A power-aware scheduling algorithm for real-time workflow applications in clouds

Cloud computing has become increasingly prevalent, providing end-users with temporary accesses to scalable computational resources. For the sake of scalability and efficiency, an increasing number of DAG-type workflow applications are deployed in cloud platforms. The energy problem has become one of the major concerns in clouds. We proposed a power-aware and real-time scheduling (PRTS) algorithm for cloud schedulers to minimize the execution cost of workflow and reduce the energy consumption. The PRTS algorithm consists of two parts: scheduling the tasks to the most cost-effective virtual machines based on critical path without missing the deadline; monitoring the dynamic slack and reclaiming them to apply the DVFS technique for energy-saving. Finally, the simulation experiments reveal that slowdown factor technique can achieve the percents of energy-saving up to 12.3% (with an average of 8.3%) compared with the baseline algorithm ESS.