Energy-Aware Fault-Tolerant Dynamic Task Scheduling Scheme for Virtualized Cloud Data Centers

As clouds have been implemented and widely used in various fields, both the size and the number of cloud data centers (CDCs) are growing rapidly. Serious problems have been raised, such as the inefficient use of resources, high energy consumption, and failure of heterogeneous task execution. The existing studies have aimed to solve these challenging problems separately, but it is difficult to optimize resources and energy efficiency while simultaneously providing fault-tolerance. In this study, a dynamic task assignment and scheduling scheme, namely, the energy-aware fault-tolerant dynamic scheduling scheme (EFDTS), is developed to coordinately optimize resource utilization and energy consumption with a fault tolerant mechanism. In the task assignment scheme, a task classification method is developed to partition the coming tasks into different classes and then allocate them to the most suitable virtual machines based on their classes to reduce the mean response time while considering energy consumption. Replication is used for the fault tolerance to minimize the task rejection ratio caused by machine failure and delay. An elastic resource provisioning mechanism is designed in the context of fault-tolerance to improve resource utilization and energy efficiency. Furthermore, a migration policy is developed that can simultaneously improve resource utilization and energy efficiency. The experimental results show that compared with existing techniques, EFDTS significantly improves the overall scheduling performance, achieves a higher degree of fault tolerance with high CDC resource utilization, minimizes the mean response time and task rejection ratio, and reduces energy consumption.