Predictive VM placement algorithm for resource optimization: leveraging deep learning forecasting and resource relationship modeling
Rajni Garg, Indu Arora, Anu Gupta · International Journal of Computers and Applications · 2025
The growing demand for cloud computing has made it imperative to optimize the utilization of cloud resources. Resource optimization can be improved through Virtual Machine (VM) Placement. In order to effectively optimize placement of VMs, it becomes necessary to anticipate future resource demand. However, accurate resource forecasting is a major challenge due to the dynamic nature of cloud applications. Furthermore, if VMs are placed on the same server, it can lead to resource contention, especially when they compete for the same resources. This contention can adversely affect VMs performance and potentially increase cost for users, as well as energy consumption by the infrastructure. This work proposes a model named Predictive Disparity-based Virtual Machine Placement (PDVMP) which aims to enhance VM placement decision. The model integrates forecasting techniques grounded in Deep Learning for estimating the future resource needs of VMs. This estimation is incorporated in VM placement decision in order to ensure long-term sustainability of the VM on the destination server. Moreover, the model used in the current research work balances resource optimization and execution performance by packing multiple VMs that exhibit complementary resource demand on a same physical server. The performance of PDVMP model is tested against benchmark placement policies using real workload traces of bitBrains datacenter. The results show that the proposed approach improves resource utilization while reducing both performance bottlenecks and energy consumption. The experimentation shows an improvement in Energy Performance Metric ranging from 49.3% to 62.97%.