Unformatted text preview: Chapter 5 Life-Cycle Phases From Software Project Management By Walker Royce (of IBM) And Slides on Spiral by Barry Boehm 1 ...a very short but VERY important
chapter... 2 Introduction On one hand, Do we spend far too much time on analyses and paper studies? Do we delay actually doing the builds: the development baselines? Very easy to do... Some feel in general that this is true... Do we jump into designs and coding, and hack the heck out of an application in attempts to get it to work? Lots of people think we do this too.... What do you do? What do you think corporations think they do? On the other hand, Class discuss: 3 Introductory Statement: Walker Royce feels that we need BALANCE
between research and development (R&D) and production activities Need some kind of balance between: Concentrating on capturing and modeling functionality and Building a robust product that has the performance, reliability, and scalability customers desire.... We are after a development life-cycle BALANCE...
4 Finer Granularity Further, we need a process that supports this
balance. Need this stated more precisely: Need a process to help balance: with with Planning, capturing, and modeling requirements and establishing a baseline architecture, Continuous assessment, measuring risk, and testing to ensure progress and quality Evolution and verification of the application's functionality through series of customer demonstrations and ultimate validation. 5 Engineering and Production Stages Royce claims to achieve the ROI for software
development, we need to use a `manufacturing process' that is characterized by the highest utilization of automated development tools and use of component-based approaches to development. He likens a desirable software process to a
6 Engineering and Production Stages He breaks all activities down into Engineering and Production Engineering work: This centers on risk reduction, prototyping, establishing architectural baseline, assessment, analysis, design, and planning... Implies a smaller team up front. Production work: programming and unit test, system and integration testing, demonstrations, assessment, base-lining (alpha, beta, ...) configuration, and releases; operations Note that production includes operations...
7 Two Stages: Far Too Abstract BUT, he argues a life cycle of two stages is far too abstract to track the many
detailed activities all with actors, activities, and artifacts.... So, he maps RUP Phases into these more comprehensive phases. Enter: Engineering Phase = Inception and Elaboration Enter: Production Phase = Construction and Transition To wit: Engineering, i.e., Inception and Elaboration, focuses on the concept (idea)
of the application and its architectural components (analysis and preliminary design perhaps a wee bit of detailed design) Artifacts are established and base-lined; (Configurations...) Production, i.e., Construction and Transition, focuses on programming, testing, releases and converting / establishing operational capabilities Implies that artifacts from earlier stage (engineering) more difficult to change as activities more `downstream activities' occur
8 Royce Claims that: These phases can be mapped into the famous
Spiral Model for Software Development developed by (Barry Boehm) (shall see ahead) Now have: Conventional Software Development Model (as
represented by the Waterfall Model and its many variants; OOSE approach (as represented by the RUP); Spiral Model...Let's discuss this important model...
9 Spiral Model - Overview Spiral Model is another incremental
model. Embraces (well known for these:) prototyping, iterative software development, and risk assessment. Model is graphed like a spiral. Development can be halted at the end of any cycle...depending on evaluation of previous `cycle.' IS ROI still looking good? Are expended costs in line with anticipated costs so far? Have risks been mitigated? Functionality delivered evolving properly via high 10 Spiral Model Very much a Risks-Driven Approach Different idea of software development. How does this project affect the developers
and the clients? How does each step in the project affect its overall development? Not used in previous development models. They were usually code-driven or document-driven.
11 An evolution of models Previous Software Process Models
Code & Fix Stagewise & Waterfall Evolutionary Development Others... Code & Fix First, elementary model Write code now; fix it later No planning involved Problems: Code is poorly structured. The software developed was usually a poor match for users' needs.
12 Stagewise & Waterfall
Born out of the shortsightedness of the Code & Fix model. - need for a design phase, requirements phase,
and a testing phase. First used to develop SAGE (Semi-Automated
Ground Environment), an early warning system for the Cold War era.
