Software Engineering Fundamentals

Introduction to Software Engineering

• The lecture introduces the fundamentals of software engineering, its scope, and various process models.
• Dr. Rasha Kashef presents the information.

Software Defined

• Software is a collection of executable programming code, associated libraries, and documentations.
• Implies it is large, robust, reusable, and evolving.
• Why studying it is important: teaches design, systematic problem solving.

Software Systems

• System software examples: compilers, editors, file management utilities.
• Application Software examples: standalone programs for specific needs.
• Engineering/Scientific software: features "scientific-related" algorithms.
• Embedded software: resides within a product or system.
• Product-line software: focuses on a limited marketplace to address mass consumer market.
• Web & network centric software: are built in sophisticated environments integrated with remote databases and business applications.
• Intelligent systems: makes use of non-numerical algorithms to solve complex problems like robotics, expert systems, and pattern recognition.

Software Engineering (SE) Defined

• The application of a systematic, disciplined, quantifiable approach to the development, operation, and maintenance of software, according to IEEE Standard 610.12.
• As defined by the Canadian Standards Association: it is the systematic activities involved in the design, implementation and testing of software to optimize its production and support."

Software Engineering Scope

• It is part of a much larger system design that includes telephone switching, banking, and hospital admin systems, and aircraft.
• Performing it correctly requires a comprehensive look at all aspects.
• Involves entities, activities, and interfaces with other systems and users.
• Key attribute: understanding the application and user needs.
• Activity support: decide what activities the system should support and how.
• Software systems development: requires emphasis on different priorities depending on application domain like telecom in time-to-market, aerospace/NASA on safety, and telecom/banking on maintainability.

The Software Problem

• "A Software" means industrial strength software that solves a specific problem.
• Three forces at play: cost, schedule, and quality.

Cost of Software

• Lines of Code (LOC) & Person-Month (PM) provides context to costs involved.
• Approximately 1000 LOC can be produced per PM.
• Cost per LOC is about : $3-$10, so the Cost = $3K to $10K/PM.
• Developing a simple business application (20KLOC to 50KLOC) can cost $100K-$2.25 Million; it can run on $10K-$20K hardware.
• Hardware costs in an IT solution are small, relative to software costs.

Software Schedule

• Shorter demanded by business requirements.
• Products have often failed to complete on time.
• Cycle time + Cost : a fundamental driving force

Productivity - Cost and Schedule

• Modeled by productivity.
• Productivity = output / input resources.
• Output is measured via LOC or KLOC.
• Input resources: effort in person months as well as overhead.
• Higher productivity results in lower cost/cycle time.
• Quality and productivity drives software projects.

Software Quality

• Developing high-quality software is a basic goal; productivity and quality drives software projects.
• Quality is harder to define.
• The ISO Standard dictates six quality attributes.
• There are multiple dimensions of quality that makes it difficult to quantify with a simple value.
• Being project specific depending on the needs.
• Reliability is generally the most important criterion

ISO Standard Quality Attributes:

• Functionality
• Reliability
• Usability
• Efficiency
• Maintainability
• Portability

Stakeholders in Software

• Customer/client: wants software built, but often doesn't know specifics.
• Managers/Designers: plan software, but hard to foresee all issues.
• Developers: write code to implement software, but hard to write complex code for large systems.
• Testers: perform quality assurance, but impossible to test all actions.
• Users: purchase and use software product, but can misunderstand it.

Software Development Fundamentals

• Process: A particular method of activities, actions and tasks performed when some product is to be created.
• Software Process: A sequence of steps/activities, accompanied by constraints on order of execution, that lead to software with the desired outcome.

Software Development Process Key Points

• Process is distinct from product.
• Products are the outcomes of executing a process.
• Software Engineering focuses on process.
• Proper processes deliver software products in a timely manner, within budget, at high quality.

Key Processes in Software Development

• Software Development -> Engineering and Project Management
• Other -> Configuration management, Change management, Inspection process, Improvement process

Two Major Processes

• Engineering: encompasses development and quality assurance needed to engineer the software
• Project management: provides the planning and controlling of the development process.
• Roles: developers execute the Engineering and Project managers execute the management process.

Common Phases in Software Development

• Requirements Analysis
• Design
• Coding
• Testing
• Delivery

Basic Software Development Process

• Phases: Requirements, Design, Implementation, Testing, Integration, Maintenance.
• Requirements is under the Software Development Paradigm
• Design is under the Software Design Paradigm
• Implementation is under Programming Paradigm
• Software Design Models: Procedural, Object Oriented, Logic, Functional.
• Involves Planning/Design & Documentation

Phase 1: Requirement Analysis

• States the problem precisely.
• Forms a basis of agreement between the user and developer.
• Specifies "what"
• Is difficult as the problem is often not well understood.
• Even medium systems can have specification documents spanning hundreds of pages.
• Output is the Software Requirements Specification (SRS) document.

Phase 2: Design

• Moves from problem domain to solution domain.
• Key tasks: Architecture design (components/connectors), High-level + Detailed design (modules and data structures).
• Methodologies: Focuses on architecture or high-level design.
• Outputs are architecture/description/logic design documents.

Phase 3: Coding

• Converts design into code in specific language.
• Goal: Implement design simply and understandably. Well-written code reduces testing and maintenance effort.
• Coding phase affects both testing and maintenance.
• Output is Code.

Phase 4: Testing & Quality Assurance

• Defects can be introduced within each phase.
• Goals include: Identifying most of defects and to Remove defects for high quality.
• Task: Very expensive and must be properly planned and executed.
• Output: Test plans/results, and final, reliable and tested code.

