Improving big data centers energy efficiency: Traffic based model and method

Abstract

Green aspects of data centers (DCs) which provide the appropriate hosting services and data storage are considered. Various approaches to transmission control of integrated data streams (IDS), which are the result of traffic aggregation, being transmitted in a DC related to Big Data processing, are analyzed. It is proposed a complex, which includes several models and a method; it uses traffic samples of several sessions as input data to perform short-term traffic prediction. For the case of elastic traffic, the application of the developed complex will reduce the DC’s resources usage, which are required to process data, down by 3%. As a result of the resources reallocation of the DC on the basis of the short-term prediction, the gain on the effectiveness of their use with such type of traffic will be 4–8%. Application of the proposed complex allows using the resources of the DC in more efficient way (reduces transmission time and, correspondingly, increases energy efficiency of the whole system).