FMonE: A Flexible Monitoring Solution at the Edge

Figure 5

Several Docker containers are involved in an FMonE pipeline. To extract the metrics, the Docker image for FMonE is started in each host, with docker as the Inplugin parameter, inout for the MidPlugin, and rabbitmq as the Outplugin. The RabbitMQ container will also be created as part of the pipeline and its hostname used as a parameter for the previous outplugin. The aggregation will be performed by a different FMonE agent that will pull the metrics with its corresponding rabbitmq Inplugin, average them with the average Midplugin, and store them in a database with the mongodb Outplugin. The pipeline will be completed with the creation of a containerized MongoDB back-end. Marathon and Mesos are used to start these containers in the desired regions.