Agile Project Management Frank Maurer University of Calgary Computer Science e-Business Engineering Group
[email protected] http://ebe.cpsc.ucalgary.ca/Frank.Maurer/
Project Management? • • • • •
What is it? Why do we need it? What is important? Best experience? Worst experience?
20-Sep-07
Agile Project Management
2
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
1
Why estimate effort?
20-Sep-07
Agile Project Management
3
Copyright © 2006 Frank Maurer. All Rights Reserved.
20-Sep-07
Agile Project Management
4
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
2
20-Sep-07
Agile Project Management
5
Copyright © 2006 Frank Maurer. All Rights Reserved.
Common Management Issues • • • • • •
Building teams Risk management Project planning Team coordination Progress tracking Quality assurance
20-Sep-07
Agile Project Management
6
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
3
Team Formation: Tayloristic Way • role-based (functional, horizontal) • follow detailed plans of entire software development lifecycle • the focus is not on individuals but on the process itself! • teams are tailored to repeatable, manufacturing-like process • tend to lead to isolated islands of knowledge • • • 20-Sep-07
what is to be done how it is to be done the exact time allowed for doing it Agile Project Management
7
Copyright © 2006 Frank Maurer. All Rights Reserved.
Team Formation: Agile Way • cross-functional teams (vertical) • require less knowledge transfer (because there is no intermediates who may loose/alter knowledge) • facilitate better knowledge transfer (informally) • rotations from one role to another are common • highly specialized experts can be shared among several teams
20-Sep-07
Agile Project Management
8
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
4
Empowerment • • • • •
Self-determination Motivation Leadership Expertise Amplify learning by feedback and frequent synchronization
20-Sep-07
Agile Project Management
9
Copyright © 2006 Frank Maurer. All Rights Reserved.
Risk Management • Risk identification – E.g. unknown technologies, tools
• Risk quantification • Risk resolution – Reserve time for overcoming troubles – Define tasks that reduce risks
• Contingency plan 20-Sep-07
http://www.pru.uts.edu.au/images/risk_management_benefits.gif
Agile Project Management
10
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
5
Typical Risks • Changing scope • Technology is immature or unknown to developers • Wrong effort estimates • Low quality • …
20-Sep-07
Agile Project Management
11
Copyright © 2006 Frank Maurer. All Rights Reserved.
20-Sep-07
Agile Project Management
12
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
6
Four Project Variables •
Cost – CHAOS Reports, Standish Group, 19942002
•
Scope – Feature creep – Requirements churn
•
Time – “Adding people to a late project just makes it later” Brooks, Mythical Man Month
•
Quality – Disasters and software bugs http://www.mtholyoke.edu/~rzdalea/cs100/software_disasters/sd.htm http://www.csl.sri.com/users/neumann/illustrative.html
20-Sep-07
Agile Project Management
13
Copyright © 2006 Frank Maurer. All Rights Reserved.
Software Project Overruns • Kjetil Moløkken-Østvold, Kristian Marius Furulund: “The Relationship between Customer Collaboration and Software Project Overruns”, Proc Agile 2007, IEEE: – – – –
About 70-80% of all projects encounter effort (cost) overruns The average magnitude of effort overruns is 30-40% Similar results for schedule overruns No apparent change the past 30-40 years
• Moløkken-Østvold and Jørgensen, "A Comparison of Software Project Overruns“, IEEE Transactions on Software Engineering, 2004: – Effort overruns based on development process • Projects using sequential processes: Median= 60% (Mean=55%) • Projects using flexible processes: Median=1% (Mean=24%) • Interviews found that flexible development processes fostered good collaboration with the customer 20-Sep-07
Agile Project Management
14
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
7
Project Management • Project management = planning, organizing, controlling of tasks & resources to accomplish a defined objective • What, who, when, how much (i.e. costs) • Command and control
20-Sep-07
Agile Project Management
15
Copyright © 2006 Frank Maurer. All Rights Reserved.
Project Management Tasks •
Planning the project – Define tasks, task dependencies and milestones
•
Estimate effort
•
Scheduling the tasks
– How long will it take to do something – Define start and finish dates
•
Assign tasks – Who will work on it – Resource leveling – Myth: “if we fall behind the schedule, we can always add more programmers and catch up later in the project”
•
Tracking progress – Conducting periodic project status meetings – Determine whether formal project milestones have been accomplished – Compare planned and actual end dates
20-Sep-07
Agile Project Management
16
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
8
Gantt Chart
20-Sep-07
Agile Project Management
17
Copyright © 2006 Frank Maurer. All Rights Reserved.
