Don't we need Table Defs to run the DSJob ?

kaps · Post by **kaps** » Fri Dec 30, 2005 2:25 pm

Guys

In one of the posts here, it's mentioned that we don't need to promote Table Definitions to other environment if we are just need to run the job in that environment rather changing it.

Is that true ? Doesn't DataStage look for Table Defs in the repository when it runs the job ? On the other hand, It makes me think otherwise...Since we are compiling it and making it as a package it does not need them again if we don't want to change what is built ?

Very Happy New Year 2006 !

Thanks

ArndW · Post by **ArndW** » Fri Dec 30, 2005 2:50 pm

Kaps,

the table definitions, or metadata, used by the job are compiled within the job and are not needed at runtime.

chulett · Post by **chulett** » Fri Dec 30, 2005 8:26 pm

Technically, the information stored under Table Definitions is not needed at all. You can build and run and maintain jobs without ever having generated any metadata, instead typing it in by hand. Over and over again.

Not saying that it isn't a good idea to have and use them, just not absolutely needed.

ray.wurlod · Post by **ray.wurlod** » Sat Dec 31, 2005 12:30 am

Aaarrggh!!

The whole rationale behind being able to import table definitions then load them into jobs, fully or partially, using mouse clicks is that they remain untouched by human hands, and therefore error-free. This maximizes your chances of creating a working job first time.

While the other posters are technically correct, they are not, repeat not, advocating manual creation of table definitions.

Arnd's point, that what you've built into your job design isn't actually needed to run the job is true.

However, if you're using MetaStage to perform lineage analysis on your production systems, you do need to move the table definitions to the production environment, because MetaStage uses these (and the links to them from job designs) to perform that analysis.

chulett · Post by **chulett** » Sat Dec 31, 2005 8:25 am

No need to revert back to your Primal Scream therapy days, Ray.

I thought it was very clear that none of us here were advocating the non-use of metadata. Just trying to be thoroughly anal, make some additional points about metadata usage and do our own versions of a Full Wurlod. Perhaps we should leave that to the professional.

Nice to know about the MetaStage linkage, makes sense.

WoMaWil · Post by **WoMaWil** » Sun Jan 01, 2006 4:33 pm

And what about using parameters? Does MetaStage still find the link if Tablenames, DBUsers etc are norished by Parameters?

Wolfgang

roy · Post by **roy** » Mon Jan 02, 2006 12:26 am

Hi,
I think that assuming you loaded the tables definitions in the columns page and that relates to the same table/s you want to see in the table name it should show regardless of using parameters for table name/s.

then again I have no means to test it right now

Anyone able to shed some light on this?

Sreenivasulu · Post by **Sreenivasulu** » Mon Jan 02, 2006 2:13 am

In short,
The table definitions are not required for the compiling and running of jobs.

Table definitions are used when you want to do metadata analysis.

Regards
Sreeni

ray.wurlod · Post by **ray.wurlod** » Mon Jan 02, 2006 5:47 am

MetaStage (and Manager Usage Analysis) don't work if table names are job parameters. It seems happy enough if user ID, password and even schema are job parameter references, however.