>
All I need is to figure out a real world applications. Open to suggestions
There are several properties to this system: (1) small size, (2) small energy costs, (3) multiple cores on multiple boards, (4) small computational powers. The size, energy, #core, #flops are absolute measures; 'small' is a relative measure (e.g. smaller than most of its alternatives). The applications can be clustered these characteristics; meaning, that a ClusterHat-based system:
* can be put on a small box of an unused space because of its small size;
* this box can serve regardless of energy bills because of its small power consumption;
* assuming you have monitoring and automation control, it is a partially stable server (stable because can automatically restart crashing apps, partially because of multiple single-point failures, e.g. single power source, single host-RPi, maybe single points of failures in the clusterhat itself.
* the services benefit from and from parallel computing,
As you can see, these are used in non-functional definition of applications, yet these can stimulate idea generation and selection of the functional apps to address personal needs. In any case, if you would work systematically, you would have to measure your system according to a sub-set of these criteria, e.g. is hardware fitting for the app. If you do not have functional ideas, you can familiarise yourself with testing hardware computational limits, build a model to check if this hardware would be fitting for a hypothetical app, familiarise yourself with generic app orchestration (e.g.
https://microk8s.io/ ).
Thus my suggestion is: if you do not know the specific product, then go for the generic technology. It will make you a stronger engineer, and you'll have a better toolbox for functional work in the future, maybe in smart-home or industrial automation that require these non-functional requirements.
Kind regards,
Dmitry.