Managers and Leaders • Managers: command and control – Define and assign tasks – Gather status reports and track progress
• Leaders: convince and steer – Help team to plan project – Coach team members – Remove obstacles
20-Sep-07
Agile Project Management
18
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
9
Agile Management Strategy • Team members accept responsibility – Tasks are not assigned but team members sign up for them
• Committed to do quality work • Not much management overhead • Coaching & mentoring (software apprentice)
20-Sep-07
Agile Project Management
19
Copyright © 2006 Frank Maurer. All Rights Reserved.
Four Values • Communication – “problems with projects can invariably be traced to somebody not talking to somebody else about something important” XP p. 29
• Simplicity – “what is the simplest thing that could possibly work?” – YAGNI – you ain‟t gone need it
• Feedback – Put system in production ASAP – “Have you written a test case for that yet?”
• Courage – Hill climbing (simple, complex, simpler,..) – Big jumps take courage 20-Sep-07
Agile Project Management
20
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
10
20-Sep-07
Agile Project Management
21
Copyright © 2006 Frank Maurer. All Rights Reserved.
Project Vision • Vision = Statement of what the business will look like once the new system is implemented. • Used to establish a project budget • Established by product owner – Provides/finds funding for projects
• Vision includes – Anticipated benefits for business – Assessment criteria for management to evaluate progress and conformance to vision Management oversight needed 20-Sep-07
Agile Project Management
22
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
11
Product Owner • • • • •
Defines project vision big picture Provides/finds funding for projects Checks ROI Prioritizes backlog One person – must represent all stakeholders
20-Sep-07
Agile Project Management
23
Copyright © 2006 Frank Maurer. All Rights Reserved.
DSDM Process Feasibility Feasibility
Agree schedule Create Functional Prototype
Functional Model
Implement Identify Functional Prototype
Review business busines s
Implementation
Train users
User approval and user guidelines
Review Prototype
Identify Design prototype Agree Schedule
Design and Build Iteration
Review Design Prototype
Create Design Prototype
20-Sep-07
Agile Project Management
24
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
12
Return on Investment (ROI) • Payback time
Software by Numbers, p. 16
• Net present value, internal rate of return SE Economics • Monetary versus non-monetary payback 20-Sep-07
Agile Project Management
25
Copyright © 2006 Frank Maurer. All Rights Reserved.
A Matter of Trust: Business Contracts •
Fixed scope/fixed price contracts – Trust by contract – Attempts to move technical risk to development side – Contract requires documentation imposes process – Opposing sides of table
• •
How are fixed prices derived by development organization? Time and expenses contracts: Fixed budget/ variable scope/early termination – – – –
Trust by feedback and involvement Collaborative environment Changes easy Issues: • No time limit on project • No guaranteed functionality
20-Sep-07
Agile Project Management
Cost of change requests
How urgently do I need the contract?
Risk multiplier (Insurance premium) Honest effort estimate
26
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
13
The Relationship between Customer Collaboration and Software Project Overruns • •
Good collaboration is subjective, and not precisely defined This paper (and presentation) highlights these collaboration issues – Communication – Contracts – Customer capability
•
In-depth analysis of 18 projects conducted by a contractor – Follow up of the large-scale study in 18 different organizations – Personal interviews
•
Overrun measure =
BREbias
(actual estimate ) min( actual , estimate )
Moløkken-Østvold, Furulund: “The Relationship between Customer Collaboration and Software Project Overruns”, slide used with permission
20-Sep-07
Agile Project Management
27
Copyright © 2006 Frank Maurer. All Rights Reserved.
