Deep Learning and Optimization Approaches in A Proactive Auto-scaling and Energy-Efficient VM Allocation Framework Using an Online Multi-Resource Capsule Shuffle Attention Network for Cloud Data Centres: A Review

Abstract

Cloud computing has become a fundamental component of modern digital infrastructure, offering scalable and flexible access to computational resources through distributed cloud data centres. As the demand for digital services continues to grow, efficient management of large-scale workloads while ensuring service performance, energy efficiency, and sustainability has become a critical challenge. One of the primary issues in cloud environments is the dynamic allocation of virtual machines (VMs) to physical servers, where inefficient allocation can lead to underutilization of resources, increased energy consumption, and violations of service level agreements (SLAs). To address these challenges, proactive auto-scaling and energy-efficient VM allocation frameworks have gained significant attention. Unlike traditional reactive methods, proactive approaches leverage predictive models to anticipate workload variations and allocate resources in advance, thereby improving system responsiveness. Recent advancements integrate deep learning techniques with optimization algorithms to enhance prediction accuracy and allocation efficiency. Models such as convolutional neural networks, recurrent neural networks, and capsule networks effectively capture complex workload patterns, while attention mechanisms further refine predictions by focusing on critical resource features, enabling more intelligent and efficient cloud resource management.