PERTINENT ASSORTMENT OF VIRTUAL MACHINES USING DEEP LEARNING COLLATERAL CHUNK IN CLOUD ENVIRONMENT

The advancement of cloud computing facilitates the sharing of computing resources effectively. However, the higher energy consumption, dynamic workload variation, and ineffective Virtual Machine (VM) assortment degrade Quality of Service (QoS) and Service Level Agreement (SLA) violations, leading to improper resource allocation. Therefore, a novel VM assortment and dynamic allocation of computational and communication resources based on Deep Learning (DL) collateral chunk is proposed. Primarily, the task and the cloud environment modeling are initialized. Then, the task arrival is carried out, followed by the task allocation strategy. Next, the load of each VM is characterized. Further, the VMs are placed dynamically based on the Lamarckian Cauchy-based Improved Moth Search (LC-IMS) method. After that, the resource is allocated in the selected VM using Deep Deterministic Policy Gradient Neural Network (DDPGNN). Thus, the proposed DDPGNN executed the task on the selected VM with a 21% task rejection rate, 99.6% resource utilization, 31% CPU utilization, 90kWh energy usage, and 8.5 seconds response time, showing better performance than the existing Convolutional Neural Network optimized Modified Butterfly Optimization (CNN-MBO) and Effective Resource Allocation Strategy (ERAS). Thus, the proposed system is adaptable to real-time cloud environments that enable rapid response to fluctuating workloads and user demands while maintaining high efficiency and low latency.