Transcript Methodology for Information Systems Project Management
Methodology for Information Systems Project Management
Chapter 4: James R. Burns
Some PM humor
A badly planned project will take three times longer than expected – a well planned project only twice as long as expected. The more you plan the luckier you get.
At the heart of every large project are many small projects trying to get out.
The nice thing about not planning is that
failure
comes as a complete surprise rather than being preceded by a period of worry and depression.
The Real Software Development Process
Out of jest and with tongue firmly in cheek, someone suggested the following software development process: 1. Order the T-shirts for the development team 2. Announce availability of the product (This helps to motivate the team.) 3. Write the code (Let’s get with it!) 4. Write the manual 5. Hire the project manager 6. Spec the software (Writing the specs after the code helps to ensure that the software meets the specifications.) 7. Ship 8. Test (The customers are a big help with this step.) 9. Identify bugs as potential enhancements 10. Announce the upgrade program
Summary of Today’s lecture
What do we mean by methodology?
What are some ‘typical’ IT Project Types?
The Generic Lifecycle model
Introduction: The Methodology Choice Becomes the Process
Indigenous to every project are its processes: the methodologies by which the project is brought to fruition METHODOLOGY becomes the PROCESS • When implementation and execution of the methodology begins In this chapter we discuss the major phases of each project type We do not give you the entire WBS--just the top level
What are some IT Project Types??
Enterprise Resource Planning Enterprise Architecture Determination Re-engineering projects Rapid application development projects Product selection Component configuration projects Conversion projects Maintenance projects Component integration projects Internet Development projects Client/server changeover projects
A Generic Project Management Process
Conceptualization and Definition --
Initiating
Planning and Budgeting --
Planning
Execution --
Executing
Termination and Closure --
Closing
Monitoring and Control --
Monitoring and Controlling
STAGE 1: Initiating STAGE 2: Planning STAGE 3: Executing STAGE 5: Closing STAGE 4: Monitoring-and-Controlling
These are stages of the meta project
Three of these stages must be funded from sunk costs. In some cases “seed money” can be obtained to enable thorough completion of stages one and two Its only at stage three that projects formally ramp up and begin spending money from an authorized source
Alternative names for Stages
Inception Lifecycle Objective Milestone Elaboration Lifecycle Architecture Milestone Construction Initial Operational Capability Milestone Transition Product Release Milestone
Initiating
Obtain user requirements Worth doing?
Ensure fit Assess consistency Identify dependencies Assess risk Determine owners/stakeholders Test alignment with strategies Test resource availability Define scope Make go/no go decision Obtain funding Review alternatives
Planning
Project team is determined Resources are negotiated Team is formed, stormed, normed and ready to perform Formal Project plans are determined Formal Budgets are prepared
Executing
(Implementation)
Where the project ramps up And begins to fulfill its phases Likened to execution in a sports activity Produces the project product Change management becomes especially important here
Closing
Document lessons learned History database Postmortem meeting Signature signoffs Get paid
Software Development Process Models--Boehm
WE’RE TALKING ABOUT THE
EXECUTING
STAGE HERE Code-and-fix model Waterfall model Evolutionary model Transform model Spiral model
Code-and-fix model
Write some code Then think about requirements, design, test and maintenance later After a number of fixes, code was poorly structured Used well before structured programming was invented
Waterfall model
Definition If DEFINITION takes 2 months, then the project is roughly ___ long Analysis Design Construction Test Operation Maintenance
Definition of Requirements Design
The Waterfall Staircase
System Acceptance Testing Implementation Operation
Waterfall model
Required in all government software contracting Document-driven Good for developing something new and complex--a new AI inference engine, a new compiler, a new database engine PROBLEMS: Expensive and time consuming because of its dependence on fully elaborated documents,
curtails testing until near the end
The Standard Waterfall Model
Better than the old “CODE AND FIX” approach A manufacturing model for software CASE tools were developed to support it: TI’s IEF—sold in 1994 to Sterling Software Basis for most software acquisition standards in government
The Standard Waterfall Model, Cont’d
Too expensive for small projects Still used for project estimation (cost and duration) of large projects Doesn’t get used in practice very often Doesn’t seem to work satisfactorily
Evolutionary Process Model
Better than WATERFALL for interactive end-user software development Ideally matched to fourth-generation language applications Also works well when users say “I can’t tell you what I want, but I’ll know it when I see it.” Gives users a rapid initial operational prototype that they can play with and react to, even improve upon
The Evolutionary Process Model
NOT MUCH MORE THAN OLD “CODE AND FIX” APPROACH Assumes
user’s operational system will be flexible enough to accommodate unplanned evolution paths
.
