文摘
Cloud Data Centres (CDCs) are facilities used to host large numbers of servers, networking and storage systems, along with other required infrastructure such as cooling, Unsupervised Power Supplies (UPS) and security systems. With the high proliferation of cloud computing and big data, more and more data and cloud-based service solutions are hosted and provisioned through these CDCs. The increasing number of CDCs used to meet enterprises’ needs has significant energy use implications, due to power use of these centres. In this paper, we propose a method to accurately predict workload in physical machines, so that energy consumption of CDCs can be reduced. We propose a multi-way prediction technique to estimate incoming workload at a CDC. We incorporate user behaviours to improve the prediction results. Our proposed prediction model produces more accurate prediction results, when compared with other well-known prediction models.