This repository contains a Jupyter notebook with a discrete-event simulation of message-queue processing built on SimPy.

• The queue contains 200,000 messages.
• 20 threads/workers process the queue.
• Each worker picks up one message, processes it, and then picks up the next.
• The processing time of a single message follows a log-normal distribution:
  • 50th percentile: 1.7 seconds;
  • 95th percentile: 3.6 seconds.

• the total throughput of all workers, in messages per second and per minute;
• the per-worker throughput, in messages per second and per minute;
• the time it takes to drain the whole queue;
• how the processing rate evolves over time.

The notebook plots:

• total processing rate in messages per second;
• total processing rate in messages per minute;
• per-worker processing rate;
• the distribution of message processing times.

Click the Open in Binder badge above, wait for the environment to start, and run all cells in simpy_message_processing.ipynb.

Run locally:

pip install -r requirements.txt
jupyter lab

Then open simpy_message_processing.ipynb and run all cells.