13 Stagewise A development process of successive phases. Phases included operational plan, operational specs, coding specs, coding, parameter testing, assembly testing, shakedown, system evaluation. Underwent two refinements in 1970. Now referred to as the Waterfall Model. 14 Waterfall Model
Introduced: Feedback loops across multiple stages: Validation and verification steps. Prototyping via a "build it twice" step alongside of requirements and design. Difficulties exposed even as revisions were made to the model. Required elaborated documents. (Documentdriven; lengthy development cycles, etc. Led to pursuing stages of development in the wrong order
15 16 Evolutionary Development
Evolution of the system in directions based on experience.
Provides rapid initial operational capability. "I can't tell you what I want, but I'll know it when I see it." Flexible, yet uncertain approach. Evolutionary Development Problems:
No formal design phase (same problem as Code & Fix). One bad assumption the unplanned paths "will" be compatible. Hard-to-change code resulted. Many problems when new software was incrementally 17 replacing old software 18 19 Spiral Model - Overview In the Spiral Model, prototyping, evaluation, planning, and all engineering
and production activities are executed in different quadrants. (Different variations of the model...) But the basic notion of iteration is very firmly established. For each `cycle': risk is assessed, more design and development is undertaken, the work products and evaluated, and planning for the next cycle in the spiral is undertaken; Iterate..... Please note that in reality, the Spiral Model curve is actually
skewed to the right, as the spirals do not `carve' out equal `areas. Area in each cycle is a determinant of effort and cost.... 20 Spiral Model - Overview Quadrants in a cycle: Creation of a prototype as a means to gather and lock in requirements; gain customer buy-in. Risk is assessed and if acceptable, Development activities then follow using the waterfall model Specifications are created from the prototyping effort, Then Requirements, Analysis, Design, Implementation, ... ensue Review and release are undertaken... Restart and iterate as above... Planning for next iteration is undertaken or not... Iterate until application is developed fully and totally released to the clients. 21 Spiral Model - more May be several cycles of prototyping...but as prototypes
evolve, these may become official releases of the product (internal or external). Before a cycle ends, the review discusses experiences, assesses risk, and decision made whether or not to proceed. Note: for each cycle, FIRST thing before embarking is to decide what are the major requirements to be handled. Adjust the architecture and requirements and/or project plan as needed. The Spiral forms the basis for entire life cycle of the product. Thus, the Spiral Model continues the spiral process for
22 Maintenance, and the model continues until the application is ultimately retired or replaced... Spiral Model and the RUP Phases
R & D Stage Inception Elaboration Production Stage Construction Transition Idea Architecture Beta Releases Products Prototypes Coarse artifacts Major risk items Creative, judgment Business Rules Stakeholder Vision Change managed baselines Elaborate artifacts Low Risk Items Engineering, reasoned Well-instrumented23 processes `Skewness: Back to our Process Model (RUP) Let's look more closely at the phases for our process
model: Inception Elaboration Construction Transition Each phase has primary objectives, essential activities, and primary evaluation criteria to judge its success at milestone time. Since the process that is underpinning our management of software processes and personnel course is the RUP, it is imperative that we understand this management 24 process, the RUP, as much as possible Inception Phase (1 of 3) Overriding Goal: achieve concurrence among
stakeholders on life-cycle objectives for the project (milestone: LCO) Primary Objectives: (by end of phase...) Establish project software scope and boundary conditions Includes operational concept, acceptance criteria, and a clear
understanding of what is and what is not intended in the product Know These! Identify critical use cases (core functionalities) of system and the primary scenarios that will drive the activities Demonstrate at least one candidate architecture against some of the primary scenarios (walk through it...) Estimate the cost and schedule for the entire project (including detailed estimates for the elaboration phase) Estimate potential risks (sources of unpredictability)
25 Inception Phase (2 of 3) Essential Activities Develop Project Scope Capture requirements and concept of operations repository Describes users' view of the requirements Repository contains information used to define problem space Repository must contain information to capture acceptance criteria Develop a Candidate Architecture and Demonstrate it Repository contains enough information to demonstrate at least a
single candidate architecture This might be at a very high level, such as deployment level; A general layered architecture may also be demonstrated. But considerable Requirements have not yet been done and almost NO Analysis and Design have been undertaken. Repository must include enough data to support make/buy decisions so that cost, schedule, resources can be `costed out.' Planning and Preparing the Business Case Risk management strategies, staffing, general iteration plans, cost/
schedule/profitability tradeoffs are all evaluated. Environmental (Infrastructure) support is defined. ROI, market share, etc.