The above assumption is bad, because frequently, several independently evolved applications must now be integrated Temporary work-arounds for software deficiencies increasingly solidify into unchangeable constraints on evolution
Transform model
Endeavors to address the difficulties associated with the code-and-fix, waterfall, and evolutionary models.
Assumes the existence of a software module that can “parse, interpret and compile” written specifications automatically into machine code There is no 3GL code (COBOL, Fortran) or 4GL code (visual languages) The specifications are transformed directly to machine code
The Transform Model
Steps are: Formal specification of the desired product Automatic transformation of the specification into code An iterative loop is necessary to improve the performance Exercise of the resulting product, and An outer iterative loop to adjust the specification based on the resulting operational experience
The Transform Model, Cont’d
Avoids the difficulty of having to modify code that has become poorly structured through repeated re-optimization, since the modifications are made to the specification Avoids the extra time and expense involved in intermediate design, code and test activities
A Pervasive Problem
How to enable users to create their own apps After all, there just aren’t enough developers around Ten years ago, the Air Force said it needed 1 out of every 5 HS graduates just to maintain its codes SOLUTION: Let users write their own specifications and then transform those same specifications directly to code
Did it work?
No! Specifications must be carefully written in a compiler-compatible format- the specifications become the program. Users have to learn a specification language--same as learning a 3GL.
Like the evolutionary model, it did not accommodate unplanned evolutionary paths well.
Won’t work for Large-Scale Development
The Spiral Model (Boehm, 1988)
Overcomes most of these shortcomings and addresses these questions What shall we do next? How long shall we continue doing it?
Radial dimension represents the cumulative cost incurred in accomplishing the steps to date Angular dimension represents the progress made in completing each cycle of the spiral.
Spiral model
Can accommodate most previous development methodologies as special cases Is a risk-driven methodology, not a document-driven one Endeavors to depict both the passage of time and the accumulation of expenditure
Spiral model, Continued
Radial dimension represents the cumulative cost incurred in accomplishing the steps to date Angular dimension represents the progress made in completing each cycle of the spiral.
The methodology is dynamic and dependent upon the relative risks remaining
Each Loop of the Spiral model
Identifies the objectives of the product being elaborated Identifies the alternatives to implementation Determines the constraints imposed on the alternatives Evaluates the alternatives relative to the objectives and the constraints Dynamically chooses methodological detail needed to alleviate perceived sources of risk
How is risk mitigated?
If there is risk associated with the specification of the product, then reference checking, administering user questionnaires, analytic modeling, or combinations of these should be used to mitigate the risk If there is risk associated with the ultimate design of the product, then prototyping, simulation, or benchmarking should be utilized to mitigate this risk.
How is risk mitigated?
If user-interface risks strongly dominate product considerations, or there are internal interface control risks, then the next step may be to use evolutionary development
Figure 2-3. Spiral Model of Software Dev.
Advantages of Spiral Model
Applies to maintenance as well as to development Incorporates prototyping as a risk-reduction option at any stage Accommodates reworks or go-backs to earlier stages Focuses early attention on reuse of existing software
More Advantages of the Spiral Model
Accommodates preparation for life-cycle evolution, growth Provides a mechanism for incorporating software quality objectives Focuses on eliminating errors and unattractive alternatives early More adaptable to projects other than development than the waterfall model, which applies only to development
More Advantages of the Spiral Model
Provides a viable framework for integrating hardware-software system development Accommodates CASE tools
Disadvantages of the Spiral Model
Doesn’t work with fixed-price contracts well Use by outside contractors and system integrators is difficult, therefore Competitive front-end contracts and level-of-effort and award-fee contracts work better Relies on project professionals with risk assessment expertise If there aren’t any, it doesn’t work well
The Dimensions
Waterfall High Ceremony Low Ceremony Iterative
Process Map
Waterfall
Few risks, late integration and testing
High Ceremony
Heavy Doc, heavy process discipline, CCB
Low Ceremony
Little Doc, light process discipline
Iterative
Risk Driven, Continuous integration and testing
Agile Software Development
Software is developed in increments using an iterative approach Backbone first User interfaces next Important functionality next Less important functionality last Learning takes place all along the way Important components may be improved before less important components are even started Provides the user with an early experience with the software.
