Cube Size Estimation formula....

We need to estimate the size of a cube that will eventually have.... which
will get to the multi-terabyte range when mature....

I have a formula I picked up on the web below, but the sizes projected seem enormous... based on the data going in...

STORAGE FORMULA ESTIMATE - measure group storage (bytes) = (((2 * total # of attributes) + (4 * # of measures)) * # of fact records) / 3
Use the formula for each measure group in UDM model and sum the results. The total # of attributes argument reflects the # of attributes in dimensions that intersect the measure group.

Does anyone have another way of estimating cube size to validate this formula?

Regards,

Ray Cochrane, CPA
BI Consultant

Well. You can use a general rule of Analysis Services data being about 1/4 - 1/3 size of the same data stored in relational database.

Edward.
--
This posting is provided "AS IS" with no warranties, and confers no rights.

|||

Hi Edward,

I assume you would apply this rule of thumb to the size of just the data, i.e. excluding indexes etc. Is that correct?

Also, do you know of any links referencing this rule?

Cheers, Rob.

|||

Hmm :)

You call me to find you some references to this rule.

Here is some data for Analysis Services 2000. http://www.sqlmag.com/Article/ArticleID/19999/sql_server_19999.html
The data for T3 relational database was about 1.2 TB and data in AS was compressed down to 471GB

AS 2005 is behaving similarly.

Hope that helps;

Edward.
--
This posting is provided "AS IS" with no warranties, and confers no rights.