Waterfall Model – Revisited• Disadvantages of Waterfall Model – 1. Real projects are rarely so straightforward and sequential – 2. It is generally not possible to completely define (and freeze) all the requirements at the start of the project – 3. Problem is discovered in testing? – 4. Freight-Train Effect, or Late, or Over-Budget
What is “Wicked Problem”• Problems we can’t really understand until we’ve developed a solution.• “That is not what I want ... but now I know what I do want!”
The Mythical Man Month- Dr. Frederick Brooks• In software projects, what will take one person ten months can not be solved by ten people in one month.• Throwing people onto a late project will just make it later• Because of Wicked Problems, “Plan to the throw one away”
Rapid Prototyping• Put together a team of “Smart Guys” from multiple disciplines• Develop the GUI on Paper• Code the GUI in a fast language (Make it look like it’s working) <=Requirements=> **<=Prototype=>**• Show it to the USERS (A Picture is worth a 1,000 words) <=Design=> <=Code=>• Get Feedback <=Test=> <=Deploy=>
Case Study- RAD Grand Community Calendar Project– Project Manager, Developer, Community Members write user requirements– Coder writes sample HTML– Shows the web page; heads bob, some changes to navigation– DBA, Coder, Project Manager determine the architecture (Design)– Coding & Review– Shifting Requirements priced project out-of- budget
Problems With Prototyping• No “Current” Documents• Functional Spec is Prototype + Feedback• Prototype is not “baseline” functionality• Same problems with Functional Spec as waterfall!
Prototyping Part II: The Rigged Demo• Re-Visit and improve the prototype to serve Listen To as a “baseline” Customer• Turns prototype into a “rigged demo” Build/Revise Mockup Customer Test• Show that to the Drives Mockup customer
At the Demo Dialogue• Customer:“This looks great, and it looks like you’re about done. When can we have it?”• Developer: “Uh, it’s only a prototype – we plan to throw it away and start over.”• Customer: “No – this is exactly what we need, and we need it now! We’ll take 50 prototypes!”– The Sales Guy begins to see $$ signs.– Under Rigged Demo scenarios, there is either a lot of wasted effort, or prototypes that were never intended to ship end up shoved into production.
Case Studies Multi-Stage Prototyping• Telecommunication – The prototype made the sale! – Was pushed into production – From user requirements to “Ship”ing in 4 month – Errors, Bugs, High Turn-Over – Had to support bug fixes plus “incremental” change• Visual Product Explorer – Prototype created for internal consumption – Feedback Cycle – Modified for trade demo – Next step: How do we write the spec? – Product is the spec; shove it into production!
Iterative Models What’s an Iteration?• Iterative Design: Code as much as you can questions surface, then start over.• Every model we’ll talk about below is a variation on the Iterative Model.
Spiral ModelDetermine Evaluateobjectives, alternatives,alternatives, identify andconstraints resolve risksPlan next Develop verifyphases next level product
Risk Assessment• Spiral Model – risk driven rather than document driven• The "risk" inherent in an activity is a measure of the uncertainty of the outcome of that activity• High-risk activities cause schedule and cost overruns• Risk is related to the amount and quality of available information. The less information, the higher the risk• What happened with Denver Airport Luggage System?
Spiral Model Strength and Weaknesses• Strengths – Introduces risk management – Prototyping controls costs – Evolutionary development – Release builds for beta testing – Marketing advantage• Weaknesses – Lack of risk management experience – Lack of milestones – Management is dubious of spiral process – Change in Management – Prototype Vs Production
Win Win Spiral Model• Win-Win Spiral Process Model is a model of a process based on Theory W, which is a management theory and approach "based on making winners of all of the systems key stakeholders as a necessary and sufficient condition for project success."
