A Survey of Methods and Architectures for A Proactive Auto-scaling and Energy-Efficient VM Allocation Framework Using an Online Multi-Resource Capsule Shuffle Attention Network for Cloud Data Centres

Abstract

Cloud computing has become a fundamental component of modern information technology infrastructure, enabling scalable computing resources, distributed data storage, and flexible service deployment across diverse application domains such as enterprise systems, big data analytics, and artificial intelligence platforms. Cloud data centres consist of large-scale server farms and virtualization environments that support dynamic and heterogeneous workloads, requiring efficient resource management strategies to maintain optimal performance and service reliability. However, the rapid expansion of cloud-based services introduces significant challenges in workload distribution, virtual machine (VM) allocation, energy efficiency, and operational cost reduction. One of the most critical issues is the optimal placement of VMs onto physical servers while ensuring balanced resource utilization and minimizing power consumption. Since cloud data centres consume substantial electrical energy for computation, cooling, storage, and networking, inefficient scheduling leads to higher energy usage, increased carbon emissions, and elevated operational expenses. Therefore, energy-aware resource management has become a key research focus in sustainable cloud computing. Additionally, auto-scaling mechanisms play an important role in dynamically adjusting computing resources based on workload variations. While reactive scaling strategies respond after performance degradation occurs, they often cause latency and SLA violations. In contrast, proactive approaches leverage predictive analytics to estimate future demand and allocate resources in advance, thereby improving system efficiency, reducing response delays, and enhancing overall cloud performance and reliability.