Fault tolerance aware workload resource management technique for real‐time workload in heterogeneous computing environment

Abstract

AbstractWorkload execution is composed by mapping larger tasks onto heterogeneous environments such as cloud platforms for enhancing the efficiency of workload resource management techniques. Execution of scientific workflow on a cloud platform is time‐consuming, expensive, and requires a fault‐tolerance guarantee. The existing methodology has emphasized on minimizing processing time to reduce costs. However, the processing cost can be reduced by minimizing energy consumption. In providing fault‐tolerance while meeting the workload quality of service requirement the task is offloaded to a new physical machine; Hence, this increases energy consumption and thereby increases the cost of workload execution. In addressing the research challenges this paper presents the fault‐tolerant aware (FTA) workload resource management (WRM) technique. First, the FTA‐WRM optimizes processing and communication costs as an energy constraint leveraging the dynamic voltage frequency scaling technique. Then, a task offloading mechanism is modeled as an energy constraint with an application delay requirement for providing fault tolerance into the FTA‐WRM. The experiment outcome shows the FTA‐WRM significantly improves processing and energy efficiency in comparison with the existing workload resource management technique.