sFlow-based resource utilization monitoring in clouds

Monitoring of infrastructural resources against a huge number of virtual machines in clouds is a challenging task to system administrators. System management tools are used to collect and provide the utilization of a variety of performance counters of each virtual machine. Such dispersed views cannot reveal the actual performance behavior of virtual machines. To this purpose, a monitoring framework is necessary particularly since cloud hosts are subject to varying load conditions. In this paper, we propose a framework based on sFlow to interpret the resource utilization correctly and in real-time to ease the workload of system administrators.