To be able to deploy a successful distributed stream application,
there are 4 pillars
- the business functionality: implemented using the Nuxeo Stream processor
- the instrumentation plane: metrics and observability, proving that the job is done
- the control plane: how to scale up and down.
- the operation plane: how errors are handled.
Most of the stuff is there but there are some missing details to reach this:
- be able to scale up or down on a specific computation inside a processor
- be able to stop a Nuxeo instance when a processor is starving to ease the scale down
Creating a small POC to demoing these aspects will enable to identify the missing part and can be used as a playground for training