An Efficient Resource Allocation Strategy Using Bandwidth-Aware Virtual Machine Placement in Cloud Datacenter Network Architectures

In recent years, cloud computing has revolutionized the delivery of computing, bandwidth, and storage by offering hardware, platforms, and applications as a service in a scalable environment. Efficient resource allocation, particularly virtual machine (VM) placement within data center networks (DCNs), remains a challenge to ensure service-level agreement (SLA) compliance and optimized energy consumption with minimum resource migrations. This paper proposes a novel self-adaptive resource allocation approach, termed SARA-BC, which integrates bandwidth and computing parameters in a VM placement decision process. Unlike traditional approaches that assume infinite resources, SARA-BC models practical constraints such as dirty memory rate, task completion time, and host energy consumption, and resource migrations. The model is evaluated using six DCN topologies (MDCube, HCN, BCN) with both base and extended parameters, leveraging real-world workload traces from Google Cluster. The SARA-BC technique is also evaluated and analyzed against static, dynamic, human-inspired, and bio-inspired techniques. Experimental results demonstrate that SARA-BC significantly improves average execution time, reduces the number of VM migrations, and lowers energy usage across different configurations, as shown in Sect. 4.5. The SARA-BC technique outperforms the Min-Min, GA, ANN, and SARA. This work offers a robust and scalable solution for energy-efficient VM management in cloud infrastructures. Hence, the bandwidth, computing-aware resource allocation provides better results in cloud datacenter network systems using BCN, HCN, and MDCube DCN topologies.