0.0 by Qwen

Chapter 6: Architectural Design

Why Architectural Design is Important?

Definition

• Architectural design (meaning: the high-level structure of a software system, showing its main parts and how they connect) is the process of defining a system’s overall organization, including its sub-systems, modules, and control flow.

Causes

• Not specified in notes

Goals / Objectives

• To create a clear framework for control and communication between system parts
• To support stakeholder communication, requirements validation, reuse, negotiation, and complexity management

Importance

• Serves as a foundation for the entire system design
• Helps manage complexity in large systems
• Enables early validation of whether the system can meet its requirements
• Acts as a shared reference point for developers, clients, and other stakeholders

Procedures

• Not specified in notes

Advantages & Disadvantages

• Advantages:

  • Supports stakeholder communication through high-level system views
  • Helps system analysts check if requirements can be met
  • Enables large-scale reuse of proven architectures
  • Serves as a design plan for negotiation and discussion
  • Acts as an essential tool for complexity management

• Disadvantages:

  • Not specified in notes

Impact / Effect

• Makes system development more organized and predictable
• Reduces risk of miscommunication between teams
• Increases chances of successful system delivery by aligning design with requirements early

Examples

• Not specified in notes

Key Takeaways

• Architectural design gives a big-picture view of how a system is built
• It’s not just for developers—it helps everyone involved understand the system
• Good architecture helps reuse, manage complexity, and confirm requirements early

---

Architectural Design Activities

Definition

• The three main activities in architectural design are:
  1. System Organization (meaning: deciding how to split the system into major parts)
  2. Modular Decomposition (meaning: breaking those parts into smaller, manageable pieces called modules)
  3. Control Modeling (meaning: defining how control flows between parts—what starts what, and when)

Causes

• Not specified in notes

Goals / Objectives

• To structure the system clearly
• To decompose large components into reusable, understandable modules
• To define control relationships so services are delivered at the right time and place

Importance

• These activities form the core workflow of architectural design
• They ensure the system is well-organized, maintainable, and functional

Procedures

• System Organization: Identify sub-systems and how they communicate
• Modular Decomposition: Break sub-systems into modules using object-oriented or function-oriented strategies
• Control Modeling: Choose a control style (centralized or event-based) and apply a specific model (e.g., call-return, broadcast)

Advantages & Disadvantages

• Advantages:

  • Provides a structured approach to designing complex systems
  • Each activity addresses a different aspect of system design (structure, detail, behavior)

• Disadvantages:

  • Not specified in notes

Impact / Effect

• Leads to a complete architectural blueprint that guides implementation
• Helps avoid chaotic or inconsistent designs

Examples

• Not specified in notes (though examples appear under each sub-activity)

Key Takeaways

• Architectural design has three key steps: organize, decompose, and control
• Each step answers a different question: What are the big parts? How are they built? How do they work together?
• Skipping any of these can lead to a weak or unmanageable system

---

1. System Organization / Structuring

Definition

• System organization (meaning: dividing the whole system into major independent units called sub-systems and defining how they interact)

Causes

• Needed when building complex systems that must handle multiple responsibilities (e.g., vision, movement, packing in a robot)

Goals / Objectives

• To identify sub-systems
• To establish a framework for control and communication between them
• To choose an appropriate system organization style (repository, client-server, or layered)

Importance

• Sets the foundation for the entire architecture
• Determines how data is shared and processing is distributed

Procedures

• Represent sub-systems in a block diagram
• Decide how sub-systems communicate and share data
• Select one of three widely used organization models:
  • Repository model
  • Client-server model
  • Layered model

Advantages & Disadvantages

• Advantages:

  • Clarifies system boundaries and responsibilities
  • Enables parallel development of sub-systems

• Disadvantages:

  • Poor choices can lead to tight coupling or inefficient data sharing

Impact / Effect

• Affects scalability, maintainability, and performance of the whole system
• Influences hardware and network choices (e.g., in distributed systems)

Examples

• Packing robot system includes sub-systems like:
  • Vision system
  • Object Identification System
  • Gripper Controller
  • Arm Controller
  • Conveyor Controller
  • Packing Selection System
  • Packing System

Key Takeaways

• System organization means splitting the system into big, independent parts
• You must decide how they talk to each other and share data
• There are three main styles to choose from, each with trade-offs

---

Repository Model

Definition

• A repository model uses a central database where all sub-systems store and access shared data.