Endeavors to deliver business value early.
More Agile Software Development
An iteration lasts one to four weeks
Each iteration passes through a full software development cycle including planning, requirements analysis, design, coding, testing, and documentation. The goal is to have an available release (without bugs) at the end of each iteration.
At the end of each iteration, the team re evaluates project priorities.
Agile methods emphasize face-to-face communication over written documentation.
Principles Behind the Agile
Manifesto
Customer satisfaction by rapid, continuous delivery of useful software.
Working software is delivered frequently (weeks rather than months).
Working software is the principal measure of progress.
Even late changes in requirements are welcomed.
Close, daily cooperation between business people and developers is strongly encouraged.
Face-to-face conversation is the best form of communication.
More Principles behind Agile Development
Projects are built around motivated individuals, who should be trusted.
Continuous attention to technical excellence and good design is required.
Simplicity is a hallmark.
Self organizing teams are always used.
Regular adaptation to changing circumstances is accommodated.
Iterative, Agile Processes
Rational Unified Process (RUP) Dynamic Systems Development Method (DSDM) Extreme Programming (XP) Crystal
Scrum
The Home Ground for Agile Software
A culture that thrives on chaos, product is of low criticality, a small number of senior developers are used, and requirements change very often, applications are small.
The Home Ground for Waterfall
For plan-driven methods (such as the Waterfall Model), the home ground is high criticality, requirements are well known and don't change too often, a large number of developers, and a culture that demands discipline and order.
Project Management & SCRUM
SCRUM is a type of agile software development, along with extreme programming (1996) , Crystal Clear, Adaptive Software Development, Feature Driven Development, and Dynamic Systems Development Method (DSDM) (1995).
More SCRUM
Scrum is an agile development methodology, implying low ceremony (little documentation, face to-face meetings).
Scrum is a process skeleton that includes a set of practices and predefined roles. There are two types of roles used in connection with Scrum, those who are committed are called ‘pigs’ and those who are involved who are called ‘chickens.’ Stakeholders are considered chickens whereas the project team and Scrum master (project manager) are called ‘pigs.’
Still More SCRUM
Scrum consists of a series of sprints. Each sprint is a period of 15 to 30 days, during which the team creates a usable module of software. Scrum is considered ‘easy to learn’ and doesn’t require a lot of training to start using it.
Sprint periods of 30 days are similar to the monthly time-boxes used in RAD.
SCRUM Meetings
Each day during the sprint, a project status meeting occurs. This is called a SCRUM Meeting. The procedure for a SCRUM Meeting is the following : 1) the meeting starts precisely on time with team decided punishments for tardiness 2) all are welcome, but only “pigs” may speak 3) the meeting is time-boxed at fifteen minutes regardless of the team’s size 4) all attendees should stand 5) the meeting should happen at the same location and same time every day
Traditional PM versus Agile Methods Traditional PM Approach Concentrates on thorough, upfront planning of the entire project.
Requires a high degree of predictability to be effective.
Agile Project Management (Agile PM) Relies on incremental, iterative development cycles to complete less-predictable projects.
Is ideal for exploratory projects in which requirements need to be discovered and new technology tested.
17–53
Some Agile definitions
Stories, story points
—these are features, portions of functionality
Elevation
—an iteration that does not result in added functionality, does not usually get released to the customer
Scrum meeting
—held daily, usually at the beginning of the day—for 15 min
Sprint—an iteration --usually 1 to 4 weeks in duration, resulting in the delivery of something of value to the customer—a small, short ‘project’
Project Uncertainty
FIGURE 17.1
17–55
Traditional Project Management versus Agile Project Management
Traditional
Design up front Fixed scope Deliverables Freeze design as early as possible Low uncertainty Avoid change Low customer interaction Conventional project teams
Agile
Continuous design Flexible scope, specifically Features/requirements Freeze design as late as possible High uncertainty Embrace change High customer interaction Self-organized project teams
TABLE 17.1
17–56
Agile Project Management
Agile PM Is related to the rolling wave planning and scheduling project methodology.
