Deadline-aware and energy-efficient dynamic flow scheduling in data center network

Abstract

The construction of energy-efficient network and achievement of green communication have garnered great attention as a promising a way to reduce network operating costs and C emissions. Moreover, recently the deadline-aware and energy-efficient routing and scheduling algorithms in data center network have been attracting a broad attention. However, the dynamic scheduling for flows has not been explicitly studied by the existing research. In this paper, we investigated the dynamic flow scheduling in data center network, and propose a deadline-aware and energy-efficient dynamic flow scheduling (DEDFS) algorithm, assuming the path of the flow could be calculated in advance and pre-stored. In addition, the number of mouse flows in data center network accounts for main proportion, but consumption is very small. In order to achieve the balance of energy-saved and efficiency, mouse flows will be directly transferred, while elephant flows will be scheduled by the Most-Critical-First static strategy based dynamic scheduling algorithm. It selects the interval of largest energy consumption density as the critical interval, and all of the flows in this critical interval will be preferentially scheduled. Finally, the feasibility and validity of the algorithm are verified by simulation.