Phase 5: Delivery

• The "Operations" group typically handles delivery.
• The mode is variable by distribution of e.g. shrink wrapped software, in house software, web based and Software As A Service (SaaS).
• From a user perspective, it can be equally important to the initial design.

Software Development Life Cycle: Typical Effort Distribution

• Requirements: 10-20%
• Design: 10-20%
• Coding: 20-30%
• Testing: 30-50%
• Costliest aspect is Coding.

Error Introduction and Detection

• Most errors introduced in Coding (50%), followed by Design (30%) and Requirements (20%).
• Defects can be injected at any of the major phases.
• Cost of defect removal increases exponentially with latency time.
• Cheapest way to detect and remove defects is by checking after every phase

Software Project and Process Models

• Project: building a software system within cost and schedule while achieving high quality and customer satisfaction.
• Process model: abstract general specifications optimized for a particular class of problems; effective practices used for project success.
• Different process models perform the five phases of the development process in different manners.
• A project may select its process through one of the process models.

Software Process Models Defined

• An abstract representation of a process and a description of a process from some particular perspective.

Process Model Categorization

• Traditional: Waterfall (oldest, widely used), Prototyping (followed by Waterfall), Iterative (used widely in product development), and Timeboxing (Iterative 2.0).
• Modern: Agile (lightweight methodologies, best for mobile applications).

Waterfall Model

• Stages: Requirements, High Level Description, Detailed Design, Code, Test, and Deploy.
• Phases proceed in a linear sequence with no opportunity for feedback. Partition project addressing specific task

Waterfall Model, More info

• The linear ordering implies each phase should have some output.
• The output must be validated/certified
• Common outputs: SRS, project plan, design/test plans, final code, and supporting documentation.

Waterfall Model: Advantages an Disadvantages

• Advantages: Natural approach for problem solving and is conceptually simple.
• Advantages: Easy to administer in a contractual setup; each phase signifies milestone.
• Disadvantages: Inflexible partitioning makes adaptation to changing requirements difficult.
• Disadvantages: Suited when the requirements are well-understood. Model describes stepwise refinement process".
• Extensively applied military, healthcare, and aerospace sectors.

Prototyping Model

• Use: risky or unfamiliar projects
• Addresses the requirement specification limitation of waterfall
• Instead of defining specifications, it is built to understand the requirements..
• Helps alleviate risk and Expect to "throw away" first version.
• Replaces requirements stage with a small waterfall model.

Prototyping Model Development

• Begins with basic requirements.
• Focuses on key features requiring the best understanding are included in the prototype.
• Including features that are already well-understood are unnecessary.
• Used to improve the understanding of requirements.

Prototyping Model: Development Cost

• Can be kept low by only including features needing clarification and omitting elements like Exception handling, and standards. A small percentage of costs is enough.
• Can contribute to building.

Prototyping Model: Advantages and Disadvantages

• Advantages: Stable requirements, satisfies user needs, early exploration of scaling issues, the ability to modify/cancel the project early, and enhanced user engagement.
• Disadvantages: Potential hit on cost/schedule, false sense of security if prototype does not focus on high-risk areas
• Applicability: difficulty in eliciting requirements, confidence is low, and in areas where design is driven by end-user requirements.

Iterative Model

• Addresses limitations of the waterfall (the "all or nothing").
• Combines the benefits of prototyping and waterfall approaches.
• Allows for the development and delivery of software in increments.
• Each increment is complete in itself.
• Feedback from one iteration is used in future iterations.
• Commonly used in customized development
• Most Software Products follow it
• Newer approaches XP, Agile relies on Iterative Development

Iterative Model Advantages and Disadvantages

• Advantages: Get-as-you-pay and Feedback for improvement.
• Disadvantages: Architecture/design may not be optimal, the amount of refactoring may increase and the Total cost may increase.
• Applicability: response time important, all requirements are unknown.
• Execution: Each iteration is a mini-waterfall (decide the specs, then plan the iteration). The amount of new functionality determines length of iteration.

Time Boxing

• Alterration of the iterative development
• Fix iteration duration + determine the specifics.
• Divides iteration in a few equal stages.
• Integrates pipelining principles.
• General approach: fixing functionality for each iteration, then plan and perform.
• In time boxed iterations – adjust functionality to fit a fixed period.

Time Boxing Model: Execution and Team Sizes

• Duration of each iteration, total workload, and productivity stays the same- Average cycle or delivery timeline cuts to a third.
• Team Size->Linear Execution
• The same team performs all stages. For a stage, it is S in linear execution the total team size is also S.
• To achieve pipeline execution, the total team size should be three times (one for each stage)
• Total team size in timeboxing is larger, reducing cycle time.
• Timeboxing allows for structured addition of manpower to minimize turnaround.
• It is worth noting time stay unchangeable.

Time Boxing Model: Advantages & Disadvantages

• Advantages: Use of manpower shortens delivery schedules while increases flexibility.
• Disadvantages: demands larger project management, and high synchronization.
• Applications: When rapid delivery schedules, flexible features and stable architecture are needed.

Agile Model

• Assumptions: Difficulty in predicting software requirements, analysis, and development.
• Assumptions: Design/construction must be interleaved.
• Solution: Process adaptability
• Example: XP, extreme programming
• The process includes planning (stories), design, programming, testing, communication is paramount.

Summary

• The process achieves project objectives (Q&P).
• Process models form the basis of the project.
• The process has an identifiable task.
• Many model for Development
• Tailor best suited for.