Reduce resource usage
With just 1 worker but more threads. Keeping the worker fork model so it is easy to scale up busy instances. The memory limit allows for up to 3-4 workers using 256MB each but it is likely that the app will leak memory over time and workers will use more.