Data Warehouse Architecture Inmon or Kimball

1
Data Warehouse Architecture Inmon or Kimball

• DW Architecture
• How do we choose ?
• Bill Inmon
• Ralph Kimball
• Or some variant ?

2
Data Requirements

• What do your users need ?

3
Data Requirements

• What do your users need ?
• Can the the various groups agree ?

4
Data Requirements

• What do your users need ?
• Can the the various groups agree ?
• Do they have a time scale in mind ?

5
Data Requirements

• What do your users need ?
• Can the the various groups agree ?
• Do they have a time scale in mind ?
• Does your chief information officer have a time
  scale and budget set ?
• These points will influence your approach

6
What does Bill Inmon say ?

• One large integrated warehouse schema

7
What does Bill Inmon say ?

• One large integrated warehouse schema
• Longer to deliver and more expensive

8
What does Bill Inmon say ?

• One large integrated warehouse schema
• Longer to deliver and more expensive
• No big bang, iterative approach

9
What does Bill Inmon say ?

• One large integrated warehouse schema
• Longer to deliver and more expensive
• No big bang, iterative approach
• Top down approach based upon organisation

10
What does Bill Inmon say ?

• One large integrated warehouse schema
• Longer to deliver and more expensive
• No big bang, iterative approach
• Top down approach based upon organisation
• Could feed departmental data marts

11
What does Bill Inmon say ?

• One large integrated warehouse schema
• Longer to deliver and more expensive
• No big bang, iterative approach
• Top down approach based upon organisation
• Could feed departmental data marts
• More complex

12
What does Ralph Kimball say ?

• Based on data marts, small subject based schemas

13
What does Ralph Kimball say ?

• Based on data marts, small subject based schemas
• Based on what user communities want

14
What does Ralph Kimball say ?

• Based on data marts, small subject based schemas
• Based on what user communities want
• Quicker to deliver

15
What does Ralph Kimball say ?

• Based on data marts, small subject based schemas
• Based on what user communities want
• Quicker to deliver
• Later combine datamarts into warehouse

16
What does Ralph Kimball say ?

• Based on data marts, small subject based schemas
• Based on what user communities want
• Quicker to deliver
• Later combine datamarts into warehouse
• Dimensional modelling

17
What does Ralph Kimball say ?

• Based on data marts, small subject based schemas
• Based on what user communities want
• Quicker to deliver
• Later combine datamarts into warehouse
• Dimensional modelling
• Based upon star schema's

18
What does Ralph Kimball say ?

• Kimball data bus
• move data to staging and clean
• then populate data marts from staging
• must have conformed dimensions

19
What does Ralph Kimball say ?

• Kimball data bus
• move data to staging and clean
• then populate data marts from staging
• must have conformed dimensions
• Approach
• select business process
• determine the granularity
• choose dimensions
• identify facts

20
Ralph Kimball versus Bill Inmon Comparison
21
User Involvement

• Keep your users informed
• Make them a part of the process
• Under stand their needs
• Consider setting up user forums
• Consider having power users
• Consider sending out news letters and updates
• Your project could fail if the users drift away !
• Periodically ask the users, does the solution
  meet your needs ?

22
Want to know more ?

• Thank you for taking the time to check this
  presentation
• Please check our web site below
• Please feel free to ask any question that you
  might have
• Semtech Solutions Ltd
• www.semtech-solutions.co.nz
• Email info_at_semtech-solutions.co.nz