Skip to content

LOAD_BALANCING_TESTS.md

Latest commit

 

History

History
289 lines (230 loc) · 7.42 KB

LOAD_BALANCING_TESTS.md

File metadata and controls

289 lines (230 loc) · 7.42 KB

Load Balancing Test Scripts

This directory contains comprehensive load balancing test scripts for the Microservices API Gateway project.

📁 Files Overview

Main Scripts

  • test-load-balancing.sh - Main load balancing test script
  • setup-load-test.sh - Environment setup helper
  • cleanup-load-test.sh - Post-test cleanup script

🚀 Quick Start

1. Prepare Environment

./setup-load-test.sh

2. Run Load Balancing Tests

./test-load-balancing.sh

3. Clean Up After Testing

./cleanup-load-test.sh

🧪 Test Types

Option 1: Quick Test

  • Uses existing single service instances
  • Tests basic load balancing functionality
  • No additional setup required
  • Good for development and CI/CD

Option 2: Multiple Instance Test

  • Starts multiple instances of selected services
  • Tests true load distribution across instances
  • Demonstrates horizontal scaling
  • Recommended for comprehensive testing

Option 3: Concurrent Load Test

  • Tests system under concurrent user load
  • Measures response times and success rates
  • Good for performance benchmarking

Option 4: Full Comprehensive Test

  • Combines multiple instance and concurrent testing
  • Most thorough load balancing validation
  • Includes detailed statistics and analysis

Option 5: Test Existing Instances

  • Tests currently running service instances
  • Good for quick validation without setup

📊 Test Features

Load Distribution Analysis

  • Request Distribution: Shows how requests are distributed across instances
  • Balance Verification: Checks if load is evenly distributed
  • Instance Identification: Tracks which instance handled each request

Performance Metrics

  • Response Times: Min, max, and average response times
  • Success Rates: Percentage of successful requests
  • Throughput: Requests per second measurements
  • Concurrent Users: Simulation of multiple users

Eureka Integration

  • Service Discovery: Verifies instances are registered in Eureka
  • Instance Health: Checks instance health status
  • Dynamic Scaling: Tests addition/removal of instances

🔧 Configuration

Environment Variables

API_HOST=localhost          # API Gateway host
GATEWAY_PORT=8080          # API Gateway port
DISCOVERY_URL=http://localhost:8761  # Eureka server URL

Test Parameters

  • Request Count: Number of requests per test (default: 20-30)
  • Concurrent Users: Number of simultaneous users (default: 5)
  • Instance Count: Number of service instances to start (default: 3)

📈 Understanding Results

Load Distribution Results

Request | Status | Response Time | Instance Info
--------|--------|---------------|---------------
      1 |    200 |         0.234s | user-service-1
      2 |    200 |         0.187s | user-service-2
      3 |    200 |         0.201s | user-service-3

Distribution Summary

Total Requests: 30
Successful: 30
Failed: 0
Success Rate: 100.00%

Requests per Instance:
  user-service-1      : 10 requests (33.3%)
  user-service-2      : 10 requests (33.3%)  
  user-service-3      : 10 requests (33.3%)

What Good Results Look Like

✅ Balanced Distribution: Requests evenly spread across instances
✅ High Success Rate: >95% successful requests
✅ Consistent Response Times: Low variance in response times
✅ All Instances Used: All registered instances receive requests

Warning Signs

⚠️ Uneven Distribution: One instance handling most requests
⚠️ High Error Rate: >5% failed requests
⚠️ High Response Times: Significant delays or timeouts
⚠️ Single Instance: Load balancer not distributing load

🔍 Monitoring During Tests

Real-time Monitoring

# Watch service status
watch ./service-manager.sh status all

# Monitor logs
tail -f logs/multi-instance/*.log

# Check Eureka dashboard
firefox http://localhost:8761

System Resources

# Monitor CPU usage
htop

# Monitor memory
watch free -h

# Monitor network
netstat -tuln | grep :808

🐛 Troubleshooting

Common Issues

Load Balancing Not Working

Problem: All requests go to single instance
Solution: 
- Check Eureka registration: http://localhost:8761
- Verify Spring Cloud LoadBalancer is enabled
- Check API Gateway routing configuration

High Error Rates

Problem: Many 5xx errors during testing
Solution:
- Check service logs for errors
- Verify adequate system resources
- Reduce concurrent load
- Check database connections

Instances Not Starting

Problem: Multiple instances fail to start
Solution:
- Check port availability
- Verify Java/Maven installation
- Check service dependencies
- Review application.yml configuration

Debug Commands

# Check running Java processes
jps -l

# Check port usage
netstat -tulpn | grep :80

# Check service logs
tail -f logs/discovery-server.log
tail -f logs/api-gateway.log

# Test direct service access
curl http://localhost:8761/eureka/apps

📚 Advanced Usage

Custom Load Patterns

Modify the test script to implement custom load patterns:

# Burst load pattern
for i in {1..50}; do
    curl "$endpoint" &
done
wait

# Gradual ramp-up
for users in {1..10}; do
    test_concurrent_load "$endpoint" $users 5
    sleep 10
done

Integration with CI/CD

# Example GitHub Actions workflow
- name: Load Balancing Test
  run: |
    ./start-services.sh
    sleep 30
    ./test-load-balancing.sh <<< "5"
    ./cleanup-load-test.sh <<< "n"

Performance Benchmarking

# Record baseline performance
./test-load-balancing.sh > baseline-results.txt

# Compare with changes
./test-load-balancing.sh > new-results.txt
diff baseline-results.txt new-results.txt

🏗️ Architecture Notes

Load Balancing Algorithm

  • Default: Round-robin distribution
  • Configurable: Can be changed in Spring Cloud LoadBalancer configuration
  • Health-aware: Automatically excludes unhealthy instances

Service Discovery Integration

  • Registration: Services register with Eureka on startup
  • Health Checks: Eureka monitors instance health
  • Dynamic Updates: Load balancer updates instance list automatically

Scaling Considerations

  • Horizontal Scaling: Add more service instances
  • Resource Limits: Monitor CPU, memory usage
  • Database Connections: Ensure adequate connection pools
  • Network Bandwidth: Consider network capacity

📋 Best Practices

Testing Best Practices

  1. Start Simple: Use existing instances first
  2. Incremental Testing: Gradually increase load
  3. Monitor Resources: Watch CPU, memory, disk I/O
  4. Baseline First: Record normal performance
  5. Clean Environment: Use fresh instances for accurate results

Production Considerations

  1. Circuit Breakers: Implement fault tolerance
  2. Rate Limiting: Protect against overload
  3. Health Checks: Robust health monitoring
  4. Graceful Shutdown: Handle instance restarts properly
  5. Monitoring: Continuous performance monitoring

🤝 Contributing

To add new test scenarios:

  1. Modify test-load-balancing.sh
  2. Add new test functions following existing patterns
  3. Update this documentation
  4. Test with various load patterns
  5. Submit pull request with test results

📞 Support

For issues or questions:

  1. Check service logs first
  2. Verify Eureka dashboard
  3. Review system resources
  4. Check network connectivity
  5. Consult Spring Cloud documentation