Causes

• Used when large amounts of data must be shared among many sub-systems

Goals / Objectives

• To enable efficient data sharing
• To provide centralized data management

Importance

• Avoids data duplication and ensures consistency across sub-systems

Procedures

• All sub-systems read from and write to a single central repository
• Sub-systems do not need to know how data is produced—only how to use it
• The repository defines a common data schema

Advantages & Disadvantages

• Advantages:

  • Allows sharing of large amounts of data
  • Enables centralized management
  • Sub-systems are independent of data production details

• Disadvantages:

  • Sub-systems must compromise on a single data model
  • No support for specific data management policies per sub-system
  • Hard to distribute across multiple machines
  • Data evolution (changing the structure) is difficult and expensive

Impact / Effect

• Makes integration easier but limits flexibility
• Can become a performance bottleneck if not scaled properly

Examples

• Integrated CASE (Computer-Aided Software Engineering) toolset:
  • Project Repository stores all data
  • Tools like Design Editor, Code Generator, Report Generator, etc., all use the same repository

Key Takeaways

• Great for data-heavy systems where consistency matters
• But hard to change or distribute later
• All tools share one central database

---

Client-Server Model

Definition

• A client-server model is a distributed system where servers provide services and clients request them over a network.

Causes

• Used when processing and data need to be spread across multiple machines

Goals / Objectives

• To distribute functionality across independent components
• To allow clients to access shared services remotely

Importance

• Enables scalable, networked applications
• Supports resource sharing (e.g., files, printers, databases)

Procedures

• Set up stand-alone servers that offer specific services (e.g., data, video, printing)
• Build clients that call on these services
• Connect them via a network (e.g., wide-bandwidth network)

Advantages & Disadvantages

• Advantages:

  • Straightforward data distribution
  • Creates an effective network system
  • Uses cheaper hardware (clients can be simple)
  • Easy to add or upgrade servers

• Disadvantages:

  • Different servers may use different data models, leading to inefficient sharing
  • Redundant management (each server manages its own data)
  • No central register of available names and services

Impact / Effect

• Makes systems more flexible and scalable
• But can lead to integration challenges if services aren’t well-coordinated

Examples

• Film and picture library system:
  • Clients (4 shown) connect via network
  • Servers include:
    • Catalogue Server (holds catalogue data)
    • Video Server (stores film clips)
    • Picture Server (holds digitized photos)
    • Hypertext Server (manages web links)

Key Takeaways

• Clients ask, servers respond—over a network
• Easy to grow by adding more servers
• But watch out for inconsistent data formats and missing service directories

---

Layered Model

Definition

• A layered model (also called Abstract Machine Model) organizes the system into layers, where each layer provides services to the one above it.

Causes

• Used when you want to isolate functionality and support incremental development

Goals / Objectives

• To interface sub-systems cleanly through layers
• To allow one layer to change without breaking others (in closed layered models)

Importance

• Supports modular, step-by-step development
• Makes testing and debugging easier by isolating concerns

Procedures

• Stack layers from bottom (e.g., OS) to top (e.g., application)
• Each layer uses services from the layer below and offers services to the layer above
• In a closed layered model, a layer only interacts with adjacent layers

Advantages & Disadvantages

• Advantages:

  • Supports incremental development
  • In closed models, changing one layer only affects adjacent layers

• Disadvantages:

  • Not specified in notes

Impact / Effect

• Promotes clean separation of concerns
• Can add performance overhead if too many layers exist

Examples

• Version Management System layers:
  • Top: Configuration management system layer
  • Then: Object management system layer
  • Then: Database system layer
  • Bottom: Operating system layer

Key Takeaways

• Think of it like a cake with layers—each layer builds on the one below
• Great for step-by-step building and testing
• Changes usually don’t ripple through the whole system

---

2. Modular Decomposition

Definition

• Modular decomposition (meaning: breaking each sub-system into smaller, manageable pieces called modules or components)

Causes

• Needed to simplify implementation and assign work to teams

Goals / Objectives

• To decompose sub-systems into modules
• To choose a decomposition strategy: object-oriented or function-oriented

Importance

• Turns big, abstract sub-systems into concrete, buildable units
• Affects code reuse, testing, and maintenance

Procedures

• Use one of two main strategies:
  • Object-Oriented Decomposition: Group data and functions into objects (e.g., Invoice, Payment)
  • Function-Oriented Decomposition: Split system into functions or processes (e.g., Read invoices, Issue reminders)

Advantages & Disadvantages

• Advantages:

  • Makes code easier to understand and maintain
  • Supports team-based development

• Disadvantages:

  • Poor decomposition can lead to tight coupling or code duplication

Impact / Effect

• Determines how flexible and reusable the code will be
• Influences how bugs are tracked and fixed