Contracts • Contracts are important since they often regulate collaboration (directly or indirectly) Target Pricing • Common contract types – Time and material – Fixed price – Target price (better: Flexible pricing) • Mutual sharing of cost overruns (and vice versa) • Floors and ceilings for cost sharing
A pricing method that involves (1) identifying the price at which a product will be competitive in the marketplace, (2) defining the desired profit to be made on the product, and (3) computing the target cost for the product by subtracting the desired profit from the competitive market price. The formula Target Price - Desired Profit = Target Cost Target cost is then given to the engineers and product designers, who use it as the maximum cost to be incurred for the materials and other resources needed to design and manufacture the product. It is their responsibility to create the product at or below its target cost. http://www.answers.com/topic/target-pricing?cat=biz-fin
Moløkken-Østvold, Furulund: “The Relationship between Customer Collaboration and Software Project Overruns”, slide used with permission
20-Sep-07
Agile Project Management
28
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
14
Contract form and overruns Boxplot of BREBias vs Contract form
1,5
BREBias
1,0
0,5
0,0
-0,5 1 - By the hour
2 - Fixed price 3 - Target price Contract form
4 - Other
N
Mean
Median
By the hour
4
0.55
0.37
Fixed price
5
0.33
0.19
Target price
7
0.10
0.21
Other
2
0.13
0.13
Moløkken-Østvold, Furulund: “The Relationship between Customer Collaboration and Software Project Overruns”, slide used with permission
20-Sep-07
Agile Project Management
29
Copyright © 2006 Frank Maurer. All Rights Reserved.
Contact frequency and overruns
BREBias
Boxplot of BREBias by Communication frequency 2,0
Level
Mean
Median
1,5
Daily
0.09
0.19
1,0
Not Daily
0.58
0.35
0,5
0,0
Daily
Not daily Communication frequency
• •
A Kruskal-Wallis test for difference results in p=0.023 The corresponding size of effect is d=1.25, indicating a large size of effect
Moløkken-Østvold, Furulund: “The Relationship between Customer Collaboration and Software Project Overruns”, slide used with permission
20-Sep-07
Agile Project Management
30
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
15
20-Sep-07
Agile Project Management
31
Copyright © 2006 Frank Maurer. All Rights Reserved.
Mike Cohn: Agile Estimating & Planning
Why we plan “Planning is everything. Plans are nothing.” Field Marshal Helmuth Graf von Moltke
• • • • •
Reduce risk Reduce uncertainty Support better decision making Establishing trust Conveying information
20-Sep-07
Agile Project Management
32
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
16
Barry Boehm’s Cone of Uncertainty
Mike Cohn: Agile Estimating & Planning, p. 4
Upfront Planning and the Cost of Change Standard SE Cost of change
Agile assumption time 20-Sep-07
Agile Project Management
34
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
17
Mike Cohn: Agile Estimating & Planning
Why plans fail • Completion of activity vs feature delivery – Activities don‟t finish early Parkinson‟s Law: Work expands so as to fill the time available for its completion – Lateness is passed down the schedule: One thing that goes wrong is passed on (while all things must go right for early start) – Activities are not independent
• Multitasking causes further delays Ibid. p 15
20-Sep-07
Agile Project Management
35
Copyright © 2006 Frank Maurer. All Rights Reserved.
20-Sep-07
Agile Project Management
36
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
18
Some Terminology
• Project planning, iteration planning, planning game (XP), sprint planning (Scrum) • Story card/index card (XP), backlog entry (Scrum), feature/feature set (FDD) • Customer, goal donor/user, gold owner/client, product owner, scrum master • Spike 20-Sep-07
Agile Project Management
37
Copyright © 2006 Frank Maurer. All Rights Reserved.
Architectural Spike • • • •
Throw-away prototype Answers technical issue Reduce technical risk or improve reliability Usually: Pair for 1-2 weeks
20-Sep-07
Agile Project Management
38
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
19
Project Planning: Agile Way • Business value focused – User stories, features
• Project scope not fixed at beginning reactive to changing business needs • Short timeframes – 1 week – 3 months
• Planning and coordination are team efforts – – – –
Planning game Product backlog Daily standup meeting (scrum) Estimates done by developers
20-Sep-07
Agile Project Management
39
Copyright © 2006 Frank Maurer. All Rights Reserved.
Time Boxes • Never slip a date change scope • Sometimes external deadlines are HARD • Advantages – Increased motivation • Successful delivery keeps developers and customers happy
– Faster feedback – Creates a constant project heartbeat – Deadlines create pressure (counters: work fills time available)
• Advantages of flexible dates – Release only when required scope is completed – Overly optimistic deadlines are made more realistic 20-Sep-07
Agile Project Management
40
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
20
Feature-Driven Development (Coad, Lefevbre, De Luca) • Deliver frequent, tangible, working results that are “useful in the eyes of the client” • A feature defines a task • Group features into business-related sets • Focus on delivering results every two weeks • Track and report progress by feature progress
20-Sep-07
Agile Project Management
41
Copyright © 2006 Frank Maurer. All Rights Reserved.
