Cloud OS Process Synchronization Simulation

This project simulates and analyses process synchronisation in cloud operating systems, focusing on semaphore-based approaches to prevent race conditions in multi-container environments.

Overview

Modern cloud environments present unique challenges for process synchronisation due to their distributed nature, multi-tenant environments, and variable workloads. This project implements a comprehensive simulation framework to evaluate the effectiveness of semaphore-based synchronisation strategies in preventing race conditions while measuring their performance impact.

Key Features

• Distributed Semaphore Implementation: Scalable semaphore mechanism with configurable permits and timeouts
• Cloud Environment Simulation: Realistic modelling of containerised applications accessing shared resources
• Workload Testing: Various concurrency patterns (low, medium, high, burst, mixed)
• Performance Metrics: Comprehensive data collection including throughput, response time, and conflict rates
• Comparative Analysis: Direct comparison between synchronised and non-synchronised operations

Project Structure

├── src/
│   ├── CloudSyncSimulation.java         # Main simulation framework
│   ├── WorkloadSimulation.java          # Concurrency pattern testing
│   └── SemaphorePerformanceTest.java    # Semaphore configuration analysis
├── results/
│   ├── simulation_metrics.csv           # Raw simulation data
│   ├── semaphore_configurations.csv     # Configuration test results
│   └── workload_test_results/           # Results from different workload patterns
├── docs/
│   └── research_paper.md                # Scientific analysis and findings
└── README.md

Key Findings

Our simulation results demonstrate that:

1. Race Condition Prevention: Semaphore-based synchronisation completely eliminates race conditions across all concurrency levels.

2. Conflict Rates Without Synchronization:

   • Low concurrency (10 clients): 52% of operations experience conflicts
   • Medium concurrency (25 clients): 92% of operations experience conflicts
   • High concurrency (50 clients): 98% of operations experience conflicts

3. Performance Trade-off:

   • Low concurrency: 24% increase in response time with synchronisation
   • Medium concurrency: 511% increase in response time with synchronisation
   • High concurrency: 1452% increase in response time with synchronisation

4. Resource Utilisation: Synchronisation enables optimal resource utilisation without exceeding safe capacity limits.

Getting Started

Prerequisites

• Java Development Kit (JDK) 8 or higher
• Any Java IDE (Intellij IDEA, Eclipse, etc.) or command-line compiler

Running the Simulation

1. Clone the repository:

   git clone https://github.com/yourusername/cloud-os-synchronization.git
   

2. Compile the Java files:

   javac src/*.java -d bin/
   

3. Run the main simulation:

   java -cp bin/ CloudSyncSimulation
   

4. Run specific test scenarios:

   java -cp bin/ WorkloadSimulation
   java -cp bin/ SemaphorePerformanceTest
   

5. Results will be saved to CSV files.

Customising Simulations

You can customise various simulation parameters in the code:

• Number of containers
• Number of resources
• Semaphore permits (max concurrent access)
• Network latency distribution
• Processing time distribution
• Simulation duration

Research Methodology

Our research follows a systematic approach:

1. Design: Implementation of a distributed semaphore with cloud-appropriate features
2. Simulation: Realistic modelling of cloud resource access patterns
3. Measurement: Collection of performance and consistency metrics
4. Analysis: Evaluation of trade-offs between consistency and performance
5. Validation: Comparison with theoretical models and existing research

Conclusion

This project demonstrates that semaphore-based synchronisation is highly effective for preventing race conditions in cloud environments, though with a performance overhead that increases with concurrency. The results provide valuable insights for designing synchronisation strategies in real-world cloud applications.

Acknowledgments

• This project was developed as part of the Operating Systems course (SCOB031) at the University of Limpopo.
• Research draws upon seminal work in distributed systems and cloud computing synchronisation strategies