xsevensinzx (11/13/2016)
Jeff Moden (11/12/2016)
To wit, I frequently go with the "no data warehouse" option in favor of just writing some streamlined code to solve actual problems instead of a system to handle all problems that have yet to be realized. Sometimes it really is better to go with a dedicated solution rather than trying to catch all possibilities in a generic fashion.Yeah, that's what I was thinking too. It would still lead to multiple data management systems though, whether you called them a warehouse or not. But it's hard to create specialized data marts if the house does not support the requirements. You constantly have to update and conform in order to have one solution be the solution to all.
I agree with the "no data warehouse" if and when it fits the bill, obviously sometimes it does and sometimes it doesn't, correlation of different reporting from those systems is much more error prone and much harder than maintain a single data warehouse. A good example of the latter are situations when working with multiple diverse systems which all hold a slice of the cake but none has the full picture. On the flip side, regulatory and legislative requirements must also be met which often can complicate the picture, certain data cannot be stored, processed or analysed with other data. That situation can force the data the business to use multiple warehouses, i.e. telecom company cannot use Mobile or POTS data for Internet marketing etc.
😎