The industrial IoT paradigm is a hierarchy of worker nodes managing local IoT devices and rolling up to a manager of managers, aka Federation Model. With RPi I see some contention between all the daemons and processes and that impacts responsiveness. With the understanding that SBC’s are supposed to be “cheap” (ignoring the current RPi4 supply chain issues), whats wrong with a dedicated SBC on a per-protocol basis with each one connecting to an also cheap ethernet switch? I say use the cheap, work it, abuse it, don’t fear separate hardware to eliminate contention and improve response time. I see a cluster of 3 RPi4’s right now, one for BT/BLE, one for Zwave/Zigbee, and one for ethernet zeroconfig/broadcast/multicast/ipv6. As an architect I look at things pretty high level and rely on engineers to say whats possible. Would you roll up to a client or a 4th RPi? I think you could kind of do a lot of it now with existing HA by only installing the BT/BLE integrations on one RPi/HASS, Zwave/Zigbee on another, and ethernet zeroconf stuff on a third but what’s missing is the rollup web page. Food for thought. Maybe someone has already had this idea.