You can watch the replay for this Geek Sync webcast in the IDERA Resource Center: http://ow.ly/Er0q50A5o7q
Join IDERA and Joseph Maggi while he discusses some of the more heated debates in the data modeling world centered around the concept of agile data modeling. Much of this debate has its roots in a misunderstanding or at least conflicting views on what Agile is in the first place. Some view agile data modeling as a haphazard approach to database “design” while others view it as a way to get applications developed more quickly and efficiently. This session will explore the merits of both sides of the argument and will discuss the technical manifestations of Agile (namely Scrum and Kanban) and where data modeling fits within these agile methodologies.
2. Agenda
Agile Data Modeling
The Controversy
Misconceptions
Applying Discipline to Agile
3
1
2
4
1
Methodology vs. Mindset-
3. About me…
• United States Air Force
– Late 1980’s – Data Analyst
• DoD Contractor
– Early 1990’s – Birth of the Relational Model
• Data Modeling (and tool) Instructor
• DW/BI Practice Director
• Founding/Managing Partner – Agilarc LLC
4. About Agilarc…
• Professional Services Consulting Firm
specializing in
• Enterprise Architecture
• Data Management
• Cloud Services
• Analytics and Business Intelligence
• Agile Services
• Building agility into your best practice driven
architecture is at the heart of what we do.
• Managing Partners with over 50+ years of
combined Data Management experience
5. The Controversy 2
• Conflicting Goals
– Faster time to market
• Versus…
– Thorough Analysis prior to coding
– Pros and Cons of each
• Code refactoring expensive
• Analysis Paralysis
6. Misconceptions…
• Agile is a technique (not)
– Agile is a mindset
– Supported by multiple frameworks/techniques
• Why we model data
– Business Rules vs. Business Questions
– Different reasons based on perspective
• Agile framework aligned with DM rationale
– Here’s were things go off the rails
7. Why we model data…
• Application Development
– Shrink wrapped software
• Multiple business rules
– Supporting internal business processes
• Organizational business rules
• Data Warehousing
• Analytics
• Master Data Management
9. Agile’s Foundational Values
• People come first, before product and before process
– Personal commitment must be made between everyone on the team
– Without the team, there is no product
• Adaptation
– Change…of mind, understanding, circumstance, is inevitable, and essential
– Be ready, willing, and able to respond to changes worth responding to
– Changes may apply to people, process, or product
• Early and frequent value delivery
– The feedback loop is more important than specs on paper
– Focus relentlessly on doing valuable work and ignoring all else
– Customers seeing early and frequent ROI are more engaged and enthusiastic
• Customer collaboration
– Delighting customers requires frequent collaboration
– Spirit of partnership, not vendor-buyer or winner-loser
11. Mindset vs. Implementation Frameworks
• Agile is a mind-set
• Frameworks are the
manifestation of the mind-
set
• Understanding the mind-set
first facilitates success with
the frameworks
13. Defined Process vs. Empirical Process
Defined process
- Command and control
- Plan what you expect to happen
- Enforce the plan, sometimes
regardless of changing conditions
- Use change control, because
change is expensive
Empirical process
- Learn as we proceed
- Expect and embrace change
- Inspect & Adapt: iterative
improvements
- Estimates are usually wrong.
Don’t commit to them
16. Agile Frameworks - Kanban
• kanban defined (small k vs. big K)
– A Japanese word that means “signal card”
– A manufacturing process that tells an upstream process
to produce more
– Upstream cannot do work until they get the signal that
the downstream is done (limiting WIP)
– Facilitates JIT automation and kaizen (continuous
improvement)
– Semi-prescriptive scheduling system of artifacts,
concepts, and processes for managing the supply chain
and providing inventory control, or managing tasks and
change…requires that some process is already in place
– Formalized by Microsoft in the early 2000s as a product
development method
17. Simple Example
MoMA Admission System
• Capacity is limited (bandwidth)
• Line is continuous (backlog)
• Cards are created to represent
upstream capacity
• Every patron is provided a card
needed to enter
• When the patron is finished, the
card is dropped off, freeing up
another spot for a new patron
(signal mechanism)
• When there are no free cards, no
more entry is permitted (limiting
WIP)
18. What’s a “Pull” System?
• Find a systematic way to achieve sustainable pace of work
• Find an approach to introducing process changes that would meet little
resistance and facilitate cultural evolution
• Avoid employee burnout, increase morale, encourage empowerment
and process refinement (Culture Change!)
