THE OLAP/OLTP CULTURAL CONFLICT

advertisement
FORUM II
THE OLAP/OLTP CULTURAL
CONFLICT
Ron Allan
Georgetown University
allanr@georgetown.edu
Sean Kelly: Required
—
A New Mindset
TV
vs
Radio
OLTP
OnLine Transaction Processing
— Technology of Live Systems
— Stands for Regimes That are:
—
—
—
—
—
—
Tightly Controlled
Hierarchal
Schedules
Milestones
Deadlines
Detailed Planning
OLAP
OnLine Analytical Processing
— Iterative Approach
— Design is Part of the Process
— Your Done When the User Stops
Calling
OLTP
CULTURE
Core Business Systems Have Become:
— Larger
— More Powerful
— More Complex
— More Integrated
Requiring at the Detail Level:
— Close Co-ordination
— Conformity
— Minimization of Creativity
OLTP
CULTURE (cont)
The Technologies are so Complex:
— Difficult to Grasp at Detail Level
— Leads to a Culture of:
— Planning
— Milestones
— Deadlines
— Providing a Common Vocabulary for
Implementers and Users
— Wisdom of Secretary Rumsfeld
OLTP
CULTURE (cont)
Culture of Procedures has Stood the
Test of Time
— Forty Year Geometric Expansion of
— Computerized Processes
— Data Stored
— Decomposition of Project Provides
Managers with a Feeling of
Control
OLTP
CULTURE (cont)
SOME DOWNSIDES
— Techies Regard Procedures as
Drudgery
— Breeds a Culture of Control
— Breeds Passive Control
— Experimentation Deemed Inappropriate
— Conflicts with Requirement for Timely
Decisions
OLTP
CULTURE (cont)
OLTP PROCESS SUCCESSFUL BECAUSE:
— Implementers are Presented
With a Design!!
— Already Worked Out
— Tested and Refined by Use
— Requiring Little Additional Creativity
OLAP
CULTURE
Characterized by What It Is Not…
i.e., Not Buttoned Down
Bite Off Small Parts – Partial Projects
Involved Users Test Results as They
Occur
— Smaller the Bites, the More Rapid the
Adjustments
— Each Partial Project Informs the Next
Partial Project–and Overall Strategy
—
—
—
—
OLAP
CULTURE (cont)
OLAP Projects are Characterized
by the Way They End
— Battle of Britain
— “We Got Up One Day and the
Germans Didn’t Come”
— OLAP Projects are Done When
the Phone Stops Ringing
OLAP
CULTURE (cont)
OLAP Approach is Not a Non-Methodology
— Users must Review Designs for
Completeness
— Typical Business Questions Must Be
Answered
— One Type of Query Not Favored over
Another
— Methodology Structured to Permit
Evolution of Design
OLAP
TECHNOLOGY
— Data Warehousing is an Iterative
Process;
— The Users Must Be Involved
— OLTP Systems Keep Records
— OLAP Systems Provide Information
OLAP
TECHNOLOGY
THE FIELD OF DREAMS SCENARIO
— OLAP Systems provide Info for the
Non-Routine Parts of User’s Work
— If an OLAP System is Not
Intuitive, Users Won’t Use It
— Just Because You Build It Doesn’t
Mean That They Will Come
OLTP
vs
OLAP
“Give Us Your Requirements…”
—
OLTP: Performs Tasks that are
— Structured
— Routine
—
OLAP: Performs Tasks that are
— Wide in Varity
— Non-specific
— Not Contemplated at Design Time
OLTP
vs
OLAP
OLAP REQUIREMENTS DEVELOPMENT
— Often a Waste of Time
— If Enough is Known to Develop
Detailed Requirements…
— Its Probably not an OLAP Project
OLTP
vs
OLAP
DETAILED PLANNING OF OLAP
PROJECTS
— Tends to be Self-Defeating Because…
— “Planning” by Staff and Users
Generates New knowledge, Which…
— Makes Plan Obsolete, Then…
— Plan Can’t be Revised without Several
Meetings
OLTP
vs
OLAP
WORKING TO THE RULE
— Technically Sophisticated Users
Reluctant to Play, because…
— They Fear that IT will Work to the
Requirement, then…
— Declare Victory, and…
— Go On to Next Project, Leaving…
— User Requirements Met, but…
— User Needs Unmet
OLAP:
DESIGN & ITERATION
— Physical Design is Crucial
— Effective and Robust Design can’t
be Planned…
— It Must be Iterated
SEAN KELLY ONCE MORE:
The physical database design for the specific
industry is THE most Important DW
Component. If you don’t have a good PHYSICAL
DESIGN, NOTHING else can make up for it:







Not
Not
Not
Not
Not
Not
Not
the RDBMS engine
the front-end tool
the data extract tool
technical ingenuity
the data loading utilities
more processors
the business vision
 Not even consultants !
OLAP:
DESIGN & ITERATION
— Concept Well Known in Data Warehouse
Community
— Why are so Many Projects Struggling?
— Creeping OLTPization of OLAP Projects
OLAP:
THE HISTORY
— Early Data Warehouse Developers
Were Left Alone
— OLTP Managers Assumed it would
Fail
— Irrational Desire for Useful
Information
OLAP:
THE SUCCESS
— Some Data Warehouse Projects
Succeeded
— OLTP Managers Feared DW Success
without Their Assistance – and
Worse…
— Not Under Their Control
— Doug Hackney: Caché Loosened
Purse Strings
OLAP:
EXECUTIVE SPONSORSHIP
— OLTP Managers Asserted Control by
Imposing Procedures
— Sean Kelly: Executive Sponsorship
Means Someone Far Enough Up the
Hierarchy to Intimidate IT
— Make IT Knock It Off when it
Interferes with the Iterative
Process
OLAP:
END OF RESISTANCE
— Resistance to Kimball Ended about
6 Years Ago
— Accepting Kimball Did Not Lead to
Accepting OLAP
— Creeping OLTPization
— Hofer: Competent Bureaucrats Take
Over; Control the New Paradigm
OLAP:
THE DECLINE
— Decline in the Importance of
Intuitiveness
— Early Technical Selling Point:
Intuitiveness
— De-emphasis on Physical Model
— Decline of “Push the Decision
Down” to Point of Impact
OLAP:
QUERY TOOL ATTITUDE
—
—
—
—
—
Vendor Accepts Kimball
Rejects ER Diagram
Suck Up to IT Departments
Commoditization of Product
Emphasis Shifts to Sale of
Services
OLAP:
WHATS TO BE DONE
— OLTP Mindset will not Go Away
— OLAP Practitioners Must Lead
— OLTP Managers Be Brought to See
that OLTP Culture is Inappropriate
— Build a Little; Give Users a
Little
OLAP:
BE INFORMED
— Know What You Know
— Don’t be Wimpy
— Keep Users Informed
— Keep OLAP Projects in Perspective
— Don’t Hype
— Teach Executives the Difference
between OLAP and OLTP
— Wisdom of Arthur Fonzerelli
OLAP:
THE NEXT GENERATION
— Iterative Process Must Gain Acceptance
— OLTP Managers Must become Comfortable
with Iteration
— OLAP Leaders will take the Majority of
the Grief Because…
— It is They Who Put the Ball in Play
THE END
Ron Allan
Georgetown University
Phone: 202-687-8967
E-mail: allanr@georgetown.edu
http://www.georgetown.edu/users/allanr/
Download