26 Inception Phase (3 of 3) Primary Evaluation Criteria Stakeholder concurrence on scope definition and cost/schedule estimates? Do critical use cases demonstrate that the requirements are understood? Do we have credible estimates for cost/schedule/resources/risks/development? Does the architectural prototype support previous items? (Does the prototype indicate that the scope of project is understood, and does the development group indicate prevalent understanding?) Are actual expenditures verses planned expenditures acceptable? <<end inception>>
27 LCO Milestone... Elaboration Phase (1 of 5) Clearly the most important phase! Overriding
goals are several, varied, and critical! At end of phase: engineering is complete, almost all use cases are designed (certainly all critical use cases and flows), a prototype for gathering requirements and to demonstrate proof of concept is accommodated, and an analysis model is constructed, and a baseline executable architecture is established and demonstrated. Risks have to have been addressed and strategies understood; Business Rules have been subscribed to closely; Cost and schedule are acceptable and predictable and updated, if necessary, Stakeholder acceptance is achieved (the vision is realized in the artifacts), etc. and We have stability... 28 We want to graduate from a low-cost effort into a full-blown production process, where costs are maxed and personnel are on staff. Elaboration Phase (2 of 5) Primary Objectives (at end of phase...) Base-lining the architecture asap (establishing configuration management procedures...for tracking all artifacts! Base-lining the Vision. It is now `solid' and accommodated in the artifacts so far. Base-lining a detailed plan for Construction Demonstrating the baseline architecture such that it clearly supports the vision at, of course, reasonable cost in reasonably time.
29 Elaboration Phase (3 of 5) Essential Activities Detail the Vision. Ensure all have this shared functional vision and that this is
reflected in the Use Cases that will drive the architecture and planning.... Discuss: What does this mean to you??? Explain! Detail the process (to come) and the infrastructure support. Must be spelled out; the plans for each iteration in Construction (project management supporting discipline) and the anticipated assessment at the end of each iteration, the functionality accommodated by each iteration must be spelled out in general. Detailed iteration planning (after first iteration or two) will come later. But overview planning is now! Discuss: What does this mean? Explain! 30 Elaboration Phase (4 of 5) Essential Activities continued Build the Architecture... Group classes into packages; subsystems; consider
existing executable components that may be reusable e.g. GUI Components (commonly available); can you reverse engineer existing components? (Should you seek to buy components? Contract for them? Develop them?) But you MUST integrate these into architectural units (layers, packages, subsystems, all with dependencies, well-defined interfaces, ...) Bounce your candidate architecture against your primary scenarios to trace that all functionality is accommodated by these artifacts. May result in a number of design choices and changes in model elements (e.g. classes, responsibilities...) May point out some requirements missing...
31 Remember, requirements are singular; but there is not a
single, perfect design. Elaboration Phase (5 of 5) Primary Evaluation Criteria Remember, at the end of Elaboration Phase we have the LCA Life Cycle Architecture Milestone. So, ... Is the Vision solid? Stable? Is the architecture stable? Demonstratable? Execute example of addressing a high risk scenario? Does the architecture indicated that all risk elements have been addressed/mitigated? Have we looked carefully into Construction and established sufficient planning detail to project credibility in our estimates? (initial iteration one or two carefully planned?) Do we have stakeholder buy-in that their vision can be accommodated if we proceed as plans indicate? Are actual resource expenditures verses planned resource expenditures acceptable so far? Achieve this Milestone! Press on with concurrence.32 Construction Phase (1 of 5) Great mindset change: now interested in
producing a deployable product!
