Energy Efficiency MapReduce Job Scheduling of Shuffle and Reduce Phases in Data Center

In this paper, job scheduling of shuffle and reduce phases is considered for data center with heterogenous servers to minimize energy consumption. Constructing task list and assigning tasks to slots are designed in a job scheduling framework. The construction of task list considers jobs’ deadlines and tasks’ processing times. Two main steps (candidate servers construction and allocate tasks) are in the proposed assignment. The set of candidate servers is constructed in terms of data size and network topology. Allocation of tasks and slots with normalized shuffle time and data size decreases completion times of jobs, in which shuffle time is calculated by two new bandwidth allocations considering deadlines. Experimental results show that the proposed job scheduling consumes less energy than other existing adapted task scheduling strategies.