Secure Aware Optimized Support Vector Regression Models Based Host Overload Detection in Cloud

The increasing need for high-performance computing, preservation, and networking capabilities to support corporate and scientific applications is driving a rapid expansion in the use of cloud computing server farms. Virtual machine (VM) consolidation plays a crucial role in this context, involving the direct migration of VMs from underutilized physical servers to optimize power consumption efficiency, operational costs, and reduce CO2 emissions. A pivotal step in VM consolidation is the detection of host overload, which aims to predict potential server over-subscription with VMs. This paper introduces an Optimized Support Vector Regression model for overloaded detection. To enhance the Support Vector Regression (SVR) performance, optimal selection of SVR parameters is achieved using the Enhanced Tasmanian Devil Optimization algorithm. Following overload detection, VM migration occurs, but this process raises concerns about system integrity and data confidentiality. To address these concerns, data is encrypted using the Cyclic Shift Transposition Algorithm before migration. The proposed approach's performance is evaluated across various metrics such as energy consumption, ESV, Migration, and SLA X0.001, and its effectiveness is compared with different existing methods.