Software design methodologies

Software Life Cycle Models 1950s Code & Fix 1960s Design-Code-Test-Maintain 1970s Waterfall Model 1980s Spiral Model 1990s R apid A pplication D evelopment 2000s Agile Methods

Development Methodologies (1/2) Agile software development Agile Unified Process (AUP) Open Unified Process Best practice Cathedral and the Bazaar, Open source Constructionist design methodology (CDM) Cowboy coding Design by Use (DBU) Design-driven development (D3) Don't repeat yourself (DRY) or Once and Only Once (O3) Dynamic Systems Development Method (DSDM) Extreme Programming (XP)

Development Methodologies (2/2) Test-driven development (TDD) Unified Process Waterfall model Worse is better (New Jersey style) Extreme Programming (XP) You Ain't Gonna Need It (YAGNI) Iterative and incremental development KISS principle (Keep It Simple, Stupid) MIT approach Quick-and-dirty Rational Unified Process (RUP) Scrum (management) Spiral model Software Scouting

Code-and-Fix “ Code-like-Hell” Specification (maybe), Code (yes), Release (maybe) Advantages No overhead Requires little expertise Disadvantages No process, quality control, etc. Highly risky Suitable for prototypes or throwaways

Activities by % of Total Effort NASA’s “Manager’s Handbook for Software Development”

Potential Deliverables by Phase

Concept Exploration The “Why” phase Not a “mandatory formal” phase Sometimes called the “pre-project” phase Collecting project ideas Then the “funneling” process Project Justification ROI Cost-benefit analysis Project Portfolio Matrix Initial planning and estimates

Concept Exploration Possibly includes Procurement Management: RFP Process Vendor selection Contract management Gathering the initial team Including PM if not already on-board Identify the project sponsor Primary contact for approval and decision making Potential Phase Outputs: Concept Document, Product Description, Proposal, SOW, Project Charter

Concept Exploration Characteristics & Issues Lack of full commitment and leadership Some frustrations: Management only getting rough estimates from development Development not getting enough specifics from customer Finding a balanced team Budget sign-off may be your 1 st major task Achieved via: Good concept document or equivalent Demonstration of clear need (justification) Initial estimates

Requirements The “What” phase Inputs: SOW, Proposal Outputs: Requirements Document (RD) a.k.a.Requirements Specification Document (RSD) Software Requirements Specification (SRS) 1 st Project Baseline Software Project Management Plan (SPMP) Requirements Approval & Sign-Off Your most difficult task in this phase

Requirements Perhaps most important & difficult phase Shortchanging it is a ‘classic mistake’ Can begin with a Project Kickoff Meeting Can end with a Software Requirements Review (SRR) For Sponsor and/or customer(s) approval

Why are Requirements so Important?

Requirements Characteristics & Issues Conflict of interest: developer vs. customer Potential tug-of-war: Disagreement on Features & Estimates Especially in fixed-price contracts Frequent requirements changes Achieving sign-off Project planning occurs in parallel

Requirements Requirements are capabilities and condition to which the system – more broadly, the project – must conform

2 Types of Requirements Functional (behavioral) Features and capabilities Non-functional (a.k.a. “technical”) (everything else) Usability Human factors, help, documentation Reliability Failure rates, recoverability, availability Performance Response times, throughput, resource usage Supportability Maintainability, internationalization Operations: systems management, installation Interface: integration with other systems Other: legal, packaging, hardware

Requirements Other ways of categorizing Go-Ahead vs. Catch-up Relative to competition Backward-looking vs. Forward-looking Backward: address issues with previous version Forward: Anticipating future needs of customers Must be prioritized Must-have Should-have Could-have (Nice-to-have: NTH) Must be approved

Analysis & Design The “How” Phases Inputs: Requirements Document Outputs: Functional Specification Detailed Design Document User Interface Specification Data Model Prototype (can also be done with requirements) Updated Plan (improved estimates; new baseline)

Analysis & Design a.k.a. Top-level design & detailed design Continues process from RD Ends with Critical Design Review (CDR) Formal sign-off Can also include earlier Preliminary Design Review (PDR) for high level design

Development The “Do It” phase Coding & Unit testing Often overlaps Design & Integration phases To shorten the overall schedule PM needs to coordinate this

Development Other concurrent activities Design completion Integration begins Unit testing of individual components Test bed setup (environment and tools) Project plans updated Scope and Risk Management conducted

Development Characteristics Pressure increases Staffing at highest levels Often a “heads-down” operation Issues Last-minute changes Team coordination (esp. in large projects) Communication overhead Management of sub-contractors

Integration & Test Evolves from Dev. Phase Often done as 2 parallel phases Partial integration & initial test Starts with integration of modules An initial, incomplete version constructed Progressively add more components

Integration & Test Integration primarily a programmer task Test primarily a QA team task Integration: Top-down: Core functionality first, empty shells for incomplete routines (stubs) Bottom up: gradually bind low-level modules Prefer top-down generally

Integration & Test Tests Integration testing Black & White-box testing Load & Stress testing Alpha & Beta testing Acceptance testing Other activities Final budgeting; risk mgmt.; training; installation preparation; team reduced

Integration & Test Characteristics & Issues Increased pressure Overtime Customer conflicts over features Frustration over last-minute failures Budget overruns Motivation problems (such as burnout) Difficulty in customer acceptance Esp. true for fixed-price contracts

Deployment & Maintenance Installation depends on system type Web-based, CD-ROM, in-house, etc. Migration strategy How to get customers up on the system Parallel operation Deployment typically in your project plan, maintenance not

Deployment & Maintenance Maintenance Fix defects Add new features Improve performance Configuration control is very important here Documents need to be maintained also Sometimes a single team maintains multiple products

