| For a Data Warehouse/Data-Mart solution, analyze well, but be decisive A Data warehouse has following decision points (ranging from business to outright
technical):
- The width of business information (type of business information needs it will satisfy)
- The time horizon of business information (last one year v/s last ten years)
- The depth of business information (the level of granularity)
- The number of source systems.
- The level of quality
- The kinds of schema
- The level of aggregations (totally granular v/s aggregations to satisfy high frequency queries)
- The approach towards slowly changing dimensions (which method to apply)
- Data integration approach (Data federation vs conventional ETL)
There is no limit to the level of discussions one can have around these and many other decision points. However one has to start somewhere. Like most of other management situations, one has to follow the following decision making principles:
- Get some external view on how other companies are doing. Be part of Industry user groups.
- Postpone decisions which can have lower impact if they are taken later.
- Start small and don't go for big-bang incase this is your first foray with Data Warehouse.
- Fix an arbitrator, whose decision can be considered as final.
- Design your DW in a way, so that changes at later stage are not over-whelming. A well-designed dimensional modeling can provide future extensibility.
|
