Designing a Data Warehouse from the Ground Up

Designing a Data
Warehouse from the
Ground Up
Dustin Ryan & Mitchell Pearson
INTELLIGENT DATA SOLUTIONS
WWW.PRAGMATICWORKS.COM
Who are these guys?
 Dustin Ryan
 7 years as BI Consultant & Trainer at
Pragmatic Works
 Author, Blogger & Speaker
 Trainer of miniature ponies
SQLDusty.com
INTELLIGENT DATA SOLUTIONS
WWW.PRAGMATICWORKS.COM
Who are these guys?
 Mitchell Pearson
 BI Consultant & Trainer at Pragmatic Works
 Blogger & Speaker
 Unnaturally passionate about model trains
MitchellSQL.wordpress.com
INTELLIGENT DATA SOLUTIONS
WWW.PRAGMATICWORKS.COM
Why a Data Warehouse?
Purpose
Primary Interaction
OLTP
Execution of business
Single transaction
Interaction Method
Temporal Focus
Design Optimization
Insert, Update, Delete
Current
Update concurrency
Design Principle
INTELLIGENT DATA SOLUTIONS
3NF
Data Warehouse
Analysis of business
Aggregated
transactions
Select
Current/historic
High-performance
queries
Star Schema
WWW.PRAGMATICWORKS.COM
Four Steps
1.
2.
3.
4.
Identify the business process
Identify the grain
Choose the dimensions
Choose the measures
INTELLIGENT DATA SOLUTIONS
WWW.PRAGMATICWORKS.COM
Identify the Business Process
1.
2.
3.
4.
Business process NOT business
department
If just starting, choose high
impact, low risk area of the
business
The business can help you here
For this example  Retail Sales
High
Impact
Low
Impact
Low
Risk
INTELLIGENT DATA SOLUTIONS
WWW.PRAGMATICWORKS.COM
High
Risk
Identify the Grain
1. What does one fact row represent?
2. Choose the most atomic level
3. We can’t predict the queries!
“One row represents a movie rented by a customer from
an employee in a store on a day.”
INTELLIGENT DATA SOLUTIONS
WWW.PRAGMATICWORKS.COM
Define the Dimensions
1. Who, what, where, when?
2. Best attributes are descriptive
3. Use smallest data types possible
INTELLIGENT DATA SOLUTIONS
WWW.PRAGMATICWORKS.COM
Define the Dimensions
1. Who, what, where, when?
2. De-normalized design focuses on high
performance reads
3. Best attributes are descriptive
INTELLIGENT DATA SOLUTIONS
WWW.PRAGMATICWORKS.COM
Define the Measures
1. How the business measures success
2. Best measures are fully additive
3. Non-additive measures should be handled in
SSAS
INTELLIGENT DATA SOLUTIONS
WWW.PRAGMATICWORKS.COM
Resources
INTELLIGENT DATA SOLUTIONS
WWW.PRAGMATICWORKS.COM
Contact Information
Dustin Ryan
Mitchell Pearson
[email protected]
@SQLDusty
http://SQLDusty.com
[email protected]
@MitchellSQL
http://MitchellSQL.wordpress.com
INTELLIGENT DATA SOLUTIONS
WWW.PRAGMATICWORKS.COM