Uses iterations (“time boxes”) to develop a workable product that satisfies the customer and other key stakeholders.
Stakeholders and customers review progress and re-evaluate priorities to ensure alignment with customer needs and company goals.
Adjustments are made and a different iterative cycle begins that subsumes the work of the previous iterations and adds new capabilities to the evolving product.
17–57
Iterative, Incremental Product Development
FIGURE 17.2
17–58
Useful in developing critical breakthrough technology or defining essential features Continuous integration, verification, and validation of the evolving product.
Frequent demonstration of progress to increase the likelihood that the end product will satisfy customer needs.
Early detection of defects and problems.
Superior to traditional plan-driven development when it comes to creating new products, evidence shows 17–59
Popular Agile PM Methods
Scrum Extreme Programming Agile Modeling Rapid Product Development (PRD) Agile PM Methods Crystal Clear RUP (Rational Unified Process) Dynamic Systems Development Method (DSDM) Lean Development 17–60
Agile PM Principles
Focus on customer value Iterative and incremental delivery Experimentation and adaptation Self-organization Continuous improvement 17–61
Agile PM in Action: Scrum
Scrum Methodology Is a holistic approach for use by a cross functional team collaborating to develop a new product.
Each
iteration
is called a
SPRINT
Defines product features as deliverables and prioritizes them by their perceived highest value to the customer.
Re-evaluates priorities after each iteration (sprint) to produce fully functional features.
17–62
Scrum Development Process
FIGURE 17.3
17–63
Key Roles and Responsibilities in the Scrum Process
Product Owner Acts on behalf of customers to represent their interests.
Development Team Is a team of five-to-nine people with cross functional skill sets--is responsible for delivering the product.
Scrum Master
(aka Project Manager) Facilitates scrum process and resolves impediments at the team and organization level by acting as a buffer between the team and 17–64
SCRUM….In Summary
In summary, scrum is an agile process to manage and control development work. Scrum is a team-based approach to iteratively, incrementally develop systems and products when requirements are rapidly changing. Scrum is a process that controls the chaos of conflicting interests and needs. Scrum is a way to improve communications and maximize co-operation. Scrum is a way to detect and cause the removal of anything that gets in the way of developing and delivering products. Scrum is a way to maximize productivity. Scrum is scalable from single projects to entire organizations. Scrum has controlled and organized development and implementation for multiple interrelated products and projects with over a thousand developers and implementers.
Another IT Project Type: System Acquisition/Conversion/Cut-over
Decide on requirements Grade commercial apps that fit Decide on one of them Install Conversion Cut-over
Another IT Project Type: Conversion
Slam-dunk Parallel Pilot
Another IT Project Type: System Integration and Test
Component selection Component integration and testing
Another IT Project Type: Process Reengineering—Michael Hammer
1) determine measures of performance 2) install measures of performance 3) delineate entire existing process in all its gory detail 4) perform process value analysis and activity-based costing 5) benchmark processes by comparison with other processes
More Process Reengineering
6) design re-invented process 7) simulate re-invented process 8) prepare report with recommendations 9) upgrade the software applications that support the re-engineered process 10) install re-invented process 11) measure improvements
Another IT Project Type: Enterprise Integration
the ability to read-from and write to all of the apps and data sources across the enterprise Application integration Use a common SOFTWARE BUS to “glue” them together Data warehousing
More Enterprise Integration
Workgroup/Workflow Apps Messaging systems Data federation performs integration while leaving the source data in place
Another IT Project Type: System Maintenance
Can use Spiral model here Software doesn’t need to be greased, like something mechanical does, so what do we mean by maintenance of software?
Selection Methodology: For Software, Processes and Projects
Could use MAUT, as we previously suggested Could use ROI Could use IRR
This is it: no more time….no more slides in this section…. This year….
