Optimal Library/Model/Application Number & Structure

[araryes]

Hi - Can any one throw some clarity around how shoudl i go about deciding on the optimal library/model/application structure - i.e. how do i decide how many libraries, models and appications (analyst and contributor) that  will need to be built to satisfy a particular planning solution implementation - just the high level guiding principles will help

[mrobby]

For contributor the number of libraries is simple.

You will have no more than 2 for any given application since this is a limitation imposed by Cognos.

Best practice is to have one library that contains shared DLists, A-Tables etc.  The second library contains the cubes and links specific to that application.  This way your first library can be shared amongst many contributor applications and will force dimensions to be the same amongst your applications.

Cognos reccommends a per E-List slice of no more than 500k cells.  Can it go bigger?  Yes but performance will start to become an issue.  How you get to the 500k cells is up to you and the needs of your application.  It could be many small cubes or a few large calculation cubes etc.  Best practice is to not try to put all the cubes into one application if it is going to become to large.  Simply split out your application into multiple applications and use admin links or system links to pull data between them.



Now if you are only going to use Analyst then this a bit different.  You can use as many libraries as you want and share items with any other library.  Typically cubes can be larger with less of a performance hit.  But the catch is that Analyst uses RAM in order to process data so you will become limited as to the size of data you can manipulate.  Analyst will end up giving you an error and force you to restart Analyst in these cases.