transportable, they can run in either Server environment. Most companies have a s...

Is this a new installation? Is this the only table you're having loading issues? Can you load the data into a another table with the same structure? There are too many variables. It sounds like you have some sort of issue writing to this table. Start with a simple design that extracts your source da...

That would have required you to build an ETL framework that captures inserts and updates and the before images of the rows updated. You would delete your inserts and update your updates with the before-image. Otherwise, you're in to using database table recovery procedures such as logging to rollbac...

Huh? 200 million network + database queries is better than a hashed equi-join lookup with preloaded/cached data? Unless you think the PX Join is doing network lookups (sparse lookup) then I don't know what you're thinking. If Server is your only option, the best choice is to find a way to use a hash...

To answer the original post, why would you even need to see values that were passed to the key expression when you already have them available? As for the ultimate question about ranged lookups against hashed files for high volumes - forget about using the UV/ODBC interface as it's better than a dat...

And that each project has an &PH& in it....

The poster has stated there's a lag between the script and the startup of the job, and then switches and states processing time is increased. Is this two issues being stated or one? If the job has a delay between the script executing the dsjob command and the job actually starting - this is probably...

Make your own version of the function and use a pathed hashed file instead.

If you're calling a before/after routine and you want the job parameters passed into the routine, you are given InputArg as the only argument to hold your parameters. One common way is to space separate the argument values (supply them as the parameters). The value of InputArg will be the space sepa...

Time dimension lookup

Oracle bitmap indexes are a well known issue with ETL. You need to drop/disable, load, and rebuild these indexes during your ETL cycle. You need to read up on bitmap indexes to understand why. The short answer is that a bitmap can have only a few entries that span the entire table. Inserting/updatin...

Since there's 30 jobs the question is complex. You may have a point of serialization that is more affected by the row count differences. You could also have logic branches based on region that have more/less overhead and thus performance may vary.

You could use a Sequence to do this all, Command stage executes the command to get the directory listing, parse the output for the first line into a variable, use that variable in another Command stage to FTP the file

ODBC/OCI loading with Sybase will never be fast, it's an inherent limitation, especially if using IQ. You must approach using the bulk loading methods for inserts and updates.

Well, don't blame me. dsjob requires credentials in most environments to be exposed on the command line. Only later functionality allows you to indirect this information into the process via a file. Command line exposure of passwords has been a long running issue for unix environments.