The Five FDD Processes
Peter Coad et al: Java Modeling Color with UML, Prentice Hall 1999, p.190
20-Sep-07
Agile Project Management
42
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
21
FDD Process
Peter Coad et al: Java Modeling Color with UML, Prentice Hall 1999, p.198
20-Sep-07
Agile Project Management
43
Copyright © 2006 Frank Maurer. All Rights Reserved.
Agile Project Planning • Project vision the really big picture • Release planning strategic picture – – – –
Chooses a few months worth of user stories/features Date and scope Can be changed Creates product backlog
• Iteration planning tactical picture – – – – –
Few weeks Set of stories prioritized by customer Creates sprint backlog Define set of tasks for each story Task granularity: 1-3 work days estimation accuracy
20-Sep-07
Agile Project Management
44
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
22
20-Sep-07
Agile Project Management
45
Copyright © 2006 Frank Maurer. All Rights Reserved.
Agile Requirements Definition • User stories/Backlog Entries Feature requests – On index cards – Short descriptions of a feature – In customer language, no techno babble – Provide value to customer – Independent of each other – Testable – Small decompose large stories
• Estimated by developers: best case, most likely, worst case • Collect story cards and prioritize them 20-Sep-07
Agile Project Management
46
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
23
When is a User Story Done? • • • •
All unit tests pass All acceptance test pass The customer accepts it All refactorings are done
20-Sep-07
Agile Project Management
47
Copyright © 2006 Frank Maurer. All Rights Reserved.
Who Decides ? • Business decisions – – – –
Scope: which “user stories” should be developed Priority of stories Composition of releases Release dates
• Technical decisions – – – –
Time estimates for features/stories Elaborate consequences of business decisions Team organization and process Scheduling
20-Sep-07
Agile Project Management
48
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
24
Mike Cohn: Agile Estimating & Planning, p 135
20-Sep-07
Agile Project Management
49
Copyright © 2006 Frank Maurer. All Rights Reserved.
Managing a Release • Value Driven Releases • Business value = f(cost, time, functionality, quality) • 80% of the business value can be derived from 20% of the functionality • Linear development: Christmas wish lists • Iterative development: prioritized wish list
20-Sep-07
Agile Project Management
50
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
25
M. Denne, J. Cleland-Huang: Software by Numbers, Prentise-Hall, 2004
Minimum Marketable Features • Components with intrinsic marketable value • Creates business value by – – – – –
Competitive differentiation Revenue generation Cost Saving Brand projection Enhanced loyalty
20-Sep-07
Agile Project Management
51
Copyright © 2006 Frank Maurer. All Rights Reserved.
Mike Cohn: Agile Estimating & Planning, p 83+85
Prioritization of features • Amount of risk removal High
High risk Low Value
High risk High Value
Avoid
Do first
Low risk Low Value
Low risk High Value
Risk
• Financial value • Development cost • Amount of learning
Do last
Low Low
20-Sep-07
Agile Project Management
Do second Value
High
52
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
26
20-Sep-07
Agile Project Management
53
Copyright © 2006 Frank Maurer. All Rights Reserved.
