DSXchange

You would have a stage variable for each data type, each variable checks the data type field and saves the value if it is the right type. Each stage variable is then written to an output field. A constraint on the transformer only outputs a row when the final data type is detected. Make sure the so...

This environment is a datastage Project - which we term Production. We also have another environment termed Development. Hmmm. Environment sounds synonymous with project in the posters world. Access to servers is handled via userids and passwords, not a problem to lock down a server that way as som...

Wow. Your production and development exists within the same server? Didn't anyone find this to be a problem? That's like saying my production database and development database co-exist on the same server. Except in DataStage's case it's worse because only a single version can be installed and execut...

Anyone even read my post? There was a working solution posted irrespective of this wild goose chase on the row splitter/merge.

Since dsjob.exe is the command line interface, you're best option is to simply use this and "wrapper" it with your front end. Here's a shell script I posted to do the wrapper so that an enterprise scheduler could interface and run jobs: m I suggest you just execute the NT/DOS program dsjob in the DS...

Unless you have obscene volumes or an obscene number of columns, you could probably just use your source file and put it into hash lookups. It sounds like each row has a type. Simply make the primary key on the hash file the natural key of the data plus the column type. Load the whole source file in...

If you used an explicit path for the hash files (something like /var/opt/ds/hashfiles/) then simply using an "rm -r /your directory" is sufficient to blow it all away. If you used the default of letting your hash files fall into your project, then you must use the Universe TCL command DELETE.FILE to...

Are you talking about a transformer stage or a "transformer job"? If you execute a unix command for every row processed, you won't be running very fast. Can you please describe what it is you need done?

Ahhhhhhhhhh........splat (sound of me jumping out the top window of the highest building in town) Sucks to be you on this project. Your observations for your performance concerns are valid. Parents are processed first, children are processed second with surrogate foreign key substitutions of the par...

In a word, SANDBOX. You are missing the concept of using a sandbox to land, stage, cleanse, and do all surrogate assignment work. Once you are happy with results, extract your inserts and updates from the sandbox and blast it into the target. You'll of course argue about performance, extra work invo...

Search this forum, unlocking jobs has been covered extensively. Search for "DS.TOOLS", that is an undocumented utility that can be used.

and 2ndly Ken, does this script covers all there is for the PX edition and PX jobs? As far as I know, dsjob works for all job types. Since this script just executes dsjob and uses it to query for job status, I hope there's no problems. All of the other stuff is used to simply demonstrate gathering ...

DataStage has to map multiple disparate data types into its own generic data types. This holds true for both inbound and outbound data. That's why it's so important to use the source and target imported metadata rather than make it up or attempt to use source metadata for a target. For the purposes ...

This is covered fully on this site. Search for anything on dsjob, the unix command line interface.

Here is a script that I posted if you're looking for examples of a fully fleshed interface:

http://www.dsxchange.com/viewtopic.php?t=85578

Okay, well then here's a last parting attempt to help before going to sleep (it's 1:30am in the morning here). If you're processing 300 million rows a day, I would hope that you have opportunities for incremental update rather than full re-aggregation. Otherwise, you must keep in mind that your unit...