1 big cube or 10 small cubes

[Former Member]

Dear Experts,

Towards set up of BW framework for a DP implementation, I have
this trade off decision between choosing 1 large infocube (to pack data from
multiple sources) or n small cubes that stores data from individual sources

Which approach is better? considering

- Performance? does it matter with size of infocube ?

- Ease of use ? e.g. process chains, corrections, locking if any

- Back up of data. ... to one cube or several cubes.

- CVC creation? wrt this cube

-Copying/data administration

...

...

If I go ahead with single cube the data volume I am talking of
(4 years +) could be of the order of 50 million records!!

Request to share some best practices of RSA1

Thanks

Guru

[Former Member · Accepted solution]

Hi Guru,

Its better to have more than 1 infocube as it affects the performance and security of the data as well as the system.

Considering :

Performance : While interacting with a single large infocube, it might take a longer time to fetch data from it and definitely the system performance will be affected, as it might be possible than on that data some activities will be depending and if that data comes late then all those activities will also be delayed.

Ease of use: its will be easy for the process chains to fetch data consistently and faster from multiple cubes as u can run multiple process chains simultaneously.

Backup of data : data will be secure for a greater extend if stored in smaller volume in differently locations than risking it with 1 large infocube.

Instead of creating 10 infocubes u can have as less than 5-6 cubes and continue ur work. In my currebt project also more than 1 infocube is used as its better for system performance and security of data.

Cheers,

Ankit Singhal

[Former Member · Accepted solution]

Thank you Ankit,

Many thanks. As clear as it gets. I am going ahead with multiple cubes.

By 10 I meant multitple.. just an exaggeration. If I do it by source I can think of 6 cubes.

I presume this  (more cubes) is also recommended best practice by SAP too. What I didnt come across was an example diagnostic on memory, speed, copy, back up times etc... on the lines we have for live cache but that I will avoid venturing into those details now.

Full marks awarded.

Is there something called "locking" of data in infocubes if e.g. I am using the infocube in multiple process chains and those process chains run simultaneouly on same selections ? or say dialog and background. in parallel

Thanks

Guru

[Former Member · Accepted solution]

Hi Guru,

As far as only fetching of data is only concerned then a single Infocube can be accessed by multiple process chains, but if any of the process chain have to load, delete or manipulate the data in a particular infocube then logically the infocube have to be locked with some pre-defined condition.

Thanks,

Ankit

[Former Member · Accepted solution]

Thanks Ankit for the reassurance. Got the point.