Abstract:
With a rise in demand for centralized data processing servers and peer to peer communication application, data centers have become a key stakeholder in upkeeping of such applications. The main hinderance in the development of data centers is its operational cost. In developing countries, surge in operational is a leading factor behind unavailability of data centers. Mainly, consumption of energy within data centers is by server racks. These server racks generate tons of heat. If the heat increases from a certain limit, it can cause physical damage to the costly computing equipment within the servers, halting the ongoing operations of a data center. Server rooms are cooled with heavy duty air conditioners. These air conditioners are second in line in terms of energy consumption. In case of small and medium scale data centers no significant and cheap way for the reduction of electricity cost is introduced. Hence our research focuses on developing and validating a model for such internet data centers. In our model we reduced the power consumption by automating the on/off control mechanism of CRACs and computing nodes and through thermal aware load balancing of workload. To estimate the power consumption and for simplification of mathematical complexity we used two different techniques of fuzzy inference system. Our model was able to show significant reduction in the energy consumption of data centers.
