TGO Budgeting Solution – Relational vs. OLAP
Posted by Alan on January 27, 2011
As we were brainstorming the design for the TGO Budgeting Solution we eventually touched upon the topic of the back-end storage model and whether we should go with a solution that is primarily relational or “cubular” in nature.
[NOTE: I am not sure if “cubular” is a real word or not, but those of us in the BI group here TGO Consulting use it in day-to-day conversation. So if it isn’t a real word then consider this post is me copyrighting the term right now. Over the course of time you will see more of these wacky words we are trying to get into the mainstream vernacular.]
So in the past, along with Excel, I have worked with solutions that were completely relational based (i.e., Forecaster), those that were primarily relational based but with some smattering of OLAP (i.e., Enterprise Reporter), those that were fully OLAP based (i.e, PerformancePoint Planning) and those that utilized a combination of the two (i.e., Clarity CPM). The fact that I have worked with 4 different structured solutions, that were created by 4 different developers and there are 4 different storage models, pretty much shows that there is no one general consensus of the best way to design the storage for an enterprise planning solution.
No clear winner in this arena meant we had to make this decision for ourselves. To aid our decision process we laid out the pros and cons of each type of storage model into an all encompassing matrix. A cut-down, bullet point version of our matrix is shown below:
Advantages of OLAP
Data is already stored in a format ideal for analysis
- Thanks to hierarchies and roll-ups, consolidation of data is automatic
- Large data sets can be retrieved quicker
- It is easier to code complex calculations that span large member sets
Capable of storing both numbers and text
- Greater availability of IT people with knowledge of T-SQL as opposed to MDX
- Better for situations where members (e.g., Employees, Projects, etc.) need to be added “on the fly”
- Supports both Windows Authentication and SQL Server Authentication
Now what does the above ultimately mean? It means that OLAP storage is better for things like expense or revenue budgeting because the focus is on the numbers and quick roll-ups are important. However, if organizations are doing compensation or capital expenditure budgeting then relational probably makes more sense as those budgeting processes tend to require a lot of input that are not strictly numbers based (e.g, yes/no questions, project descriptions, adding new employees or new asset purchases to budget against, etc.).
In our experience, the priorities for budgeting for organizations have tended to be first and foremost expenses, then compensation, then revenue and after that everything else. So that means that neither OLAP nor relational on their own will fit the bill for an enterprise wide planning solution. Therefore, TGO has decided that our solution will utilize a hybrid approach and utilize both OLAP (via SSAS) and relational components. It is more development work and costs but ultimately will provide users a more powerful solution.