WinWin Spiral Model••Identify Nextproduct & process definitions ••Validate commitment holders process Identify Stake holders win conditions Evaluate Product of Process and Reconcile Level Stake Define next level & product Alternatives Review, Win conditions
Win Win Spiral Cont• Identifying the systems stakeholders and their win conditions and• reconciling win conditions through negotiation to arrive at a mutually satisfactory set of objectives, constraints, and alternatives for the next level.• Evaluate Product and Process Alternatives. Resolve Risks• Define next level of product and process - including partitions• Validate Product and Process Definitions• Review, commitment
WinWin Spiral- Anchor Points• Life Cycle Objective(LCO) – What should the system accomplish?• Life Cycle Architecture(LCA) – What is the structure of the system?• Initial Operational Capability(IOC) – The first released version
Key Elements of IOC Milestone• Software preparation – Including both operational and support software with appropriate commentary and documentation – data preparation or conversion – the necessary licenses and rights• Site preparation – including facilities, equipment, supplies and vendor support• User, Operator and Maintenance preparation – including selection – team building – training
Win Win Spiral - Case Study • Extending USC Integrated Library System to access multimedia – Flexibility and Discipline let the projects teams adapt to challenges while staying on schedule – Use of risk management helped team focus on CSF for their projects – One cycle for each milestone – Communication and trust between stakeholders, shared vision – Don’t finish negotiations before prototyping – Client acceptance
Another Extreme CleanRoom Methodologies• From Hardware Cleanrooms• An incremental process that encourages continuous improvement;• Technical reviews that prevent defects and significantly reduce costs• Design and coding practices that make it easy to adapt as requirements change• Testing techniques that focus on measuring quality;• Solution-oriented teams that encourage cooperation, reduce the dependence on "gurus," and promote flexibility• Documentation structures that reveal the big picture and help team members maintain intellectual control.
Clean Room Continued• REAL Peer Review Mathematical proof of correctness (Challenges associated with it?)• Functional Specifications as Box Diagrams (State, Black, Clear)
Yet Another Extreme: Hacking• Hacking: – Code ‘n Fix – More Common than you thought• Makes Sense for: – Low-Risk, Small Project – We know exactly what we want (not “Wicked”) – Use once, then throw away – Bugs can be tolerated/fixed• Problem: – “Why not just re-use Hack X here with change Y …” – Hack Code is hard to maintain, but appealing from a management perspective.• Case Study: – I’m guessing … just about every project you ever did as an undergraduate.
Summary Summary• Waterfall – good for budgeting, but doesn’t analyze risk or have a good way to manage errors found later in the process.• Iterative – Models attempt to solve this by coding “as far as possible”, gathering feedback, and coding again.. – Prototyping “Plan to throw one away”, then re-build it “right.” – Incremental (“Staged”) Delivery Builds the software by a series of waterfalls
Summary• Spiral: – Addresses Risk at every stage & let the stakeholders determine the outcome.• Win/Win – Seeks ways to provide customer feedback through anchor points, manages risk for management, and provides win conditions for developers.• Cleanroom / Hacking – Are alternative models that work for large projects that must work right the first time, and small projects with little risk.
Resources• Generally Interesting Theories for REAL-WORLD Development:• Wicked Problems/State of Coding: – http://www.unidata.ucar.edu/staff/caron/collab/wicked.html – http://www.chc-3.com/pub/beautifulsoftware.htm • Mythical Man Month – ( http://www.amazon.com/exec/obidos/ASIN/0201835959/ref=bxgy_sr_text_a )• Code Complete – ( http://www.amazon.com/exec/obidos/ASIN/1556154844/ref=bxgy_sr_text_a )• Joel Spolsky on Real-World Software Development – http://www.joelonsoftware.com• Software Engineering, A Practitioner’s Approach – http://www.mhhe.com/engcs/compsci/pressman/
Resources (2)• Spiral Model – Using the WinWin Spiral Model: A case study, Boehm Barry, July 1998, Computer• Spiral Development workshop – www.sei.cmu.edu/cbs/spiral2000/february2000/BoehmSR.html• Anchoring the Software Process, Boehm Barry – http://www.csis.gvsu.edu/~ferguson/classes/cs641/papers/ASP.pdf• Denver Airport Project – http://www.time.com/time/magazine/archive/1994/940516/940516.tr ansportation.html• Cleanroom Model – http://www.cleansoft.com/cleansoft_mgrguide.html – http://www.sei.cmu.edu/pub/documents/96.reports/pdf/tr022.96.pdf• Hacking – http://www.plethora.net/~seebs/faqs/hacker.html
Homework• Objective Question – One major difference between the Waterfall and iterative models is that the iterative models address risk. How do they do that?• Subjective Question – Which of these models is the best for the Customer? The Seller? Why?