Low fi prototypes describing product vision • • • •
Sketches Storyboards Pictive Wizard of Oz
20-Sep-07
Agile Project Management
54
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
27
Courtesy of Dr Sharlin/Greenberg (CPSC 481)
Sketching and Prototyping Early design Brainstorm different representations Choose a representation
Sketches & low fidelity paper prototypes (LO-FI)
Rough out interface style Task centered walkthrough and redesign Medium fidelity prototypes Fine tune interface, screen design Heuristic evaluation and redesign Usability testing and redesign High fidelity prototypes Limited field testing Alpha/Beta tests
Working systems
Late design
Courtesy of Dr Sharlin/Greenberg (CPSC 481)
Sketches & Low Fidelity Prototypes • Paper mock-up of the interface look, feel, functionality – quick and cheap to prepare and modify
• Purpose – brainstorm competing representations – bring out user reactions – bring out user modifications / suggestions
Desktop Technology Program
28
Courtesy of Dr Sharlin/Greenberg (CPSC 481)
Sketches – drawing of the outward appearance of the intended system – crudity means people concentrate on high level concepts – but hard to envision a dialog‟s progression Computer Telephone Last Name: First Name: Phone:
Place Call
Help
Courtesy of Dr Sharlin/Greenberg (CPSC 481)
The attributes of sketches • Quick – to make
• Timely – provided when needed
• Disposable – investment in the concept, not the execution
• Plentiful – they make sense in a collection or series of ideas
• Clear vocabulary – rendering & style indicates it‟s a sketch, not an implementation
Desktop Technology Program
• Constrained resolution – doesn‟t restrain concept exploration
• Consistency with state – refinement of rendering matches the actual state of development of the concept
• Suggest & explore rather than confirm – value lies in suggesting and provoking what could be – sketches are the medium to conversation and interaction
29
Courtesy of Dr Sharlin/Greenberg (CPSC 481)
Storyboarding
– a series of key frames as sketches • originally from film; used to get the idea of a scene • snapshots of the interface at particular points in the interaction
– users can evaluate quickly the direction the interface is heading
Excerpts from Disney’s Robin Hood storyboard
Courtesy of Dr Sharlin/Greenberg (CPSC 481)
note how each scene in this storyboard is annotated
Desktop Technology Program
30
Courtesy of Dr Sharlin/Greenberg (CPSC 481)
Initial screen
Change the color ->
Scan the stroller ->
Place the order ->
Courtesy of Dr Sharlin/Greenberg (CPSC 481)
Alternate path…
Touch previous item ->
Desktop Technology Program
Scan the shirt ->
Delete that item->
31
Courtesy of Dr Sharlin/Greenberg (CPSC 481)
Pictive plastic interface for collaborative technology initiatives through video exploration Muller, CHI 1991
• Designing with office supplies – multiple layers of sticky notes and plastic overlays – different sized stickies represent icons, menus, windows etc.
• interaction demonstrated by manipulating notes – new interfaces built on the fly
• session videotaped for later analysis – usually end up with mess of paper and plastic!
Courtesy of Dr Sharlin/Greenberg (CPSC 481)
Pictive • Can pre-make paper interface components buttons
menu
alert box
combo box tabs
list box entries
Desktop Technology Program
32
Desktop Technology Program
33
Desktop Technology Program
34
Courtesy of Dr Sharlin/Greenberg (CPSC 481)
Medium Fidelity Prototypes • “With a computer” • Many different types – from simple computer drawn images to partially working systems
• May take longer to generate and change than low-fi • Benefits – Seems more like the completed detailed system, provides a clearer idea of how it works – May allow user testing (not true for all medium fidelity prototypes).
• Pitfalls – User‟s reactions are usually “in the small” • Blinds people to major representational flaws because of a tendency to focus on more minor details
– Users more reluctant to challenge/change the design itself • Designs are too “pretty”, developers‟ egos…
– Management may think its real!
Courtesy of Dr Sharlin/Greenberg (CPSC 481)
Medium Fidelity Painting/drawing packages • draw each storyboard scene on computer – very thin horizontal prototype (across features, no functionality) – does not capture the interaction “feel” Control panel for pump 2
Control panel for pump 2
DANGER! coolant flow 45 %
coolant flow 0 %
next drawing
retardant 20% speed 100%
Shut Down
Desktop Technology Program
retardant 20% (for shut down condition)
speed 100% Shut Down
35
Courtesy of Dr Sharlin/Greenberg (CPSC 481)
Medium Fidelity Scripted simulations • create storyboard with media tools on a computer – scene transition activated by simple user inputs – a simple vertical prototype – Can use PowerPoint…
• user given a very tight script/task to follow – appears to behave as a real system Control panel for pump 2 – script deviations blow the DANGER! simulation coolant flow 45 % coolant flow 0 % retardant 20% speed 100% Shut Shut Down Down
Courtesy of Dr Sharlin/Greenberg (CPSC 481)
Medium Fidelity Interface builders –Tools for letting a designer layout the common widgets –Construct mode • Change attributes of objects
–Test mode: • Objects behave as they would under real situations
–Excellent for showing look and feel • A broader horizontal prototype • But constrained to widget library
–Vertical functionality added selectively • Through programming
Desktop Technology Program
36
Courtesy of Dr Sharlin/Greenberg (CPSC 481)
• “pay no attention to the man behind the curtain!”
Courtesy of Dr Sharlin/Greenberg (CPSC 481)
Wizard of Oz • A method of testing a system that does not exist – the listening typewriter, IBM 1984
Dear Henry
Speech Computer
What the user sees From Gould, Conti & Hovanvecz, Comm ACM 26(4) 1983.
