Hannes KropfProject Manager| ITERGO GmbHHamburg, Hamburg, Germany
Hello,
I am looking for someone who has experience in building a data warehouse with agile methods.
Where have been success stories and where have been failures. What do I have to keep in mind building a new team. Are there any specialities in contrast to "ordinary" agile projects.
What are the risks and what are the chances? What lessons have you learned?
All Data Warehouses should be built incrementally, which is right in line with Agile. There are two traditional DW experts, Bill Inmon and Ralph Kimball. Inmon advocates building an all-inclusive Enterprise Data Model and then building out the various subject areas of that model. Kimball endorses an approach to build subject area data marts (again incrementally) but insuring that each data mart can interface with the others via "conformed dimensions". Examples of subject areas include sales, purchases, services, products, etc. That's about as in-depth as I can go here, as there are (literally) books written about each of these methods.
The first iteration of a data warehouse should be either the one with the highest value to the business, or the one which will have strong sponsorship. The first iteration will likely result in the need for changes, as lessons will be learned. Don't fret if you have to "throw away" much of the first iteration. Plus, when the user community starts to work with the data, they will want additions and changes.
The data warehouse should definitely be built incrementally, and it is often an "evolutionary" project management methodology. This methodology is sometimes called "spiral" and is often used in prototyping. Saving Changes...
Stefan SchneiderBusiness Analyst| OTTOGROUPHenstedt-Ulzburg, Sh, Germany
Keep in mind that users not always can image what the outcome will look like. Prototyping can help here. You could try this: http://sourceforge.net/projects/palo/ (a poor TM1-clone, but free) to easily input some example data and metadata to quickly prototype some views and the desired navigation people need. This tool is not matching enterprise needs at all but is enough for the analysis task. Then you could take gained insights into user stories and build those using your enterprise bi suite like Cognos, TM1 or SAS. Saving Changes...
Stefan SchneiderBusiness Analyst| OTTOGROUPHenstedt-Ulzburg, Sh, Germany
"Life is to be lived. If you have to support yourself, you had bloody well better find some way that is going to be interesting. And you don't do that by sitting around wondering about yourself."