Examples

• Invoice Processing System:
  • Object-Oriented:
    • Objects: Customer (with fields: CustomerNo, Name, Address, CreditPeriod)
    • Invoice (fields + methods: Issue(), sendReminder(), acceptPayment(), sendReceipt())
    • Payment and Receipt objects
  • Function-Oriented:
    • Functions grouped as:
      • Read issued invoices
      • Identify payments
      • Find payments due
      • Issue payment reminder
      • Issue receipts
    • Data flows between Invoices → Payments → Reminders → Receipts

Key Takeaways

• Modular decomposition turns big parts into small, buildable pieces
• You can group by objects (data + actions) or by functions (tasks)
• Choice affects how easy it is to change or fix the system later

---

3. Control Modeling

Definition

• Control modeling (meaning: defining how control flows between sub-systems or modules—what triggers what and when)

Causes

• Needed because sub-systems must coordinate to deliver services at the right time and place

Goals / Objectives

• To establish control relationships between components
• To ensure correct timing and sequencing of operations

Importance

• Without proper control, the system may miss events, crash, or behave unpredictably
• Critical for real-time and concurrent systems

Procedures

• Choose one of two general control styles:
  1. Centralized Control
     • a. Call-Return Model
     • b. Manager Model
  2. Event-Based Control
     • a. Broadcast Model
     • b. Interrupt-Driven Model

Advantages & Disadvantages

• Advantages:

  • Provides predictable system behavior
  • Matches control style to system type (sequential vs. concurrent vs. real-time)

• Disadvantages:

  • Wrong choice can lead to bottlenecks (centralized) or unpredictability (event-based)

Impact / Effect

• Determines whether the system runs smoothly under load
• Affects responsiveness and fault tolerance

Examples

• Call-Return Model:
  • Main program calls Routine 1 → Routine 1.1, etc. (like a family tree of function calls)
• Manager Model:
  • System Controller manages Sensor Processes, Actuator Processes, Fault Handler, User Interface, Computation Processes
• Broadcast Model:
  • An Event and Message Handler sends an event to all sub-systems; only those that can handle it respond
• Interrupt-Driven Model:
  • Used in real-time systems with an Interrupt Handler

Key Takeaways

• Control modeling decides who’s in charge of running the system
• Centralized = one boss (good for simple or sequential systems)
• Event-based = react to signals (good for distributed or real-time systems)

---

Centralized Control

Definition

• A centralized control style means one sub-system has overall responsibility for starting, stopping, and coordinating others.

Causes

• Used in systems where predictable, top-down execution is needed

Goals / Objectives

• To simplify coordination by having a single point of control
• To ensure orderly execution of tasks

Importance

• Reduces race conditions and unpredictable behavior in simpler systems

Procedures

• Two models:
  • Call-Return Model: Top-down subroutine calls (like a tree)
  • Manager Model: One system manager coordinates concurrent processes

Advantages & Disadvantages

• Advantages:

  • Simple to understand and debug
  • Good for sequential or moderately concurrent systems

• Disadvantages:

  • Can become a bottleneck
  • Less suitable for highly distributed or reactive systems

Impact / Effect

• Makes system behavior deterministic
• May limit scalability in large distributed setups

Examples

• Call-Return: Main Program → Routine 1 → Routine 1.1 (used in sequential systems)
• Manager Model: System Controller managing sensors, actuators, UI, etc. (used in concurrent systems)

Key Takeaways

• One component runs the show
• Call-return = function calls in order
• Manager = boss that handles multiple tasks at once

---

Event-Based Control

Definition

• An event-based control style means the system responds to events (from environment or other components) rather than following a fixed sequence.

Causes

• Triggered by external inputs, user actions, or system signals

Goals / Objectives

• To build responsive, reactive systems
• To support loose coupling between components

Importance

• Essential for real-time, interactive, or distributed systems
• Enables asynchronous behavior

Procedures

• Two models:
  • Broadcast Model: Event is sent to all sub-systems; those that can handle it do so
  • Interrupt-Driven Model: Special interrupt handler responds immediately to urgent signals (used in real-time systems)

Advantages & Disadvantages

• Advantages:

  • Highly responsive to changes
  • Broadcast model works well across networked computers

• Disadvantages:

  • Can be hard to debug (events may come in any order)
  • Risk of unhandled events or multiple handlers reacting

Impact / Effect

• Makes systems more flexible and dynamic
• But can introduce timing bugs if not designed carefully

Examples

• Broadcast Model:
  • Event Handler sends event to Subsystem 1–4
  • Each subsystem checks if it can respond (e.g., E1 → SS1, E2 → SS2)
• Interrupt-Driven Model:
  • Used in real-time systems (e.g., robotics, medical devices)

Key Takeaways

• System reacts to signals, not a fixed plan
• Broadcast = shout to everyone, let relevant parts respond
• Interrupt-driven = drop everything for urgent tasks (real-time use)