Data Warehouse Design Vs Data Warehouse Design

Introduction
There are many different approaches to a data warehouse’s design and its development. Two approaches that stand out amongst the rest are the Top-Down approach versus the Bottom-Up approach, both of which come with their own advantages and disadvantages.
Data Warehouse vs Data Mart
Data Warehouse
• Enterprise wide.
• Amalgamation of all data marts.
• Data is gathered from the staging area.
• Makes queries on the presentation resource.
• Provides a structure of the corporate view of the data.
• Follows the Entity Relational model.
Data Mart
• Focuses on individual departments.
• Follows a singular business process.
• Makes use of a STAR join that focuses on the facts and dimensions of the data.
• It provides the relevant technology

Ralph Kimball has taken the stance that a data warehouse is a collection of conformed data marts, where the key aspect is the conforming of the dimensions among the individual data marts. The data marts are created first (rather than the data warehouse) to provide analytical and reporting capabilities for a business subject based on the dimensional model of the data. A data mart houses the lowest level of granularity data as well as the summaries that are needed for the analysis of the data. By making use of Joins nor Unions, the data marts are conformed across their individual dimensions. (Ponniah,

Table 2 - Advantages and Disadvantages of the Bottom-Down Approach (Ponniah, 2010)
Comparative Table between Bill Inmon and Ralph Kimball
Characteristic Ralph Kimball Bill Inmon
Business decision support Tactical Strategic
Data integration Focuses on individual business requirements Tackles the project from an enterprise-wide perspective
Overall structure of the data Makes use of Key Performance Indicators (KPI), scorecards as well as business performance measures Makes use of data that cannot have metrics applied as it meets multiple different information needs
The persistence of data from source systems The source systems are relatively stable The source systems are dynamic and have a high rate of change
Team skill set Smaller team containing generalists Bigger team containing specialists
Time constraints There is a pressing need for the first data warehouse Less pressure to create the first warehouse and is based off the business’ needs.
Cost to company to build Relatively low costs to get the project going Relatively high costs to get the project going.
Table 3 - Table adapted from (ZenTut,