Enterprise Data warehouse

Article: Enterprise data warehouse

Article Number: #0012

Date:31-03-2018

Writer: Salman Abdulkarim (Data Engineer "ETL/BI").

"Article is written in my own words"

Dear readers, My series of article is mostly related to Core knowledge of Data warehouse and Business Intelligence, in my previous articles, I tried to cover some areas related DWH* but not all such as I consider DWH as a deep sea where the depth is endless or unpredictable from the aspect of information in it. but I am trying my best to cover these areas.

before reading my this article I highly encourage to read my previous articles that help you to understand this article more clearly.

the basic purpose of writing this article to decide is our data warehouse is Enterprise data warehouse or not or let me say what is the requirements need to be fulfilled to design enterprise data warehouse. this is how we can differentiate data warehouse with Enterprise data warehouse.

Enterprise Data warehouse

(1) what is enterprise data warehouse

basically the basic purpose of the data warehouse to integrated data from the different source system and maintaining a single version of the truth that helps decision support system to use this data for the better business decision in the form of reporting or visualization etc.

but when we talk about Enterprise data warehouse so EDWH servers a centralized repository to provide services to entire enterprise

That's it I am not writing the whole paragraph to write about EDWH but let's understand from quoted lines what I mean by "EDWH servers a centralized repository to provide services to entire enterprise"

In simple words, you design a data warehouse in such a way the EDWH own the data of the enterprise systems related different subject areas and provide services from the data they own to the enterprises.

EXAMPLE OF EDWH

this is the EDWH for entire university but having different subject areas, and from this way, we can access a complete information of a related person, courses, finance, parking, campus police system, etc

the basic purpose of this EDWH having the

collection of data from different subject areas in one single place.

Now the enterprise(university) own the end to end information of different subject that help them to analyze each area in the form of reports or visualization, all the questions can be answered easily from this EDWH.

but the only one question is left!

What are the requirements or attributes that required for Enterprise data warehouse?

1) EDWH Should have a single version of the truth

that means while we having different subject areas the information can be duplicated, I would like to mention words of Inmon, The DWH servers as a single version of the truth which means keeping a single copy of the records if they are extracting from different sources, once we want to consolidate this data we make sure it's has a single copy of the record in DWH.

this concept comes under the master data management in case if you want detail information related MDM with the example you may read my article Master Management System.

2) EDWH Should have a multiple subject area

that is a major part such as EDWH own the data from different subject areas, look at the diagram above we have student system, parking system, online course system etc, the organization own complete pictures of its own data of the different subject area.

3) EDWH Should have normalized design

it's long debate point where to decide the DWH should be kept normalized or denormalized, will from my perspective its depend upon modal you are approaching for the data warehouse such as if your modal is star schema then it's denormalized or if your modal is snowflake then its own both normalized or denormalized.

it's recommended to design EDWH in a normalized form to avoid redundancy and anomalies which exist in data.

4) EDWH Should be Implemented as a mission cretical point

one of the important factor while we designing EDWH to implement MCP(mission Critical Point) such as the design of EDWH is capable to handle any failure if exist in system and disaster recovery feature because The entire underlying infrastructure should be able to handle any unforeseen critical conditions because failure in the data warehouse means stoppage of the business operation and loss of income and revenue.

5) EDWH Should be Scalable

this means that EDWH should handle the growth of data while the company also growing with the passage of time like business rules can be changed with time or any major change required in this EDWH according to business rules so EDWH should be scalable to handle this change.

End of the article,

I don't consider this is complete or in-depth information I wrote about EDWH but I consider as Initial information that I have right now, in future, I would like to share more about EDWH ones I have more professional experience in EDWH environment.

I read few articles in order to include more information references: epublications.marquette.edu,learn.geekinterview.com

This is it for today, I hope readers found this information useful.

My Previous articles :

1) OLTP vs OLAP 2) what is ETL 3) the importance of dimensions and facts 4) Master Data Managment 5) Structured data vs Unstructured data,6) Internet of things and Big Data.7)Change Data Capture 8)Merge Query9)Alert System In ETL. 10)Semantic Layer 11)Modeling Layers