Page 1 of 1
CompareStage
Posted: Thu Jun 02, 2005 5:19 am
by srekant
hi,
I trying to use compare stage .I went to developers guide in it it was given
We recommend that you use runtime column propagation in this stage
and allow DataStage to define the output column schema for you.
but when i practically do it in Datastage 7.5 EE in output tab columns are not getting populated.
Posted: Thu Jun 02, 2005 5:22 am
by ArndW
srekant,
the runtime column propagation does not populate output tabs - they are implicitly passed on between stages. This allows you to actually write jobs with stages that show no columns, because they are passed onwards downstream via the propagation.
Posted: Thu Jun 02, 2005 5:32 am
by ray.wurlod
Be very careful if using runtime column propagation. Propagated columns can not be traced using MetaStage, since there is no linkage between them and the table definition in the DataStage repository.
Compare
Posted: Thu Jun 02, 2005 6:19 am
by srekant
If i am runtime column propagation then how will be my target dataset .do i have to manually create the column names.
i.e
i created a job :
Code: Select all
Seq
Compare Dataset
Oracle Enetrprise
then how am i going to have my target definition.
Re: Compare
Posted: Thu Jun 02, 2005 6:22 am
by srekant
srekant wrote:If i am runtime column propagation then how will be my target dataset .do i have to manually create the column names.
i.e
i created a job :
Code: Select all
Seq
Compare Dataset
Oracle Enetrprise
then how am i going to have my target definition.
Posted: Thu Jun 02, 2005 7:34 am
by ArndW
srekant - instead of getting an explanation that would be a copy of what the documentation states, just do it - run the job and then go into the dataset explorer or use orchadmin to see what the resultant dataset looks like. Column propagation is a wonderful and powerful tool that, unfortunately, can really mess up job design if you are not aware of what you are doing.
Comapre
Posted: Thu Jun 02, 2005 8:19 am
by srekant
ArndW wrote:srekant - instead of getting an explanation that would be a copy of what the documentation states, just do it - run the job and then go into the dataset explorer or use orchadmin to see what the resultant dataset looks like. Column propagation is a wonderful and powerful tool that, unfortunately, can really mess up job design if you are not aware of what you are doing.
Thanks for u r Advice and it is working fine i able to see the data in the dataset using dataset explorer.
What should be the structure of Oracle if i want to compare two inputs and write the output to a oracle table.
Propagated Columns n linkage to Table definiton in Repositor
Posted: Mon Jun 20, 2005 11:54 am
by thebird
Hi Ray,
Can you explain this a little more. I thought that when you had RCP enabled the linkage to the table definition is not broken unless forcibly done.
The "Table Definition Reference" in the columns tab does seem to show the linkage to the table definitions, even when the metadata is propagated.
Thanks in advance.
Regards,
The Bird.
ray.wurlod wrote: Be very careful if using runtime column propagation. Propagated columns can not be traced using MetaStage, since there is no linkage between them and the table definition in the DataStage repository.
Posted: Mon Jun 20, 2005 4:59 pm
by ray.wurlod
If you use RCP there are columns "appearing from nowhere" as far as the Repository is concerned. That is, since the table definition on the output link was not loaded from the Repository, there is no link established in the first place, so no way for MetaStage to do its thing.
RCP and Metadata linkage.
Posted: Mon Jun 20, 2005 11:35 pm
by thebird
Hi Ray,
Correct me if i am wrong, please.
As far as columns "appearing from nowhere" are concerned, I believe this happens only when RCP is enabled on stages where certain columns are being dropped, as the RCP carries forward the metadata to the stages appearing downstram. If RCP is disabled on such stages, no such incidents take place. This is what I have observed so far.
Again, once we have loaded the table definition in the source, if we have RCP enabled on all stages, i have seen that the metadata linkage to the original table definition is maintained up till the target, where the developer might want to create a new target table definition, for a metadata lineage analysis.
In this scenario, how do we say that - "Propagated columns can not be traced using MetaStage, since there is no linkage between them and the table definition in the DataStage repository."
Regards,
The Bird.
ray.wurlod wrote:If you use RCP there are columns "appearing from nowhere" as far as the Repository is concerned. That is, since the table definition on the output link was not loaded from the Repository, there is no link established in the first place, so no way for MetaStage to do its thing.