this isn't a syntax/ coding specific question per se, but I'm unable to find any real articles online of any significance that speak to it directly. As I begin to delve into data mining using Sql Server Analysis Services (2008R2), I am wondering if I should be taking a relational data approach (as my data currently exists) or begin with an OLAP model. I am studying daily stock market data on 5,000 symbols. This translates to 12 tables or so, low millions of rows each. At my basic, current understanding, OLAP seems like an extra layer of complexity added that I don't see justified. Any thoughts?
Some thoughts...
Putting in place an OLAP model is offereing great potential for sophisticated data analysis using MDX. MDX is quite a powerfull language with higher level concepts than SQL (e.g., different types of aggregations (e.g, open/close), hierarchical dimensions, functional language, drilldown, calculated measures etc...). Perhaps in the future, as the volume of your data will increase you might get some performance issues with raw SQL; the OLAP cube should be faster then.