• Upstream “pulls” from downstream only when bandwidth is
available, rather than pushing from source with no regard for
upstream availability
• Limits WIP, balances demand on the team
• Only workers at the bottleneck remain fully loaded; everyone
else experiences slack
Why?
19. Kanban Board Development
Backlog Plan Develop Test
2 3 5
User
Story
Defect Task
Inprogress
Done
Done
Done
Inprogress
Inprogress
Done Rules:
Plan: hours
assigned, test case
agreed
Develop: Coding
done, increment in
staging ready for
review with staging
link published
Test: Test case
completed and
verified by Product
Owner
20. The Kanban Board
Visualize the Workflow
Backlog Plan Develop Test Done Rules:
Plan: hours
assigned, test case
agreed
Develop: Coding
done, increment in
staging ready for
review with staging
link published
Test: Test case
completed and
verified by Product
Owner
5 3 5
User
Story
Defect Task
Bottleneck
Under-utilized
Idle
Slack
Lead Time
21. WIP Limits
Backlog Plan Develop Test Done Rules:
Plan: hours
assigned, test case
agreed
Develop: Coding
done, increment in
staging ready for
review with staging
link published
Test: Test case
completed and
verified by Product
Owner
3 9 1
User
Story
Defect Task
22. Core Concepts & Principles
• Start now, with what you have
• Target incremental change
• Current processes, people,
titles, roles
• Every team member is a
leader
• Limit WIP
• Visualize the workflow
• Release frequently
• Identify work types
• Improve continuously
• Process visibility
• Develop sustainable pace
• Create slack
Stop starting…start finishing
23. Scrum vs. Kanban
Which is “Better?”
Kanban
- Continuous development
- “Support” based
- Ongoing timeframe
- Lead-time-based
- Specialized teams are
okay
- WIP limited directly
- Shared backlog by
multiple teams
- Less prescriptive
Scrum
- Iterative development
- “Product” based
- Discrete timeframe
- Velocity-based
- XF teams
- WIP limited by time-box
- Backlog owned by one
team
- More prescriptive
Bottom line:
The team is more important than the process. Like-minded thinkers
that believe in the Agile mind-set can make either work in any situation.
24. Data Modeling and Data Management
• Data modeling equals Emergence
– Example: resolving a many-to-many relationship
– Doctors see many Patients, Patients see many Doctors
• Adaptation in the world of Business Rules
– Driving consensus in a data model
– The importance of meta data and taxonomy
• Early and frequent value delivery
– Value vs. features
25. Data Management
• Data modeling as a component of Data Management
• Data Management “engagement” types
– Master Data Management
– Data Warehousing
– Enterprise Data Model Development
– Meta Data Management
– Data Governance and Stewardship
• Which Framework works best
– Is the team fully dedicated?
– Is there a discrete timeframe?
– Dedicated Product Owner?
– Will we be fixing defects?
– Is work in progress an issue?
– Bottleneck visibility…
26. Framework Mapping
Domain Agile Framework
Shrink Wrapped Software Development Scrum
Master Data Management Kanban
Data Warehousing Kanban
Internal Application Development Scrum/Kanban
Data Governance Kanban
Analytics and Reporting Kanban
Enterprise Data Modeling Scrum
27. Summary…
• Misconceptions lead to Controversy
– Agile a much misunderstood concept
• Agile is a mindset
• Data modeling with a purpose
• Data modeling is a component of Data Management
• Frameworks should match problem domain
28. Questions…
• Joseph Maggi
Joseph.maggi@agilarc.com
412.335.3027
Service Offerings:
• BIDW Architecture
Review/Strategy
• Custom Development
• Data Modeling/Governance
• Daily / Multi-week Scrum
Workshops
• Scrum Coaching
• Process Auditing
29. IDERA
Download any of our tools free for 14 days
at IDERA.com
Join the conversation at community.idera.com