Nyx: A Fault-Tolerant Fully Distributed Message Queue for the IoT-Edge-Cloud Continuum
Juan Aznar-Poveda, Marlon Etheredge, Stefan Pedratscher, Patrick Künig, Thomas Fahringer · 2025
Message queues play a critical role in enabling reliable, asynchronous communication across distributed systems, particularly within the IoT, Edge, and Cloud continuum, where workloads are highly dynamic and decentralized. Traditional message queue systems often rely on centralized brokers, which can introduce scalability bottlenecks and single points of failure. In this paper, we present Nyx, a novel distributed message queue that eliminates the reliance on individual brokers to provide fault-tolerant, scalable, and robust communication across heterogeneous nodes. Nyx leverages RAFT-based broker clusters to distribute message processing responsibilities across all participating nodes, ensuring efficient and resilient operation. We conduct a comprehensive performance evaluation of Nyx using geo-distributed clusters spanning the IoT-Edge-Cloud continuum and compare it against existing message queue systems. Experimental results demonstrate that Nyx achieves up to 55% lower latency and 90 % higher throughput than state-of-the-art baselines, highlighting its effectiveness in meeting the demands of modern distributed applications.