Abstract: To perform data acquisition for complex systems like Digital Twins, protocols such as Message Queuing Telemetry Transport (MQTT) are commonly used. We recognize the limitations of this protocol when using high traffic conditions and explore alternatives to increase data flow while minimizing message loss and resource usage. We use a native cluster mode of an MQTT broker implementation, adding a component to the system architecture which we call “Cluster Manager”. This component was developed with the objective of monitoring the health of a cluster of brokers and adjusting the composition of the cluster to the message flow. The results of the implementation of this component helped determine loss margins and latencies in sending critical messages, using different strategies to scale the cluster under representative workloads of Digital Twins applications. We have also studied how different parameters used to scale the system can cause inefficient use of computing resources.

Authors: Rafael Antonio Ventura Trevisan, Francisco Paiva Knebel, Juliano Araújo Wickboldt and Mara Abel

Journal/Conference: Network and Service Management and Operation

Publication: https://dx.doi.org/10.5753/wgrs.2022.223478