I am aware of the following general protocol properties for inter-system (task,process,device,etc) communication, both computing unit address/cache-local and distributed: - Local system bus/IPC interfaces/system challenges 1. CIA 2. (re-)identification 3. performance requirements (latency, throughput, bandwidth) 4. semi-reliable systems schedule with sane error behavior 5. debugging concept (simulation, reversible computing, recording, tracing, stepping, validation etc) - Distributed system bus/interface system challenges same as local ones, but with typical distributed system properties on top 1 idempotency 2 consistency 3 backpressure (4 observability) 5 fault-tolerance (6 service discovery) (less often) 7 consensus 8 replication One problem is that most of (for example) TCP and run-time behavior is hidden from the user, so you have to guesstimate, if it works. I'm only aware of bare-bone hypervisors and micro-kernels making run-time behavior explicit (all Kernels go into that direction for IO, see io_uring etc), but I have no numbers on how reliable NICs or peripherials are (or can be used) outside of SPS field buses.