(implementation) iterate, iterate..., integrate/assess/plan as we go. No longer `engineering;' rather, production!!! Need to manage resources, control operations to
optimize costs, schedules, and quality. Emphasis on the development of intellectual property shifts to the reality of usable products. 33 Construction Phase (2 of 5) One very nice attribute in Construction: Parallel development Based on architecture...do homework up front!!!! Accelerates delivery of deployable releases Downside: complicates project management and synchronization of teams, integration, and workflow. Architecture will drive this A good architecture will support parallel development Emphasized during Elaboration planning for
Construction. 34 Construction Phase (3 of 5) Primary Objectives Develop the system rapidly but with high quality that is, construction (programming and unit testing) `implements' the design; `realize' the design... Take advantage of the process, versioning, reviews, assessment, etc. to minimize costs due to needless rework and scrap. Develop alpha, beta, `or what have you' releases for Transition phase.
35 Construction Phase (4 of 5) Essential Activities Manage resources; control development (via plans, configuration management, change management, ...); Perform unit testing (component testing) against requirements (verification) Assess releases against acceptance criteria cited in vision. (validation) V&V ...Discuss....
36 Construction Phase (5 of 5) Primary Evaluation Criteria (at end) Milestone is `Initial Operational Capability' (IOC) Is the product reliable enough for deployment? Does not mean everything must be perfect...Showstoppers? Does it fail frequently?? Is product `stable' enough for deployment? Pending changes are okay But are we getting change requests a-plenty? Are defects being identified rapidly `as we speak?' How significant are the changes?? Is the stakeholder community ready to transition? Are actual expenditures reasonably close to planned expenditures? 37 Transition Phase Recall end of phase milestone: product release to
user domain. Implies product is stable, has high quality, has accompanying user documentation (on-site or web-based training...), customer support is `ready', etc... Phase `could' include any of these... Beta testing to validate new system against expectations Beta testing in parallel with legacy system to be replaced Installation Conversion of operational data bases Training users, maintenance team, customer support...
38 Transition Phase Phase concludes when the baseline realizes the
original vision and we are ready to put it in the users' hands. Might be end of project development or starting point for next cycle, or starting point for next version of deployable version... Might be forwarding `whole shooting match' over to the maintenance group or third party for future work...
39 Transition Phase Transition is not uncomplicated May involve several iterations including Beta1, beta2, ... testing and `levels' of releases (all releases may not be equal...) Custom software? Conversion software? Development of user documentation, User training, especially in initial use of product Web-X; on developer's site; on client's site. Who pays for what? How does this work? Millions!!! Usability problems and tuning, (Un)solicited feedback, and more....
40 Transition Phase Essential Activities "Synchronization and integration of concurrent construction increments into consistent deployable baselines" ensure all flows... Installation / Conversion Cut over (complete switch to new application) Run in parallel, or Phased... Assessment against vision and acceptance criteria. Is user satisfied? Are actual expenditures reasonably close to planned expenditures?
41 Evaluation Criteria Summary - Know These Recognize each phase has one or more
iterations Phases end with major milestones; (Know these!) Iterations within phases have minor milestones. Each have deliverable(s) and undergoes assessment against criteria Each iteration entails a sequence of activities that culminate in a minor milestones or major milestones (if iteration ends phase) Scope and results of iterations are captured via artifacts produced. 42 Summary (continued) Know! Major Milestones (phase end): Approved by stakeholders Map to significant management/business decisions rather than to completion of a specific software development activity. Minor milestones (iteration end): Approved internally and Realized by artifacts / new versions of artifacts in repository; internal synchronization, internal assessment, Additional planning take place... `Executable' releases (not necessary deployable...) 43 Lessons Learned Organizational Change Middle management is where the war is won ROI on first implementation Championed by respected leaders who own the plan and execution. Project Management: Can be immense pressures from above (Sr. Management) and different sets of problems/issues from below (actual developers) Disruption costs must be absorbable in the benefit This is where the A-players are Implemented on business critical project Success breeds success (like the NFL) Results drive incentives First increment needs to be ambitious, but realistic Such as: milestone demonstrations, release timeliness, release content, etc 44 Not: processes, methods, expended energy, reuse, audits, meetings, subjective assessments, document production,... ...
View Full Document
- Spring '09
- Software development process, Elaboration phase, Walker Royce