Architectures of Modern Operating Systems

A Comprehensive Analysis of Six Core Architectures

Monolithic
Layered
Microkernel
Modular/Hybrid
Exokernel
Distributed

Understanding the Criteria:

Performance: Speed of system calls and I/O operations
Reliability: System stability and fault tolerance
Extensibility: Ease of adding new features and modules
Fault Isolation: Ability to contain failures and prevent system crashes
Dev Complexity: Difficulty of development and maintenance

Why Architecture Matters

Architecture is the blueprint. It defines performance, reliability, extensibility, and security. No single design wins everywhere; it's all about trade-offs.

OS Architecture
Performance
Speed and efficiency of operations
Reliability
System stability and fault tolerance
Scalability
Ability to handle growing demands
Security
Protection against threats

The Six Architectures

These are the six architectures we'll compare in detail.

Monolithic

All services in kernel space

Layered

Hierarchical organization

Microkernel

Minimal kernel, user-space servers

Modular/Hybrid

Loadable modules

Exokernel

Library operating systems

Distributed

Multiple machines, single image

Monolithic Architecture - Concept

All core OS services—process, memory, file systems, networking, and drivers—run together in kernel space. That tight integration is why it's fast.

User Space
App 1
App 2
App 3
Kernel Space
Scheduler
Memory Manager
VFS
Network Stack
Device Drivers
Hardware
Single protection domain for core services

Monolithic - Pros & Cons + Examples

Advantages

  • Low overhead communication
  • Great performance
  • Simple control flow
  • Mature and stable

Disadvantages

  • Kernel-space driver bugs can crash system
  • Harder maintenance
  • Large trusted computing base
  • Limited fault isolation

Examples

LinuxLinux
UNIXTraditional UNIX
FreeBSDFreeBSD

Simulation - Monolithic read()

Step-by-step visualization of how a file read operation works in a monolithic kernel

Layered Architecture - Concept

The OS is arranged in tiers—each layer only talks to its immediate neighbors. It's clean, testable, and predictable.

User Space
Applications
System Call Interface
API Layer
Kernel Services
Process Mgmt, Memory Mgmt, File System
Device Drivers
Hardware Abstraction
Hardware
Physical Resources
Narrow, well-defined interfaces between layers

Layered - Pros & Cons + Examples

Advantages

  • Clean separation of concerns
  • Easy to test and debug
  • Modular design
  • Well-defined interfaces

Disadvantages

  • Performance overhead
  • Rigid structure
  • Difficult to change layer boundaries
  • May not fit all problems

Examples

THE OS
Venus OS
Educational Kernels

Simulation - Layered read()

Watch how a request flows through each layer in sequence

Microkernel - Concept

Keep the kernel tiny: only scheduling, IPC, and minimal memory management. Everything else runs as user-space servers.

File Server
Network Server
Driver Server
Window Server
Microkernel
Scheduling IPC Memory Mgmt
Hardware
Faults are isolated—if a driver dies, restart it like a normal process

Microkernel - Pros & Cons + Examples

Advantages

  • Strong fault isolation
  • Easy to extend and modify
  • Better security
  • Modular architecture

Disadvantages

  • Performance overhead from IPC
  • Complex implementation
  • More context switches
  • Debugging can be harder

Examples

QNX
MINIX 3
L4
Mach

Simulation - Microkernel read()

See how IPC enables communication between user-space servers

Hybrid Architecture - Concept

Best of both worlds: critical services in kernel space, others as user-space servers

Network Services
Print Spooler
Window Manager
Hybrid Kernel
Process Mgmt
Memory Mgmt
File System
Core Drivers
Hardware
Performance-critical components in kernel, others in user space

Simulation - Hybrid read()

See how hybrid kernels balance performance and modularity

Hybrid - Pros & Cons + Examples

Advantages

  • Good performance
  • Reasonable modularity
  • Practical compromise
  • Widely adopted

Disadvantages

  • Complex design decisions
  • Still some kernel bloat
  • Not pure architecture
  • Harder to verify

Examples

Windows NT
macOS
BeOS

Exokernel - Concept

Minimal kernel that only manages resources—applications implement their own OS abstractions

LibOS A
UNIX-like
LibOS B
Real-time
LibOS C
Database
Exokernel
Resource Allocation Protection Multiplexing
Raw Hardware Resources
Applications get direct, protected access to hardware

Simulation - Exokernel read()

Experience direct hardware access with minimal kernel overhead

Exokernel - Pros & Cons + Examples

Advantages

  • Maximum performance
  • Application-specific optimization
  • No forced abstractions
  • Innovation in OS design

Disadvantages

  • Very complex to program
  • No standard interfaces
  • Security challenges
  • Limited practical adoption

Examples

MIT Exokernel
Nemesis
Research Systems

Distributed OS - Concept

Multiple machines appear as a single system—transparent resource sharing across the network

Node 1
Process Mgr
File Server
Node 2
Memory Mgr
Name Server
Node 3
I/O Server
Security Mgr
Network Communication Layer
Single system image across multiple machines

Simulation - Distributed read()

Watch how file operations span multiple networked nodes

Distributed - Pros & Cons + Examples

Advantages

  • High availability
  • Scalability
  • Resource sharing
  • Fault tolerance

Disadvantages

  • Network dependency
  • Complex consistency
  • Security challenges
  • Performance variability

