The Fault-Tolerant Cluster-Sending Problem

The emergence of blockchains is fueling the development of resilient data management systems that can deal with Byzantine failures due to crashes, bugs, or even malicious behavior. As traditional resilient systems lack the scalability required for modern data, several recent systems explored using sharding. Enabling these sharded designs requires two basic primitives: a primitive to reliably make decisions within a cluster and a primitive to reliably communicate between clusters. Unfortunately, such communication has not yet been formally studied.In this work, we improve on this situation by formalizing the cluster-sending problem: the problem of sending a message from one resilient system to another in a fault-tolerant manner. We also establish lower bounds on the complexity of cluster-sending under both crashes and Byzantine failures. Finally, we present worst-case optimal cluster-sending protocols that meet these lower bounds in practical settings. As such, our work provides a strong foundation for the future development of sharded resilient data management systems.