JAD, Joint Application Development
Involves the use of GROUPWARE to jointly develop software Developers can be geographically dispersed Analogous to concurrent engineering McConnell considers it a best practice
Another IT Project Type: Rapid Application Development
The Dupont time-bucket approach
RAD, Rapid Application Development
Show the user/client a piece Get feedback and revise Show user/client another piece Get feedback and revise POWERBUILDER, VISUAL BASIC fit this kind of development Utilizes a prototyping tool, usually
RAD and eCommerce Internet Development
Projects last any where from 3 weeks to 6 months max Most are around 5 weeks in duration Not much analysis is done VB, XML, ASP and Visual InterDev are the technology choices About four or five team members Front-end developer, back-end developer, DB specialist, PM and or PL
RAD Basics
Uses time-boxing Is focused on getting at least some functionality delivered by a specific due date Often, this approach skips the analysis phase and goes immediately to design Fits the paradigm of a software factory
RAD Disadvantages
Does not scale up Would not use this on any large project, especially one with lots of indigenous risk
Visual Basic and PowerBuilder
Facilitates both RAD and SAD Must
REUSE
with the tool the base objects that come to get the full benefit Must create a culture and an environment that encourages use of the base objects Sometimes, both the waterfall and the spiral models are too slow
How do we get applications out the door in a timely manner?
Reuse as much as possible--data and presentation components Intuitive Object-Oriented approach is the answer Use Dupont’s approach Calendar-driven development Trade time for functionality Use time boxes (three months, usually) can drop low-priority functionality
One of our Goals: Flexibility
We want a methodology that is so flexible that the requirements can change all during the development process, yet we can still meet the needs of the end users at the time of cut-over (i.e., deliver the product on time and within budget)
User Prototypes
First couple of iterations are really prototypes--fifth iteration is the final product Limit cosmetic iterations to just two MDI frames should not be coupled, should possess very low cohesion
Teams
Are in competition with each other Use a JAD session to plan the project Provide training as needed Use a pilot project
Change Management
If end-user’s want to add something, they must then be willing to give something up There can be instances in which the time box must be aborted But the end-users will have to eat the cost
The Goal
Get the application living quickly Learn from it Then enhance it Time is more important than functionality Use two three-month time-boxes When the final version of the product is delivered the end of the second three-months, users will have a much greater understanding of their requirements
Creating object repositories
Must use the old software factory model Deposit well-documented objects, COMPONENTS in the repository Encourage developers to reuse the objects by incentive/reward mechanisms Must avoid re-inventing objects within all the functions of an organization
Standards should be devised
Version control--allowing for all versions of the module to be tracked
Screens/Pages
Each member of the team delivers from 1 to three screens (pages) daily A significant release should be forthcoming each week Each developer’s assigned tasks should be broken down into chunks of functionality that must be delivered by certain due dates
Selection Methodology: For Software, Processes and Projects
Define the application to be computerized Develop a list of COTS packages that is available to support the app Gather information about the COTS packages Narrow the list of possible choices down to less than a half dozen Obtain hands-on demonstrations of the few remaining packages
Selection Methodology, Cont’d
Of those that remain, perform a final detailed evaluation Make a decision Purchase the selected COTS package Learn to use the package Implement the package
Project Standards
Operational Issues -- Brooks
Another Software guru like Boehm
Software Development -- Brooks
What is the problem with hiring additional people on a project that is behind schedule?
Under what circumstances are men and man-hours interchangeable?
How much time should be set aside for testing and integration, according to Brooks?
What are some good ways to organize work packages to avoid some of the problems Brooks is talking about?
Questions
All programmers are ____.
When schedule slippage occurs, the natural thing is to ____.
What are the three stages of creative activity?
These are analogous to ____?
Brooks says that the programmer builds from ____.
pure thought stuff; concepts and flexible representations. However, today this is not true.
Questions, Cont’d
Cost varies as the number of persons and the number of months • However, progress does not.
Can we use the man-month as a unit for sizing a job?
• No. It is too dangerous People and months are inter-changeable only when a task can be partitioned among many workers with _____.
• No communication between them The added burden of communication is made up of _____.
• two parts: training and intercommunication.
Long projects can expect a turnover of ___% a year.
Questions, Cont’d
How much of the total time does Brooks devote to Definition, Analysis and Design?
• 1/3 How much time to coding?
• 1/6 to Coding How much time to testing?
• 1/4 to component test and early system test • 1/4 to total system test