Elastic Multi-Controller based BCube Connected Crossbars (BCCC) for Higher Energy Efficiency

Abstract

It is a challenging task to propose an efficient cost-effective network topology to provide consistent performance for large networks consists of more number of Servers and Nodes. From the literature survey, this Research Work noted a few popular Server-Centric Network Topologies such as BCube, Bidimensional Compound Network (BCN) and FiConn were proposed. However, it was noticed that these topologies are not expandable and demand high cost during expansion. This is the major issue and concern to expand or upgrade the Data Centre. With the massive growth in the modern Data Centres, the energy consumed by the network constitutes a significant amount of total power cost in the data centre. One of the straightforward and efficient ways for varying the size of DCN in accordance with traffic demands is by turning OFF the idle network devices to reduce the energy consumption. The Software Defined Networking (SDN) is the current promising solution for controlling the resources of DCN. But, the SDN has not been leveraged completely to enable the power saving in large-scale DCN. To address the above mentioned issues, it is highly needed and important to propose an efficient model that needed to improve the power efficiency for Data Centre Network (DCN). This research work proposed a Hybrid Model called Elastic MultiController (E3MC) based BCube Connected Crossbars (BCCC) Model to achieve high energy efficiency. The proposed E3MC based BCCC Model was implemented simulated and studied thoroughly. From the simulation results, it was noticed that the proposed model achieves high performance as compared with the existing BCCC in terms of Energy Efficiency, Path length, Server Resource Utilization, Throughput, Link Failure Rate and Server Failure Rate.