network-simulator

A multi-server queueing system simulator written in Python.

Models M/M/c queues with probabilistic routing: inter-arrival times and service times are exponentially distributed (Poisson arrivals, exponential service). Incoming tasks are routed to one of n servers by a probability vector and dropped if the target queue is at capacity.

Queueing model

Arrivals (rate λ, exponential inter-arrival)
         │
         ▼
     ┌────────┐
     │ Router │  probabilistic routing (p₁ … pₙ, Σpᵢ = 1)
     └────────┘
      /   |   \
     ▼    ▼    ▼
  [S₁] [S₂] [Sₙ]   capacity Kᵢ, service rate μᵢ (exponential)

• Arrival is routed to server i with probability pᵢ
• If that server's queue is full, the task is rejected (finite-buffer loss)
• Service times are independent exponential per server

Usage

python3 main.py <T> <n> <p1> ... <pn> <λ> <K1> ... <Kn> <μ1> ... <μn>

Parameter	Description
T	Simulation duration
n	Number of servers
p1..pn	Routing probabilities (must sum to 1.0)
λ	Global arrival rate
K1..Kn	Queue capacity per server
μ1..μn	Service rate per server

Example

Two servers, 60/40 routing split, arrival rate 5.0, capacity 10 each, service rates 3.0 and 4.0:

python3 main.py 1000 2 0.6 0.4 5.0 10 10 3.0 4.0

Output (space-separated):

<accepted> <rejected> <sim_end_time> <avg_wait_time> <avg_service_time>

Output metrics

Field	Description
accepted	Tasks admitted to a server queue
rejected	Tasks dropped due to full queue
sim_end_time	Time when the last queued task completes
avg_wait_time	Mean time a task waits before service begins
avg_service_time	Mean service duration

Requirements

• Python 3.x

License

MIT