Wednesday, June 26, 2013

Coformed dimension

Conform dimension
What is conformed dimension?
A Conformed Dimension is any type of dimension that has a single meaning or value across a data warehouse. Literally word “conform” means ‘to follow rules and guidelines’.  Taking that a conformed dimension is a dimension which follows the same rule and guidelines across the data warehouse.

Example:
Time dimension is one of the best examples that fit with Conformed dimension.
In a fact table if a key ‘20130605’ pass the staging area it should have column for below information
Year, Month, day, Quarter, Calender_week
This dimension table is conformed for all the business area which is using the data warehouse.

Another example is a Dealer details in in Automotive industry. Dealer information is one of the central dimensions for all the sales, after sales, finance, warranty.

Why conformed dimension is important?
Conformed dimension is best fit for the centralized system. Now a day’s large companies are centralizing the data which are spread across the organization in different warehouse application.

Used dimension in different application may have same or slightly different data. That creates lots of redundancy and unnecessary usage of resource.

Conformed dimension proposed by Kimball is to analyze and finalize the dimension which can be reuse in different application and can be referenced in several disconnected mart.

Which factors is holding back and how to overcome this?
To confirm the conform dimension a major collaborative effort is needed across the organization stakeholders to business analyst and data modeler.
All potential dimensions which can be conformed dimension may have some difference in attribute and stakeholders needs to compromise or change the process for the sake of modeling.
This happens rare.

Kimball suggested a way that you don’t need to do conformed dimension in first place. You don’t have to agree with all the attributes. At a minimum, you should identify a subset of attributes that have significance across the enterprise. These commonly-referenced descriptive characteristics will become the starter set of conformed attributes.

Who needs to be involved to confirm a conformed dimension?
Practically conformed dimension cannot be done enterprise wide at beginning. You need to start with the two major applications which is similar in nature. Then identify commonly referenced dimensions which in future may convert to conformed dimension.

·         To confirm a conform dimension you need:
·         Data Modeler
·         Data Architect
·         Business Analyst
·         Business Owners and stakeholders from each of the application.
·         Business SMEs who have deep knowledge in all the application.