Sunday, May 17, 2009

Ralph Kimball vs. Bill Inmon

I came Across this wonderful comparison between Ralph Kimball and Bill Inmon in the way data marts Should be built I decided to share it on my blog for your reference

Areas of agreement

  • Consensus on need for solid business requirements and end-user validation
  • Agreement that it is rarely feasible to build an entire warehouse at once - incremental development with a focus on high-priority elements
  • Conformed dimensions are desirable
  • Warehouse data needs to be tracked as atomically as possible
  • Star schema is most desirable format for data marts

Areas of differences

  • Bill Inmon
    • Approach is known as Top-Down, or Corporate Information Factory
    • Warehouse data should be stored in a centralized relational structure
    • Dependent data marts should be created from central warehouse
      • These data marts will employ star schemas
      • Additional transformations may be employed between the warehouse and the data marts
    • ODS is used for transaction level detail with little to no history


Inmon Kimball

  • Ralph Kimball
    • Approach is known as Bottom-Up, or Kimball Bus Architecture
    • “… The data warehouse is nothing more than the union of all the data marts …”
    • Conformed dimensions are the glue which unite disparate data marts, while Inmon maintains conformed dimensions within centralized relational structure
    • ODS may be integrated directly into warehouse

kick it on DotNetKicks.com

1 comment:

牛牛 said...

thank u very much for ur sharing. very helpful~