DSXchange

Version 5 introduced the following: Hashed File Stage and Shared Memory Disk Caching The Hashed File stage has been enhanced to use a write-shared disk cache available in the DataStage server. This improves performance by using in-memory cached files and lets a single instance of a file be shared be...

Do you have an Oracle client installed on your datastage server? If so then you should be able to use the Oracle OCI stage instead of the ODBC stage. This will give you more update options and should have faster performance. There are some configuration instructions in the PDF Upgrade guide and the ...

There are a few options. You could extract it into a local hash file then put it into the database. The primary key on the hash file should ensure you only get distinct rows added to the database. If you are using Oracle you could use the option to only add new rows and not update existing rows. You...

Appears as though your quotes are in the wrong spot.

strSQL = "echo select 'FLAG-'||flag from mytable; | sqlplus " : dbLogin : "/" : dbPassword : "@" : dbName

DSExecute(Unix, strSQL, strFlag, strStatus)


Vincent McBurney
Data Integration Services
www.intramatix.com

Thanks Craig! Got my array and transaction fields mixed up.


Vincent McBurney
Data Integration Services
www.intramatix.com

If you are using an Oracle OCI stage to write to Oracle then have a look at the Array Size field on the Input - General tab. This controls how many rows are processed before a commit is performed. Fiddle with this figure to increase performance, anything from 1000 to 50000 should give you faster per...

You could do this with a sleep command embedded in a loop. The following statements can be put inside a routine and retrieves the flag from an Oracle table in what I hope is a 10 minute loop. You can pass in the login details: loop until strFlag = "N" strLogin = dbLogin : "/" : d...

Agree with Ray all his answers. I've got a some additional thoughts on when to use database views: - if it is to be used as a source by multiple jobs. You face greater risk if you are trying to maintain the same custom SQL in multiple jobs. - your query uses complex SQL and you want it managed withi...

Reporting Assistant needs to downloads project metadata into an odbc database on a client machine before the data appears in the standard reports. By default your installation comes with an Access database in the Reporting Tool directory and creates an ODBC for this database. You choose which object...

You can run DataStage jobs from other scheduling tools. The command for running a job is: dsjob -run -param paramname=paramvalue -param paramname=paramvalue jobname Setting parameters is optional. Another way is to schedule the job to run once a month and in the job control code exit straight away i...

Could you clarify your constriant? It sounds like you are only interested in making one decision and choosing a path based on that decision. If you use a constraint you will be running that statement against every row of data, it may be more efficient to run your condition statement in a sequence fi...

You see one of the drawbacks of the merge stage, also true of the aggregation stage, that DataStage builds the merge within the temp directory space which adds a performance overhead. You may find it faster to load file 2 onto file 1. Getting back to shared containers, if you want to process two fil...

You can use DataStage to FTP the files from the NT server onto the Unix server. Just write a Unix script which ftps the files and call this from a DataStage sequence or batch job or use the FTP file stage. If you want to place the data directly onto the Unix system you could consider connecting to t...

By the way, my knowledge of Unix administration functions is pretty woeful. Be interested to know if my theory works!


Vincent McBurney
Data Integration Services
www.intramatix.com

Generally speaking each DataStage job will take up about one CPU at any one time. If you restrict your sequence/batch jobs to under 4 server jobs at any one time you should leave a lot of server capacity free. To assign processors to particular jobs you will probably need the version 6 parallel exte...