Thanks, sidecars, like envoy and linkerd-proxy, already do protocol detection even if it is just for metrics. Maybe their stack can be augmented to do latency a

Thanks, sidecars, like envoy and linkerd-proxy, already do protocol detection even if it is just for metrics. Maybe their stack can be augmented to do latency and fault stimulation. For plain http nginx might be an option indeed.