Are you faced with justifying the required time and effort investment of dimensional modeling to people who believe you can ‘just point your BI tool at the data and let it do the work’?
If you are, and I hope you resist this crazy notion, then the following might just help.
I had to do an exec presentation a few months ago and was wondering how best to convey the flexibility of a dimensional model or star schema in terms of report development potential. I went back to basic maths and discovered that the enormous number below is the number of potential unique reports that could be produced from our current student admissions model - just one of many in the warehouse.
The conditions are that each report has 6 unique columns which are attributes of a dimension or measures in a fact. These are not permutations (where I can repeat numbers) but unique combinations.
If this doesn’t get their attention, try a derivation of it that can be related to. Maybe suggest that if you could produce these reports at the rate of one per second then it would take over 115 years to produce them all, or if they were each printed on a single A4 page (no! no! no!) and put end to end they would stretch over 1,084,865KM, or 27 times round the earth.
So next time you’re advised that ‘we know what reports we want already’ or that ‘this flexible system comes pre-packaged with hundreds of reports’ you have some ammunition to put those volumes in perspective.
We know that tomorrow, someone, somewhere will need to view the data in a different way, responding to a different demand of a different market. Dimensional models have the inherent agility to deliver to that need. Invest in them, unlike the ones on Wall St, these don’t go South.
