Huge Number of Tables ..

[NaviGator]

Hi guys,

I have a client which has a huge number of tables ( > 1000 ) in their data warehouse. All these tables are considered as indicators and should be modeled as dimensions in the FM.

What is the fastest way to model them. I know that I will have to import them all to the FM and this is done in one step but in the Dimensional view I will need to rename them and adjust some hierarchies for some of them. This seems to me as a long manual process.

Do you have any recommendations ??

Thanks .. 

[MFGF]

There is a crafty way of being able to rename things - don't know whether it will help you or not:

1. Add an extra language to the project (eg if the design language is English, add English (Zimbabwe) as a second language) - do this in Project > Languages > Define Languages
2. Export a translation file containing both your original and new languages - do this from Project > Languages > Export Translation File - select both languages
3. Edit the translation file in a tool that supports mass changes or edits. Change the English values to the desired new names, leaving English (Zimbabwe) values intact.
4. Import the translation file - do this from Project > Languages > Import Translation File. Set "Translate from" to English (Zimbabwe) and "Translate into" to English.

Cheers!

MF.

[blom0344]

The tricky part of the translation by additional language is that there is no contextual reference. If you need multiple translations for a given string, then ALL identical items are translated to the first occurance in the textfile.  If Cognos detects a name convention conflict, then it will add an index to the string: Name, Name1,  Name2


So, if you need to translate 'Name' to various more meaningfull strings like 'Clientname' ,'Tenantname' 'Customername' you will need to make manual adjustments, but using the textfiles may unload the bulk of the work..

[Lynn]

The tricky part of the translation by additional language is that there is no contextual reference. If you need multiple translations for a given string, then ALL identical items are translated to the first occurance in the textfile.  If Cognos detects a name convention conflict, then it will add an index to the string: Name, Name1,  Name2


So, if you need to translate 'Name' to various more meaningfull strings like 'Clientname' ,'Tenantname' 'Customername' you will need to make manual adjustments, but using the textfiles may unload the bulk of the work..

Oh so true! Another way around this dilemma is to do one query subject at a time. I realize you have bajillions to do, but using the export/import approach so you can work in a spreadsheet  may still be faster/easier than click-click-clicking all over the model. Plus you could share the work load with someone else if there are other resources available.

[NaviGator]

Thanks guys.

I know about the language trick, I have used it before but I never thought of using it here. I should give it a try.

Lynn, I remember that I can export the whole model but I don't know if I can export part of the model like a query subject. I will need to check this as well.


Thank you all again ..

[MFGF]

Yes you can - select a query subject in the metadata tree then go to Project > Languages > Export Translation File and choose the "Selected objects" option

Cheers!

MF.

[NaviGator]

Thanks MFGF ..

Will give it a try ..