DESDSim: A dynamic energy-saving deployment simulation platform for web server cluster

To facilitate researchers to evaluate a web cluster dynamic energy-saving deployment strategy or choose key parameters for it, this paper proposes a CouldSim-based simulation platform called DESDSim. It can simulate web cluster´s actual operation, web request´s lifecycle, and three important energy-saving measures, namely server dynamic switching on/off, CPU core´s dynamic switching on/off, and CPU dynamic frequency scaling. Rational design of interfaces and classes allows users to easily define their own request load features, energy-saving deployment algorithm, and request scheduling algorithm. Some important indicators, such as average response time, drop rate, and energy consumption, can be calculated by the platform. Specially, for considering various possible states and substates of server, the energy consumption calculation is more accurate. In addition, a threshold-based greedy deployment strategy is proposed and simulated by the platform. It bases on thresholds and greedy idea to schedule requests and adjust cluster´s deployment. The simulation results accord with our expectations for the strategy on both energy saving and QoS, which demonstrates the feasibility of DESDSim.