Optimizing Memory Allocation Using Intelligent Deep Data Mining on an IaaS Cloud Platform

Cloud computing enables users to access virtualized services as needed, offering an on-demand service model that has driven significant adoption in the field of information technology infrastructure. In a cloud environment, applications run on dedicated virtual machines (VMs), with each VM hosted on an independent server. Given the ondemand nature of cloud services, effectively managing memory utilization within the cloud network is critical, as these platforms must support the concurrent management of multiple applications. Current resource allocation methods in cloud computing primarily rely on probability distribution approaches, which do not ensure optimal RAM utilization across the network. This study focuses on developing a more efficient method for dynamic memory allocation within VMs operating in cloud environments. The proposed approach introduces a Memory Over-subscription Granter (MOG), a Cloud Vertical Elasticity Manager (CVEM), and a memory reporter (MR) to manage memory allocation among VMs on cloud platforms. MOG includes an intelligent scheduler that allocates host RAM by analyzing system data, optimizing memory distribution. Additionally, the study employs an elasticity rule to facilitate dynamic memory allocation across available cloud hosts, ensuring balanced memory utilization. This method supports improved memory efficiency and enhances the capacity to manage varying application loads within cloud-hosted VMs.