Abstract
Internet scale distributed consensus takes advantage of randomized algorithms to cope with functional resilience, reliability and access while accomplishing decentralized agreement at expenses of costly unicast communication. This paper provides experimental insights on randomized communication patterns for distributed consensus systems. We define, measure and approximate communication overhead against key performance indicators like latency and communication overhead, leading to a discussion on how network innovations may mitigate the identified issues. While this paper does not promote specific solutions or DLT implementations, our initial insights invite the wider community working on DLT and network solutions alike to deepen those insights to aid future research and development into solutions, concepts, and technologies.
