Host Detection and Classification using Support Vector Regression in Cloud Environment

Having the potential to provide global users with pay-per-use utility-oriented IT services across the Internet, cloud computing has become increasingly popular. These services are provided via the establishment of data centers (DCs) across the world. These data centers are growing increasingly with the growing demand for cloud, leading to massive energy consumption with energy requirement soaring by 63% and inefficient resource utilization. This paper contributes by utilizing a dynamic time series-based prediction support vector regression (SVR) model. This prediction model defines upper and lower limits, based on which the host is classified into four categories: overload, under pressure, normal, and underload. A series of migration strategies have been considered in the case of load imbalance. The proposed mechanism improves the load distribution and minimizes energy consumption and execution time by balancing the host in the data center. Also, it optimizes the execution cost and resource utilization. In the proposed framework, the energy consumption is 0.641kWh, and the execution time is 165.39sec. Experimental results show that the proposed approach outperforms other existing approaches.