Features, yes, provided one has time to actually think of the protocol and express it in terms of some formalism (CSP, TLA+) to at least informally sketch a pro

Features, yes, provided one has time to actually think of the protocol and express it in terms of some formalism (CSP, TLA+) to at least informally sketch a proof of major global properties. Or do some extensive testing (https://aphyr.com/posts/291-jepsen-zookeeper https://aphyr.com/posts/291-jepsen-zookeeper https://aphyr.com/posts/316-jepsen-etcd-and-consul https://aphyr.com/posts/316-jepsen-etcd-and-consul ).