multi_server.py

#!/usr/bin/env python3
"""
Multi-Server Example

This example demonstrates a multi-server queueing system
and compares performance with different numbers of servers.
"""

import pdq


def analyze_multiserver(servers, users, think_time, service_time):
    """Analyze a multi-server system with given parameters"""

    model_name = f"MultiServer_{servers}_servers"
    model = pdq.PDQModel(model_name)

    # Create multi-server node (MSC for closed networks)
    model.create_closed("Users", pdq.TERM, users, think_time)
    model.create_multinode(servers, "CPUPool", pdq.MSC, pdq.FCFS)
    model.set_demand("CPUPool", "Users", service_time)

    # Solve using exact method for closed networks
    model.solve(pdq.EXACT)

    # Get metrics
    response_time = model.get_response(pdq.TERM, "Users")
    throughput = model.get_throughput(pdq.TERM, "Users")
    utilization = model.get_utilization("CPUPool", "Users", pdq.TERM)

    return {
        'servers': servers,
        'response_time': response_time,
        'throughput': throughput,
        'utilization': utilization,
        'utilization_pct': utilization * 100
    }


def main():
    """Run multi-server comparison example"""

    print("=== Multi-Server Comparison Example ===\\n")

    # Model parameters
    users = 20
    think_time = 5.0  # seconds
    service_time = 2.0  # seconds per request

    print(f"Model Parameters:")
    print(f"  Users: {users}")
    print(f"  Think time: {think_time} sec")
    print(f"  Service time: {service_time} sec")
    print()

    # Compare different numbers of servers
    server_counts = [1, 2, 3, 4, 5]
    results = []

    print("Analyzing different server configurations...")
    print()

    for servers in server_counts:
        result = analyze_multiserver(servers, users, think_time, service_time)
        results.append(result)

        print(f"Servers: {servers}")
        print(f"  Response time: {result['response_time']:.3f} sec")
        print(f"  Throughput: {result['throughput']:.3f} req/sec")
        print(f"  Per-server utilization: {result['utilization_pct']:.1f}%")
        print()

    # Show comparison table
    print("=== Performance Comparison ===")
    print("Servers | Response Time | Throughput | Utilization")
    print("--------|---------------|------------|------------")

    for result in results:
        print(f"{result['servers']:7d} | "
              f"{result['response_time']:11.3f}s | "
              f"{result['throughput']:8.3f}/s | "
              f"{result['utilization_pct']:9.1f}%")

    print()

    # Calculate improvement ratios
    base_response = results[0]['response_time']
    base_throughput = results[0]['throughput']

    print("=== Improvement over Single Server ===")
    print("Servers | Response Ratio | Throughput Ratio")
    print("--------|----------------|------------------")

    for result in results:
        response_ratio = base_response / result['response_time']
        throughput_ratio = result['throughput'] / base_throughput

        print(f"{result['servers']:7d} | "
              f"{response_ratio:12.2f}x | "
              f"{throughput_ratio:14.2f}x")

    print()

    # Show detailed report for optimal configuration
    print("=== Detailed Report for 3-Server Configuration ===")
    model = pdq.PDQModel("Optimal MultiServer")
    model.set_comment("Multi-server system with optimal configuration")
    model.create_closed("Users", pdq.TERM, users, think_time)
    model.create_multinode(3, "CPUPool", pdq.MSC, pdq.FCFS)
    model.set_demand("CPUPool", "Users", service_time)
    model.solve(pdq.EXACT)
    model.print_report()


if __name__ == "__main__":
    main()