Raspberry Pi: Combine the power of up to seven Compute Modules with the Turing Pi Cluster Board

[Redaktion]

The Turing Pi Cluster Board turns up to seven Raspberry Pi Compute Modules into a powerful cluster device. The board can not only be used for network storage, but also as a platform for video games.

https://www.notebookcheck.net/Raspberry-Pi-Combine-the-power-of-up-to-seven-Compute-Modules-with-the-Turing-Pi-Cluster-Board.461964.0.html

[ProDigit]

Do all units work in unison, or do they work as separate entities (like we'd have to install one is per board)?

[Garry]

The question remains does any software exist that can use the parallel advantage or do you have to write it yourself. What would be the practical use of such a device ? I guess you could use it as a graphic rendering device in Linux but your Ram would maybe limit you ? what else do you think could be used ?

[John Morley]

Each Computer Module runs as a separate entity, with its own operating system (HypriotOS is a good choice, as it's essentially a stripped down Rapbian with docker support included) and IP address, as is the case with a typical cluster. On top of this you can run orchestration software, such as Kubernetes (Rancher do a small footprint "k3s" version, specifically for low powered ARM cluster) which maintains high availabilty "pods" (essentially docker containers, allocated across the available resources) in which the application software runs. If there's a docker image available for whatever applications you want to run, then you don't have to write anything and there are high quality tools available, such as Ansible and Helm, to allow you to manage the system. One of the nodes would be designated as master and the other six as workers and you would control the whole cluster through the master node. The main purpose of this device is as an affordable educational tool for learning about Ansible and Kubernetes and running them on real hardware, but it is possible to do useful work with it, too. For example, you could run WordPress or Drupal pods and use it to serve and maintain a wiki.