Desktop Technology Program
37
Courtesy of Dr Sharlin/Greenberg (CPSC 481)
Wizard of Oz • A method of testing a system that does not exist – the listening typewriter, IBM 1984 Dear Henry
Dear Henry
Speech Computer
What the user sees
The wizard
From Gould, Conti & Hovanvecz, Comm ACM 26(4) 1983.
Courtesy of Dr Sharlin/Greenberg (CPSC 481)
Wizard of Oz • Human „wizard‟ simulates system response – – – –
interprets user input according to an algorithm controls computer to simulate appropriate output uses real or mock interface wizard sometimes visible, sometimes hidden • “pay no attention to the man behind the curtain!”
• good for: – adding simulated and complex vertical functionality – testing futuristic ideas
Desktop Technology Program
38
20-Sep-07
Agile Project Management
77
Copyright © 2006 Frank Maurer. All Rights Reserved.
Scrum Flow (Sutherland, Schwaber and Beedle)
Ken Schwaber, Agile Project Management with Scrum, Microsoft Press: 2004.
Scrum: 15 min daily meetings Team members respond to basics: -What did you do since last Scrum? -Do you have any obstacles? -What will you do before next meeting?
Sprint: 30 days
Features assigned to Sprint
Desktop Technology Program
Potentially Shippable Functionality
39
Cohn‟s Iteration Planning (p 145ff) • Tasks are not allocated during iteration planning devs pick 1-2 at start of iteration and then the next when these are done built “we‟re all in this together” attitude • Iteration vs Release Planning (p. 149 Release Plan
Iteration Plan
Planning horizon 3-9 months
1-4 weeks
Items in plan
User stories
Tasks
Estimated in
Story points or ideal days
Ideal hours
20-Sep-07
Agile Project Management
79
Copyright © 2006 Frank Maurer. All Rights Reserved.
Agile estimation process
20-Sep-07
Agile Project Management
80
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
40
Size estimation • Story points or Ideal Days (Gummy Bears, Effort, …) – “Complexity” or “size” of task – Relative to other tasks
• Based on experiences from the past • Team effort – Optimism wins – Team usually does not overrule the estimate of programmers responsible for a task
Estimates are not commitments
• Presumed Issue: Effort estimates done by developers might lead to slack 20-Sep-07
Agile Project Management
81
Copyright © 2006 Frank Maurer. All Rights Reserved.
Ideal days • How long is an American football game: – 4 x 15min = 60min (ideal time – Approx. 3h (elapsed time)
• Elapsed time is influenced by – – – – – –
amount of none-development tasks estimation accuracy available developer time experience number of concurrent projects …
20-Sep-07
Agile Project Management
82
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
41
Mike Cohn: Agile Estimating & Planning, p 69ff
Story points vs ideal days • Story points – – – – –
Help drive cross-functional behavior Estimates do not decay Pure size measure Often faster My ideal days are not your ideal days
• Ideal days – Easier to explain to outsiders – Easier at first 20-Sep-07
Agile Project Management
83
Copyright © 2006 Frank Maurer. All Rights Reserved.
Mike Cohn: Agile Estimating & Planning, p 49ff
Techniques for estimating “Prediction is very difficult, especially about the future” Niels Bohr
• You will NOT be 100% accurate • Diminishing return of more estimation effort • Estimation scale: stay within one order of magnitude – User stories, epics, themes
• Deriving an estimate – – – –
Expert opinion Analogy Disaggregation Planning poker
20-Sep-07
Agile Project Management
84
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
42
Mike Cohn: Agile Estimating & Planning, p 121ff
Splitting user stories • Split along the boundaries of the data supported by the story – E.g. Loan summary List of individual loans List of loans with error handling
• Split based on operations performed within a story – E.g. separate CRUD operations
• Remove cross-cutting concerns – E.g. story without and with security
• Separate functional from non-functional requirements – Make it work, then make it fast
• Tracer bullet through all layers with partial story functionality 20-Sep-07
Agile Project Management
85
Copyright © 2006 Frank Maurer. All Rights Reserved.
