Bill Inmon supports the Top-Down approach for
data warehouse design whereas, Ralph Kimball supports Bottom-Up design. Now let
us discuss what each are in detail, but before that it is important to answer the
following questions before starting the design of a warehouse-
- Do we need a Top-Down/Bottom-Up approach for our warehouse?
- Should the warehouse be enterprise wide or be departmental?
- What do we create first- Data Mart or the Data Warehouse?
- Should we capture dependent data marts/independent marts?
Whenever a large data warehouse is channeled to
small data marts then it’s a Top-Down approach as here the data repository
feeds into the local, departmental marts. However, when we have data marts
created first and then consolidating them into a repository then it’s called
the Bottom-Up approach.
Few Characteristics of both-
Bill Inmon’s Top-Down Approach
- Has an enterprise wide view of data.
- Is inherently architected.
- Has a single, central data storage
- Has centralized rules and control
- Gives quick results if implemented with iterations
- Takes quite a long time to build
- Has a high risk of failure
Ralph Kimball’s Bottom-Up
- Has a fast and easy implementation of manageable pieces
- Has a very less failure risk
- Id inherently incremented
- Each data mart has its own narrow view of data
- Allows redundant data
- Has inconsistent data
- Has unmanageable interface
- But, most severe drawback is Data Fragmentation
No comments:
Post a Comment