Should a data warehouse be the reporting database?

[jmatt]

I have developed a data warehouse for our company that includes an accumulating snapshot for an application process very similar to Chapter 12, "Education" in "The Data Warehouse Toolkit" (Second Edition). The intention was to create BI on the pipeline, find bottlenecks and create KPI's etc. So far so good.

Now however, my boss wants to use this warehouse to generate reports, such as paper copies of applications, including things like personal statements, emergency contacts etc. These things are not in the warehouse by design, they have nothing to do with BI of the application process, which is the formal reason for the warehouse.

I say he is confusing a warehouse with a reporting database. I want to create a replicated (from production) db for these reports and not bloat the warehouse.

Am I being unreasonable? How have others handled this situation?

Thanks

[ngalemmo]

It depends.

A DW is something that grows over time, that may eventually encompass the entire enterprise as need dictates. Done correctly the various subject areas would be complete and integrate with other subject areas to provide avenues for analysis not otherwise possible.

Doing it correctly means not bringing is minimal data just to satisfy a few reports, but rather to implement a new subject area and all data related to it.

There are, of course, a lot of other considerations. Data latency being the most significant. If this is an operational requirement, then an ODS may be the correct choice.

[hang]

I also think ODS might be a good starting point, paving the way for bringing the data into dimensional data store. I don't think your boss is unreasonable, as a reporting requirement could evolve into an analytic requirement when historical reporting data needs to be part of DW. From architectural standpoint, any interesting business process should be considered as part of the DW system with dimension conformance in mind even though there might not be immediate BI requirements.

[jmatt]

I appreciate your thoughts. It is not as clear cut as i supposed. I like the idea that latency considerations as well as BI are important in deciding what goes in the warehouse.