Reflecting plan uncertainty • (Best case), most likely case, worst case • Project buffer: – Max 70% of must have features – MoSCoW rule (must have, should have, could have, won‟t have) in DSDM – Alternative: calculate standard deviation n
• Slack needed for learning Tom DeMarco: Slack
(wce mlce )
2
i
i
i 1
20-Sep-07
Agile Project Management
86
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
43
Velocity • Measures rate of progress of a team • Amount of story points completed in the last iteration • Best guess: next iteration = same as last iteration (“yesterday’s weather”) • Story points (or ideal days) + velocity duration – Velocity corrects estimation error – Accommodates developer optimism – overcomes the issue of if story points are measured based on pairs or individuals 20-Sep-07
Agile Project Management
87
Copyright © 2006 Frank Maurer. All Rights Reserved.
Comments on story points and velocity • Task effort depends on the current development context – developer experience with technology – “likeness” of task to others – availability of reusable code
• Story points is not well defined what does 1 story point really mean? – Changing velocities over time can‟t use “old” numbers
• Customers sometimes prefer estimates in hours • Velocity maps story points to person-hours available in iteration – blurs development effort for customers open & honest communication? 20-Sep-07
Agile Project Management
88
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
44
Observations from Past Planning Exercises • Effort: 2-4h for one week of work – Brainstorming user stories usually not done – Assignment of responsibilities missing
• Language for specifying requirements – Often too much IT oriented • useful for communicating with customer?
– Often to fine grained • user stories need to have business value
– Testing tasks are not user stories • Required to be done – no choice for customer • Business value?
• Interaction with customer is NOT finished 20-Sep-07
Agile Project Management
89
Copyright © 2006 Frank Maurer. All Rights Reserved.
Sprint Review Meeting Rules • • • •
2-3 hours Maximum 1 hour preparation No PowerPoint presentations Done on equipment where software was developed and tested • Presented by team to Product Owner and customers/users • Basis for planning next Sprint • Must represent potentially shippable increment of product functionality 20-Sep-07
Agile Project Management
90
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
45
Scrum Study (Mann/Maurer) • 2 year longitudinal case study • Researcher embedded in development team Average Percent Overtime Worked By Team • Overall results: Windows App 1 support and Windows App 2 Development
100.00
Scrum Introduced New Windows App Release
80.00
% Hours Overtime
60.00
Website Release
40.00 20.00 0.00
02-13-2005
01-09-2005
12-05-2004
10-31-2004
09-26-2004
08-22-2004
07-18-2004
06-13-2004
05-09-2004
04-04-2004
02-29-2004
01-25-2004
12-21-2003
11-16-2003
10-12-2003
09-07-2003
08-03-2003
06-29-2003
05-25-2003
04-20-2003
03-16-2003
02-09-2003
-20.00
01-05-2003
– Reduced overtime – Increased customer satisfaction
Week
20-Sep-07
Agile Project Management
91
Copyright © 2006 Frank Maurer. All Rights Reserved.
20-Sep-07
Agile Project Management
92
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
46
Daily Scrums (Stand-up Meetings) • • • • •
Daily 15 minute status meeting Same place and time every day Meeting room Chickens and pigs Three questions; – What have you done since last meeting? – What will you do before next meeting? – What is in your way?
• •
Impediments and Decisions
Based on Ken Schwaber‟s Certified Scrum Master course
20-Sep-07
Agile Project Management
93
Copyright © 2006 Frank Maurer. All Rights Reserved.
Based on Ken Schwaber’s Certified Scrum Master course
Chickens and Pigs • A chicken and a pig are together when the chicken says, "Let's start a restaurant!“ • The pig thinks it over and says, "What would we call this restaurant?“ • The chicken says, "Ham n' Eggs!" • The pig says, "No thanks. I'd be committed, but you'd only be involved!"
20-Sep-07
Agile Project Management
94
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
47
Benefits of the Daily Meeting • Focuses people to think about what has to be done in the short term • Puts peer pressure to see who is working to accomplish goals • Surfaces roadblocks quickly • Forces managers to not interfere with the project team
From: http://www.controlchaos.com/old-site/meeting.htm 20-Sep-07
Agile Project Management
95
Copyright © 2006 Frank Maurer. All Rights Reserved.
End-of-Sprint Review
Proceed or terminate?
Proceed: define next iteration Based on Ken Schwaber‟s Certified Scrum Master course
20-Sep-07
Agile Project Management
96
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
48
Sprint Review Meeting Rules • • • •
2-3 hours Maximum 1 hour preparation No PowerPoint presentations Done on equipment where software was developed and tested • Presented by team to Product Owner and customers/users • Basis for planning next Sprint • Must represent potentially shippable increment of product functionality 20-Sep-07
Agile Project Management
97
Copyright © 2006 Frank Maurer. All Rights Reserved.
