Resource-Intensive Fuzzing for MQTT Brokers: State of the Art, Performance Evaluation, and Open Issues

Abstract

Pub/sub messaging is a promising design pattern that relies on a centralized component called the broker, which routes messages to different devices. Due to its prominent role, the broker must be capable of providing high quality services under heavy workload. In this letter, we evaluate the resource consumption of state-of-the-art fuzzing frameworks, thereby understanding the degree to which brokers are tested before deployment. Our results attest that only one framework shows the most promise for memory-intensive testing, outperforming its counterparts by more than 20%. We conclude this letter by highlighting shortcomings that prevent existing frameworks from consuming considerable system resources.