SOFTWARE ENGINEERING

   (Session 2)

AGENDA

· Project Management Concepts

o Managing People

o Managing Product

o Managing Process

o Managing Project

· W5H Principle of Boehm

· Critical Issues related to success

· Software Metrics , Size Oriented

· Software Metrics, Function Oriented

· Software Quality Metrics

· General Metric Issues

Project Management Concepts - I

· Things to be managed

o (People, Product, Process, Project).

· People

o PM-CMM (People Management CMM) : Attract, Grow, motivate, deploy, retain

o Senior Manager, Project Manager, Practitioner, Customer, End-Users

o Team Leader should motivate, organize, Inject ideas & Value innovations.

o Team can be

§ Democratic Decentralised

§ Controlled DeCentralised

§ Controlled Centralised

o Organisation Paradigms

§ Closed (Similar to past efforts)

§ Random (Innovation Oriented)

§ Open (Collaborative working)

§ Synchronous (Problem Division)

o Personal Communication can take precedence over formal communication

· Product

o Scope & Domain of services to be chosen.

o Features and performance criteria to be chosen.

o Problem to be decomposed to create separate measurement for components.

· Process

o The process model has to be chosen

o Each framework activity shall be related to group of tasks.

o Process decomposition is done in detail

Project Management Concepts - II

· Project

o Understand What can go wrong.

§ Understanding, scope, change management,  technology, need change, deadline, user resistance, funding, skills, practices.

o First 90 % takes 90 % effort, rest 10 % takes again 90 % . (90-90 syndrome)

· BOEHM Approach ( W5HH)

o Write Details of

§ Why?

§ What?

§ When?

§ Who?

§ Where?

§ How to do?

§ How much resources?

· Critical Practices for organizational success.

o Risk Management

o Cost Schedule Estimation and tracking

o Metric principles

o Defect Analysis

o People and Program Management

Software Metrics - I

·  Project Management process Layer includes

§ Objectives and scope

§ Measurement and metrics

§ Estimation

§ Risk Analysis

§ Schedule tracking and control

· Metric include the unit definition of

§ Productivity, Quality and Technical metrics w.r.t. size, function and human effort requirement

· Measurement types

§ Direct (Lines of code LOC, Execution Speed, Memory used)

§ Indirect (Complexity, Maintainability, Efficiency)

Software Metrics - II

· SIZE Oriented metrics

§ Productivity = KLOC/person-month

§ Quality = defect/KLOC

§ Cost = Amount/LOC

§ Documentation = pages/KLOC

§ LIMITATION: LOC method is language dependent, Efficient coding gets less money, pre-estimation is difficult, doesn’t work for non-procedural language.

· Function Oriented metrics

§ Indirect: based on function point approach.

§ Inputs, Outputs, Queries, Files, Interfaces are counted for degrees of complexities.

§ FP = total count x [ 0.65 + 0.01 x Sum (Fi) ]

§ Fi is calculated as degrees of complexity against a set of questions.

§ Productivity = FP/person-month

§ Quality = defect/FP

§ Cost = Amount/FP

§ Documentation = pages/FP

§ Another variation of this called FEATURE POINT METRICS also include algorithm as metric. FP is calculated by adding linear weights in this case.

§ LIMITATION: It is still quite subjective, FP is domain dependent, FP is indirect.

Software Quality Metrics

· Indirect Metrics

§ Program Complexity

§ Effective Modularity

· Direct Metrics

§ Program Size

§ Defect Reports

· Above relates to

§ Product Operation

§ Product Revision

§ Product Portability

· Measurement of

§ Correctness (defect/KLOC)

§ Maintainability (low Mean Time To Change MTTC)

§ Integrity ( Sum [ 1 – (Threat x (1-security))]

§ Usability (Skill required to learn, time to become efficient, productivity increase, attitude/perception of users)

General Metric Issues

· LOC & FP relationship is attempted for various languages.

§ Assembly , 300 LOC/FP

§ COBOL, 100 LOC/FP

§ FORTRAN, 100 LOC/FP

§ Pascal, 90 LOC/FP

§ Ada, 70 LOC/FP

§ OOP, 30 LOC/FP

§ 4GL, 20 LOC/FP

§ Code Generators, 15 LOC/FP

· Productivity Issues. The variation due to various factors are as follows.

§ People , 90%

§ Problems , 40%

§ Process , 50%

§ Product (Reliability and Performance Req.), 140%

§ Resources , 40%

· Understanding, that measurement is important for improvement is required.

· Start to make a measurement baseline now. This can be easily based on past projects.