Where are we going with "The Data Warehouse"? this is a million dollar question. When everybody started building a data warehouse, the goal was a consolidated version of truth and an enterprise data repository. Then we moved on to build layers of applications and called it business intelligence. From all this chaos we delivered another vision with master data and metadata management.
Among all these issues and discoveries, somewhere we have created a series of stovepipes in the data warehouse environment. Fast forward to today, we are looking at "consolidation" of all the different silos into one data warehouse. To me it looks like we are running behind a tiger by holding its tail. How do we get this right? there is no one architecture that will fit everybody's needs nor is there one methodology that will answer every facet of developing and deploying a data warehouse for all situations.
Looking at the bigger picture, it is clear that everybody who has followed any architecture or methodology has implemented "their" own version of the methodology or altered some facet of the implementation to suit their needs. One can argue that it was the right thing to do. But on introspection, if we had paid some extra time to the architecture efforts, we could have avoided some mistakes. Well better late than never.
As we move to re-build and build new data warehouses, let us look at what we need
- Business Case
- Strong Methodologies - CIF, CIFE, DW2.0, DW Lifecycle, BI Pathway any of these are good methodologies
- Robust Roadmap - A strong roadmap to build the subject areas for the data warehouse is needed
- Robust data architecture
- Robust testing framework
- Business Applications
- USERS
If you need more information on any sibject area, you can check the following
BeyeNETWORK - for more general information
TDWI - the upcoming Chicago conference has dedicated courseware on this subject
DAMA - the DAMA conference next month has dedicated courseware on this subject
There are several fellow experts of mine who have published books on this subject. Before you use a book to follow the implementation, ensure that you have all the other material mentioed here available. Remember as Jack Welch said - "the team with the best players wins", empower yourself and your teams with the right knowledge and approach. By selecting the right architecture and approach, you can make the build and test cycles more manageable and possibly have cycles to extend the implementation to topics like MDM
I will be starting a forum on this network on "DW Architectures" soon, please post topic of interest to start discussions on.