Deployment & Maintenance Characteristics & Issues Lack of enthusiasm Pressure for quick fixes Insufficient budget Too many patches Personnel turnover Regression testing is critical Preferably through automated tools

Lifecycle Planning a.k.a. Lifecycle Management or SDLC Greatly influences your chance of success Not choosing a lifecycle is a bad option Three primary lifecycle model components Phases and their order Intermediate products of each phase Reviews used in each phase

Lifecycle Planning Different projects require different approaches You do not need to know all models by name You should know how that if given a certain scenario what sort of SDLC would be appropriate There are more than covered here A lifecycle is not a design, modeling or diagramming technique The same technique (UML, DFD, etc) can be used with multiple lifecycles

Pure Waterfall The “granddaddy” of models Linear sequence of phases “Pure” model: no phases overlap Document driven All planning done up-front

Waterfall Risk Why does the waterfall model “invite risk”? Integration and testing occur at the end Often anyone’s 1 st chance to “see” the program

Pure Waterfall Works well for projects with Stable product definition Well-understood technologies Quality constraints stronger than cost & schedule Technically weak staff Provides structure Good for overseas projects

Pure Waterfall Disadvantages Not flexible Rigid march from start->finish Difficult to fully define requirements up front Can produce excessive documentation Few visible signs of progress until the end

SSADM Only covers part of the system development process, i.e. analysis and design. It emphasises the importance of the correct determination of systems requirements.

SSADM Stages Feasibility Study Stage 0 – Feasibility Requirements Analysis Stage 1 – Investigation of current requirements Stage 2 – Business Systems Options Requirements Specification Stage 3 – Definition of Requirements

SSADM Stages Logical System Specification Stage 4 – Technical System Options Stage 5 – Logical Design Physical Design Stage 6 – Physical Design

Spiral Emphasizes risk analysis & mgmt. in each phase A Series of Mini-projects Each addresses a set of “risks” Start small, explore risks, prototype, plan, repeat Early iterations are “cheapest” Number of spirals is variable Last set of steps are waterfall-like

Spiral Advantages Can be combined with other models As costs increase, risks decrease Risk orientation provides early warning Disadvantages More complex Requires more management

Evolutionary Prototyping Design most prominent parts first Usually via a visual prototype Good for situations with: Rapidly changing requirements Non-committal customer Vague problem domain Provides steady, visible progress Disadvantages Time estimation is difficult Project completion date may be unknown An excuse to do “code-and-fix”

Staged Delivery Waterfall steps through architectural design Then detailed design, code, test, deliver in stages Advantages Customers get product much sooner Tangible signs of progress sooner Problems discovered earlier Increases flexibility Reduces: status reporting overhead & estimation error Disadvantages Requires more planning (for you the PM) More releases increase effort (and possible feature creep) How’s this differ from Evolutionary Prototyping?

V Process Model Designed for testability Emphasizes Verification & Validation Variation of waterfall Strengths Encourages V&V at all phases Weaknesses Does not handle iterations Changes can be more difficult to handle Good choice for systems that require high reliability such as patient control systems

RAD Rapid Application Development Popular in the 80’s 1. Joint Requirements Planning (JRP) 2. Joint Application Design (JAD) 3. Construction Heavy use of tools: code generators Time-boxed; many prototypes 4. Cutover Good for systems with extensive user input available

XP: eXtreme Programming Not a Microsoft product Part of movement called “Agile Development” A “Lightweight” methodology A bit counter-culture Currently in vogue Motto: “Embrace Change” Highly Incremental / Iterative

eXtreme Programming Suitable for small groups Attempts to minimize unnecessary work Uses an “on-site” customer Small releases Pair programming Refactoring Stories as requirements You want good developers if you use this

Other “Agile” Methodologies Agile here means “lite”, reduced docs, highly iterative Agile Software Development SCRUM Features 30-day “Sprint” cycles Feature Driven Development (FDD) XP with more emphasis on docs and process

Other “Agile” Methodologies Adaptive Software Development (ASD) Dynamic System Development Method (DSDM) Popular in Europe Homegrown: developers often hide their “agile adventures” from management

Other “Agile” Methodologies Pros Similar to XP, can reduce process overhead Responsive to user feedback Amenable to change Cons Requires close monitoring by PM May not “scale” to large projects Often requires better quality developers

Rational Unified Process RUP From Rational Corporation “Generic” version is the Unified Process Commercial Extensive tool support (expensive) Object-oriented Incremental Newer

Rational Unified Process Develop Iteratively Manage Requirements Uses UML (Unified Modeling Language) Produces “artifacts” Use component-based architecture Visually model software Complex process A “framework” Suitable for large scale systems

Choosing Your Lifecycle Varies by project Opt for “iterative” or “incremental” How well are requirements understood? What are the risks? Is there a fixed deadline? How experienced is the team or customer?

IEEE 1074 A standard for developing software processes Lifecycle model selection Project management process Predevelopment processes Development processes Post-development processes Integral process

Use cases diagram UML 2 Use cases diagrams describes the behavior of the target system from an external point of view. Use cases describe "the meat" of the actual requirements. Use cases . A use case describes a sequence of actions that provide something of measurable value to an actor and is drawn as a horizontal ellipse. Actors . An actor is a person, organization, or external system that plays a role in one or more interactions with your system. Actors are drawn as stick figures. Associations . Associations between actors and use cases are indicated by solid lines. An association exists whenever an actor is involved with an interaction described by a use case.

Class diagram UML class diagrams show the classes of the system, their inter-relationships, and the operations and attributes of the classes Explore domain concepts in the form of a domain model Analyze requirements in the form of a conceptual/analysis model Depict the detailed design of object-oriented or object-based software

Software design methodologies

In this document