Data Warehouse Structure

INTRODUCTION &OBJECTIVE:
In today’s scenario, there are two data warehouse methodologies that are extensively followed by most of the Multinational organizations, Financial Sectors and almost in all the sectors where there is a requirement of a Data Analysis & Reporting requirements. The two methodologies are top-down approach and bottom-up approach. The most common used top-down architecture is the hub and spoke architecture (i.e., centralized data warehouse with dependent data marts) that is advocated by Bill Inmon, who is commonly referred to as “the father of data warehousing” (Reference number).The Data bus architecture which uses the methodology of a bottom-up approach was introduced by Ralph Kimball who is also referred as “the Father

When it comes to the modelling of the data warehouse bill Inmon uses the Entity Relational modelling .The Overall Data warehouse can be split into four parts which are Operational, Atomic Datawarehouse, Departmental and Individual.
KIMBALL Model:
Ralph Kimball developed the bottom-up approach which uses the Data Bus architecture. According to Kimball the data warehouse needs to be developed from the business/process oriented, this introduces the concept of Data marts. The integration of data marts to create the Data warehouse is achieved by the data warehouse bus in the BUS architecture (Reference).In this approach the data from the external sources are being fed into the data marts through an ETL process via a staging .The data marts are being created first to provide the analytical & report capabilities and then the data marts are been integrated together to create an enterprise data

There are two ways in which a dimensional modelling can be modelled and they are star schema and snowflake schema modelling. There are four essential components that need to be taken care before designing a dimensional modelling which are choosing a business process, declaring the grain, choosing the dimensions and identifying the facts (ref).

SIMILARTIES: INMON vs KIMBALL
The most common similarities between these two models are the time -stamped data and the extract, transform and load process (ETL).We will now drill down in brief about the two similarities.

Similar Time Stamped Information:
The External sources and the operational systems usually can store a numerous data which can be assumed from a week data to two years data .Similarly a data warehouse can store a data of 5 to 10 years data volumes. The time parameter is the most important factor in a decision support system which allows to analyses a Product 1 sales for the last week, month, year and even the time parameter can be drilled down to day and hours, hence it is important to capture the time parameter in order to analyse various