An algorithm for minimizing energy consumption with reliability goal on multiple processors in a cloud computing environment

The energy and time constrained task scheduling on multi-processing environments with different frequency levels is considered as an important optimization issue in a cloud computing. In addition, a processor executed with a reduced frequency will dynamically increase transient faults, which will weaken the reliability of the applications. Reliability is an important figure of merit of the system and it must be satisfied in safety-critical applications. In this paper, the parallel task scheduling problem in multi-processors has been explored to reduce energy and to increase reliability in the scheduling. Multi-processing and Dynamic Voltage and Frequency Scaling (DVFS) techniques can decrease the processor’s frequency level as much as possible; therefore, the voltage consumption of the processor will be reduced. Here, a two-phase algorithm is proposed to minimize energy consumption with reliability goal on multiple processors. The first phase is for initial assignment and the second phase is for either satisfying the reliability goal and improving energy efficiency. Specifically, when the application’s reliability goal cannot be achieved via initial assignment, based on our defined current reliability ratio, an enhanced algorithm is designed to satisfy application’s reliability goal. The proposed algorithm compared with existing algorithms. Experimental results demonstrate that the proposed algorithm consume less energy while satisfying the application’s reliability goal.