Stress-Testing MQTT Brokers: A Comparative Analysis of Performance Measurements

Presently, Internet of Things (IoT) protocols are at the heart Machine-to-Machine (M2M) communication. Irrespective radio technologies used for deploying an IoT/M2M network, all independent data generated by IoT devices (sensors and actuators) rely heavily on special messaging M2M communication in applications. As demand services is growing, need reduced power consumption also growing to ensure a sustainable environment future generations. The Message-Queuing Telemetry Transport or short MQTT widely protocol. It low-resource-consuming solution based publish–subscribe type model. This paper aims assess performance several broker implementations (also known as servers) using stress testing, analyze their relationship with system design. evaluation brokers performed realistic test scenario, analysis results done three different metrics: CPU usage, latency, message rate. main contribution this work, we analyzed six (Mosquitto, Active-MQ, Hivemq, Bevywise, VerneMQ, EMQ X) detail, classified them properties. Our showed that Mosquitto outperforms other considered solutions most metrics; however, ActiveMQ best performing one terms scalability due its multi-threaded implementation, while Bevywise has promising resource-constrained scenarios.