Where process clustering deals with running instances, machine clustering deals with physical hardware and task delegation between machines.

In short, machine clustering is when you take the different services that makes up your service-stack (read: collection of services) and physically place them on different machines. This way Node.js is able to maximize the performance of the process clustering, typically allocating a maximum of 3 workers per CPU core at any given time.


This type of setup requires the machines that make up your stack to all be available on the same network, for example plugged into the same router, or occupying the same switch.


This is where network coding become exciting - because in order to have a reliable machine cluster, your services must be written to communicate purely through the network.

But once that is achieved, the service can literally be hosted anywhere. You can place all the services on a single computer, or spread them out over multiple computers for better performance.