Featured Post

Practical Data Warehouse – An inside look!

In Today’s world, there are two prevalent strategies of building a Data Warehouse suggested by the 2 well known Data Warehouse Gurus Bill Inmon (Top Down Approach) and Ralph Kimball (Bottom Up Approach). Both these prevalent approaches have their own advantages and disadvantages. In this article, we...

Read More

Bill Inmon: Top down

In the top down approach by Bill Inmon, we build a centralized data repository to house enterprise wide business data. This data repository is called Enterprise Data Warehouse (EDW). The data in the EDW is stored in a normalized form in order to avoid any data redundancy and enforce a centralized version of the data.

Data Warehouse Top Down Approach by Bill Inmon

Data Warehouse Top Down Approach by Bill Inmon

The data in the EDW is stored at the most granular level enabling:

  • Flexibility to be leveraged by multiple groups within an organization
  • Flexibility to enhance for future requirements

Some of the perceived disadvantages of this approach can be summarized into:

  • Design complexity may increase with the increasing level of granularity
  • Space storage/consumption may spike, resulting into increased cost

In the top down approach, after the successful implementation of EDW, subject areas specific data marts build up is started. These data marts are de-normalized forms of subject area specific data that are also called star schema. These data marts are mostly summarized based on the end consumer data & analytical needs.

De-normalization helps with providing easier & faster access to the target data needs of these end consumers. Depending on the type of data warehouse platform that you are on, these marts can in some shape or form be virtual views and not necessarily consume the physical space while in others it can be physical tables. Key driver of creating subject area specific data marts is to help end consumers with easier & faster access. Key drivers for you to decide on virtual & physical data marts are:

  • Ease of use
  • Performance

Top Down approach helps build a centralized data warehouse that is perceived to be the single version of truth for business analytics. This helps data reliability across subject areas and helps establishing an easier centralized data reconciliation platform.

Top Down approach may require more time and initial investment into building the centralized platform. It can turn out to be a long cycle for business to perceive immediately value.