ENHANCED ONDEMAND PREEMPTIVE SELF-SCHEDULING BASED ON MAX-QUEUING ENERGY-AWARE REQUEST PROTOCOL FOR LOAD BALANCING IN A DECENTRALIZED CLOUD ENVIRONMENT

Abstract

In recent years, the adoption of cloud computing has surged, with organizations and individuals increasingly leveraging cloud solutions for data storage and processing. This growing reliance on cloud services necessitates effective load balancing in decentralized cloud environments to optimize resource utilization, enhance system performance, and improve user experiences. However, current load-balancing techniques often fail to address scheduling performance issues, primarily due to task-shifting problems stemming from improper priority assignments. These deficiencies can lead to deadlocks, task collisions, and failures, increasing time complexity and hindering overall system efficiency. To address these issues, we propose a novel On-Demand Pre-Emptive Self-Scheduling (ODPSS) approach incorporating a Max-Queuing Energy-Aware Request Protocol (MQEARP) tailored for decentralized cloud environments. Our methodology begins with collecting virtual incoming request tasks from the web request handler within the cloud server, creating cloud request logs. We utilize a Task Time Intensity Rate (TTIR) to estimate task completion times, which are subsequently integrated into each task by establishing a max queue priority system. We employ Particle Swarm Optimization (PSO) to refine our approach further and evaluate critical server handling parameters, including energy consumption, processing time, queuing length, demand, and request delay dependencies. This evaluation yields an absolute mean rate based on the maximum count of requests the server handles, enabling us to prioritize and queue tasks effectively within the Max-Queuing Energy-Aware Request Protocol (MQEARP) framework. The OnDemand service is activated as the web server receives dynamic requests, allowing tasks to await processing in the queue. To manage this influx efficiently, our Enhanced Preemptive Self-Scheduling strategy facilitates the transition of OnDemand requests into a prioritized queue, enabling the deployment of temporary Virtual Machines (VMs) to alleviate the cloud server's workload. After the extended OnDemand service is completed, these virtual servers are terminated, effectively reducing the operational burden on the cloud infrastructure. The objectives of the proposed technique are to enhance productivity, optimize server resource allocation by considering the varying importance of user tasks, and mitigate the risk of server failures. By addressing the identified challenges and filling existing research gaps, our proposed method aims to contribute significantly to the field of cloud computing, paving the way for more efficient and reliable load-balancing solutions in decentralized environments. Based on the performance analysis of the proposed method, such as delay tolerance, energy consumption, execution cost efficiency, self-scheduling, time complexity, and resource allocation number, cloud request logs are generated, and evaluation of the ODPSS method improves to 96.9%.