Data Warehousing Benefits, Challenges, Recommendations

1
Data WarehousingBenefits, Challenges,
Recommendations

• Presentation to WNYIR
• July 15, 2005

2
What is a Data Warehouse?

• According to two of the pioneers in the industry
• William Inmon A subject-oriented, integrated,
  time-variant and non-volatile collection of data
  in support of management's decision making
  process.
• Ralph Kimball A copy of transaction data
  specifically structured for querying and
  reporting.
• Workable Compromise
• An integrated repository of well-defined
  information with surrounding delivery tools used
  to drive the management reporting of an
  organization.
• In recent years, the term Business Intelligence
  has begun to replace the term Data Warehousing

3
Benefits of Data Warehousing

• Using information as an asset to sustain
  competitive/educational advantage
• Information self service, reduced dependency on
  IT
• Ad hoc information access on demand
• Consistent, well defined, single source of
  information
• Merging of historical and current information for
  analysis
• Merging of data from disparate sources
• Professionals can focus on analysis and judgment
  rather than processing data
• Reduction in the cost to deliver and administer
  information
• Users develop their own reports
• Future distributed systems can feed from the Data
  Warehouse
• Semi-isolation from future operational system
  changes
• Option to control distribution and security
  centrally

4
Making the Case for Data Warehousing

• Form a Data Warehouse Steering Committee
• Cross-department and cross-college
• Collect common requirements
• Make a unified, justified recommendation to
  proceed
• Schedule site visits
• Conduct a Data Warehouse Proof of Concept with
  real production data and a leading BI tool
• Highlight where higher ed is heading
• Moving into Data Warehousing
• Higher ed software vendors are increasingly
  packaging Data Warehousing solutions with their
  products
• Perform return on investment analysis by
  projecting future cost savings

5
Extract, Transform, and Load (ETL)

• The guts of the Data Warehouse
• Can consume upwards of 50 of the project
  resources
• Source file and field analysis
• Database design
• Mapping, coding, testing
• Moving history over
• Can be custom developed or use off the shelf
  tools. According to one survey, 39 of 53 higher
  ed institutions custom developed their ETL.
• Key vendors
• Informatica
• Ascential

6
Data Stores

• Central Data Repository
• Highly detailed
• History captured
• Designed for flexible, unpredictable access
• Query performance not initially a focus
• Data Marts
• Fed from the Central Data Repository
• Subset of Central Data Repository
• Structured and tuned for access by a particular
  department
• Can be a different database product

7
Business Intelligence Tool

• Front porch of the Data Warehouse
• Common Features
• Semantic layer isolating users from data files
• Metadata/data definitions
• Strong administrative and security function
• Ad hoc, parameterized, or scheduled reporting
• Web based delivery
• Key Vendors
• Business Objects
• Cognos
• Hyperion (acquired Brio)
• Oracle
• SAS

8
Typical Project Steps

• Scope and High Level Requirements
• According to one survey, initial scope was fairly
  evenly split between Student, Financial, and HR.
• Architecture Development
• Database, ETL, and BI Tool decisions
• Requirements and Detailed Design
• Programming and Testing
• Conversion, Training, and Rollout

9
Implementation Strategies

• Initially - Small, value-added scope
• Early proof of concept
• Frequent, manageable releases
• Data Warehouse Steering Committee throughout
• Small, versatile project team
• Project Manager
• Data Analyst
• ETL/Database Developer
• End User Delivery Analyst
• IT Technical Support

10
Common Challenges

• Visualizing the end result
• Agreeing on common definitions
• Availability of source system data experts
• Scope creep
• Data security / FERPA
• What to do next - analysis paralysis
• Sustaining long term momentum

11
Recommendations / Tips

• Because of its nebulous nature, a Data
  Warehousing project will always generate
  uncertainty and noise.
• The following tips are often key to navigating
  through a project successfully
• Hire/contract an experienced Data Warehouse
  professional
• Find a leader/champion and a good, interpersonal
  data analyst
• Find 2-3 key power users and rely heavily upon
  them
• Do a (reusable) proof of concept with the leading
  BI tool
• Manage scope and priorities with the Steering
  Committee
• Keep scope small, but architect for the long term
• Dont reinvent the wheel

12
Fidelis Partners

• Data Warehouse Service Offerings
• Project Assessment
• Data Warehouse Proof of Concept
• Architecture Development
• Software Selection
• Full Cycle Implementation Services
• Requirements and Design
• Programming
• Rollout
• Data Mart Development

13
Open Forum

• What are your expectations of Data Warehousing at
  your institution?
• What issues or opportunities would a Data
  Warehouse help you to address? What types of
  queries/reports would be beneficial?
• What challenges do you face in Data Warehousing
  becoming a reality?
• Other Questions?