A software approach to improving cloud computing datacenter energy efficiency and enhancing security through Botnet detection

摘要

This work presents positive experiment results on the efficiency and security potential of an optimized and novel approach to an Autonomous Management Distributed System (AMDS) running in a Cloud Computing environment. The results validate the AMDS software design and demonstrate its potential as an industrial application to be used in modern datacenters. On one hand, from an operational performance point of view, they show the AMDS' ability of reconfiguring itself on the fly, thus resulting in 14 percent increased efficiency over the lifetime of the first experiment. On the other hand, they show an overall malicious (Botnet) data packet detection rate of over 52 percent, a significant percentage for only 5000 network data samples analyzed by the Botnet software module plugged into the AMDS. Both experiments have been performed in a VMWare run cloud environment, however due to the AMDS' abstract architecture, it has the potential to interface with any existing cloud management system that exposes an API.