Examples

Plan 9
Amoeba
Sprite

Architecture Evolution Timeline

How OS architectures evolved over time

1960s

Monolithic

Early UNIX, simple and effective

1970s

Layered

THE OS, structured approach

1980s

Microkernel

Mach, focus on modularity

1990s

Hybrid

Windows NT, practical compromise

2000s

Exokernel

MIT research, maximum flexibility

Performance Characteristics

Relative performance across different metrics

System Call Overhead

Monolithic
Layered
Microkernel
Hybrid
Exokernel

Fault Isolation

Monolithic
Layered
Microkernel
Hybrid
Exokernel

Real-World Usage Today

Where each architecture is actually used

Desktop/Server

Linux (Monolithic) Windows (Hybrid) macOS (Hybrid)

Mobile

Android (Linux) iOS (Darwin)

Embedded

QNX (Automotive) Linux (IoT) VxWorks

Cloud/Research

Unikernels Container Orchestration

Side-by-Side Comparison

Let's compare head-to-head across the same criteria

Architecture Performance Reliability Extensibility Fault Isolation Dev Complexity
Monolithic
★★★★★ Excellent
★★★☆☆ Average
★★★☆☆ Average
★★☆☆☆ Poor
★★★☆☆ Average
Layered
★★★☆☆ Average
★★★★☆ Good
★★★☆☆ Average
★★★☆☆ Average
★★★☆☆ Average
Microkernel
★★★☆☆ Average
★★★★★ Excellent
★★★★★ Excellent
★★★★★ Excellent
★★★★☆ Good
Hybrid
★★★★★ Excellent
★★★★☆ Good
★★★★☆ Good
★★★☆☆ Average
★★★★☆ Good
Exokernel
★★★★★ Excellent
★★★☆☆ Average
★★★★★ Excellent
★★★☆☆ Average
★★★★★ Excellent
Distributed
★★★★☆ Good
★★★★★ Excellent
★★★★☆ Good
★★★★☆ Good
★★★★★ Excellent

Trade-off Analysis

Understanding the fundamental trade-offs between architectures

Performance

High Performance
Low Complexity

Monolithic

High Performance
High Complexity

Exokernel Hybrid

Lower Performance
Low Complexity

Layered

Lower Performance
High Complexity

Microkernel Distributed
Complexity

Future Trends & Emerging Patterns

Where OS architecture is heading

Container-Native OS

Lightweight, container-optimized kernels

CoreOS, Container Linux

Unikernels

Single-purpose, library operating systems

MirageOS, IncludeOS

Serverless Runtime

Function-as-a-Service optimized kernels

AWS Lambda, Firecracker

Edge Computing

Distributed, low-latency architectures

EdgeX, K3s

Security-First Design

Hardware-assisted isolation

Qubes, Genode

When To Use Which?

Rule-of-thumb guidance for picking an approach

High Performance + Mature Drivers

Choose: Monolithic/Hybrid

Linux, Windows, macOS

Strong Isolation/Regulatory

Choose: Microkernel

QNX, MINIX 3, Embedded Systems

Research/Custom Abstraction

Choose: Exokernel/LibOS

MIT Exokernel, Unikernels

Scale-out Cluster

Choose: Distributed

Plan 9, Amoeba, Cloud OS

Teaching/Clean Interfaces

Choose: Layered

THE OS, Educational Kernels

Case Study: Linux vs QNX vs Windows

Real-world examples of different architectural choices

Linux

Monolithic
General-purpose servers, desktops
Strengths:
  • Excellent performance
  • Huge driver ecosystem
  • Mature and stable
Challenges:
  • Kernel bugs can crash system
  • Large attack surface

QNX

Microkernel
Automotive, medical devices
Strengths:
  • Hard real-time guarantees
  • Fault isolation
  • Safety certification
Challenges:
  • IPC overhead
  • Complex development

Windows

Hybrid
Desktop, enterprise servers
Strengths:
  • Good performance
  • Modular subsystems
  • Rich ecosystem
Challenges:
  • Complex architecture
  • Legacy compatibility

Key Design Principles

Fundamental principles that guide OS architecture decisions

Separation of Concerns

Each component should have a single, well-defined responsibility

Example: File system separate from network stack

Abstraction

Hide complexity behind simple, consistent interfaces

Example: System calls hide hardware details

Modularity

Design components that can be developed and tested independently

Example: Loadable kernel modules

Performance

Minimize overhead in critical paths

Example: Direct hardware access for drivers

Reliability

Isolate failures and provide recovery mechanisms

Example: Process isolation, fault tolerance

Security

Enforce access control and protect system integrity

Example: User/kernel mode separation

Key Takeaways

No silver bullet - each architecture has trade-offs
Hybrids dominate general-purpose systems
Microkernels shine where safety and reliability matter
Distributed OS extends the idea across nodes
Exokernel remains a research path informing cloud/unikernel thinking

Questions & Answers

Thank you for your attention!

Happy to take questions about OS architectures

Presented by: JOMARI REBADULLA

MASTERS IN INFORMATION TECHNOLOGY

ADV OPERATING SYSTEM & NETWORK MANAGEMENT

Adamson University "Lux et Veritas" - Light and Truth Manila, Philippines

Modern Operating System Architectures

CS 4XX - Advanced Operating Systems

Academic Year 2024-2025