Reporting: Tracking Progress & Metrics • Two questions – How many hours/days have you worked? – How many more does it take?
}
Which one is more important
• Project metrics – Actual time worked on a task – Work burndown graph • Per iteration • #Backlog project
– – – – – –
#Bugs #Stories completed #Acceptance tests defined and passing #Unit tests Test coverage …
20-Sep-07
Agile Project Management
98
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
49
Iteration tracking Ibid 228
20-Sep-07
Agile Project Management
99
Copyright © 2006 Frank Maurer. All Rights Reserved.
Based on Ken Schwaber’s Certified Scrum Master course
Project Tracking: Work Burndown Charts No one home
2500
2000
1500 No one home 1000
500
Underestimating
31
29
27
25
23
21
19
17
15
9
13
7
11
5
3
1
0
Overestimating
3000
1800 1600
2500
1400 2000
1200 1000
1500
Underestimating
Overestimating 800
1000
600 400
500
200
20-Sep-07
Agile Project Management
31
29
27
25
23
21
19
17
15
13
11
9
7
5
3
0
1
31
29
27
25
23
21
19
17
15
13
9
7
11
5
3
1
0
100
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
50
Progress Tracking with FDD
http://www.togethercommunity.com/coad-letter/Coad-Letter-0070.html
20-Sep-07
Agile Project Management
101
Copyright © 2006 Frank Maurer. All Rights Reserved.
Parking Lot Diagram
Pg. 201, Java Modeling in Color with UML
20-Sep-07
Agile Project Management
102
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
51
Agile Practices • • • •
Agile methods lay out a vision and then nurtures project resources to do the best possible to achieve the plan. Agile is the “art of the possible.” “better to beg forgiveness than ask permission.” Agile employs the following practices: – Frequent inspection and adaptation – Emergence of requirements, technology, and team capabilities – Self-organization and adaptation in response to what emerges • Creativity • Let the team figure out what to do and then do it
– Incremental emergence – Dealing with reality, not artifacts – Collaboration Based on Ken Schwaber‟s Certified Scrum Master course
20-Sep-07
Agile Project Management
103
Copyright © 2006 Frank Maurer. All Rights Reserved.
Scrum – Tips, Tricks, Observations • • • • • • •
Pay $1 for being late for a Scrum meeting Always deliver a vertical slice of user functionality Scrum backlog entries tend to be coarser grained than XP user stories Keep things visible in customer terms In Scrum meetings NO information is passed that is not potentially of interest for everybody Best case plan versus minimum promised to customer Sprint review & planning meeting: – –
• • • •
Planning horizon: do not overlook the big picture Process improvement entries in backlog Inter-team learning: informal meeting of Scrum masters Scrum is scalable – – –
•
1.5days initially Later 1day
Scrum of Scrums Multiple teams working together: 20% overhead Infrastructure teams (often: virtual teams): other teams are customers
Architecture diagram for reporting progress visually
20-Sep-07
Agile Project Management
104
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
52
Common Impediments • • • • • •
Workstation, network, and/or server are down; Network or server are slow; Required to attend human resource training session; Required to attend status meeting with management; Asked by management to do something else; Asked to do something other than what this team member committed to for this Sprint; • Unsure about how to proceed; • Unsure of design decision; and • Unsure how to use technology. Based on Ken Schwaber‟s Certified Scrum Master course
20-Sep-07
Agile Project Management
105
Copyright © 2006 Frank Maurer. All Rights Reserved.
Ibid p 249ff
Why does it work? • • • •
Replanning occurs frequently Separation of size and duration Plans at different levels Small cycle time (i.e small stores) keeps work flowing • Fuzzy states (e.g. 70% done) are elimnated • Tracking at the team level • Uncertainty is planned for 20-Sep-07
Agile Project Management
106
Copyright © 2006 Frank Maurer. All Rights Reserved.
Desktop Technology Program
53
Summary • Vision, release, iteration • Short horizon for detailed planning • Reporting needs to tie in with vision and business value • Adaptive and flexible • Team effort
20-Sep-07
Agile Project Management
107
Copyright © 2006 Frank Maurer. All Rights Reserved.
Discussion
[email protected] http://ebe.cpsc.ucalgary.ca/Frank.Maurer
Desktop Technology Program
54