VMMISD: An Efficient Load Balancing Model for Virtual Machine Migrations via Fused Metaheuristics With Iterative Security Measures and Deep Learning Optimizations
Madala Guru Brahmam,R Vijay Anand,Vijay Anand R
DOI: https://doi.org/10.1109/access.2024.3373465
IF: 3.9
2024-03-19
IEEE Access
Abstract:As cloud computing continues to evolve, the need for efficient and secure management of virtual machine (VM) migrations has become increasingly evident. Traditional models often fall short in optimizing load balancing and energy consumption while ensuring a high level of security. In this work, we propose the load balancing and energy-efficient migration model, an innovative approach that leverages bioinspired algorithms and advanced security measures to optimize VM migrations. The initial novelty of our model is the integration of Genetic Algorithms with Ant Colony Optimization for resource scheduling operations. These algorithms were specifically chosen for their proven effectiveness in solving complex optimization problems by simulating natural processes. Additionally, our model incorporates a deep reinforcement learning-based iterative-learning contextual side chaining model to enhance security measures. This approach not only learns and adapts to new security threats over time but also utilizes contextual side-chaining to link related security events, thereby providing a robust defense mechanism against potential threats. The affinity between VMs and physical machines is quantified using K Means clustering and fuzzy logic, which ensures optimal load balancing while accounting for the uncertainty inherent in the migration process. Furthermore, we employ bidirectional long short-term memory networks with recurrent graph neural network, for accurate workload prediction and informed migration decision making process. The selection of these techniques is grounded in their proven capability to analyze historical data and predict future trends with high accuracy levels. Our proposed model demonstrates marked improvements in several key performance metrics. We achieved a 4.5% reduction in makespan, a 4.9% increase in deadline hit ratio, and a 3.9% improvement in task diversity. Furthermore, computational complexity was reduced by 8.3%, VM migration efficiency improved by 2.5%, and the delay of computation was significantly reduced by 9.5%. Importantly, the integration of the Iterative-learning Contextual Side chaining Model significantly enhanced the security and quality of service (QoS) under attack scenarios, resulting in a 10.4% improvement in response speed, a 2.5% reduction in energy consumption during block mining, a 3.9% improvement in throughput, and an 8.5% reduction in storage costs. This load balancing and energy-efficient migration model represents a significant advancement in addressing the challenges of load balancing, energy efficiency, and security in VM migrations. Through the meticulous integration of bioinspired algorithms, advanced security measures, and machine learning techniques, our model provides a comprehensive and innovative solution that markedly improves system performance, reduces energy consumption, and fortifies security, thereby paving the way for a more efficient and secure cloud computing ecosystem.
computer science, information systems,telecommunications,engineering, electrical & electronic