A Queuing Theory Approach to Task Scheduling in Cloud Computing with Generalized Processor Sharing Queue Model and Heavy Traffic Approximation

Cloud computing has transformed data storage, management, and processing by offering scalable and flexible resources via the internet. A key component of this technology is the efficient allocation and management of resources, particularly through task scheduling at the level of virtual machines (VMs). Task scheduling is critical for maximizing resource utilization and system performance in cloud environments. However, it presents significant challenges due to the dynamic and distributed nature of these environments. Effective task scheduling algorithms are necessary to balance load, minimize response time, and optimize resource usage, making it a crucial area for ongoing research and development in cloud computing. This paper addresses the challenge of task scheduling in cloud computing by employing an analytical approach based on queuing theory. We model the system using a generalized processor sharing (GPS) queue and evaluate its performance through heavy traffic approximation. This method allows us to derive performance metrics for queuing systems prone to congestion, considering general interarrival and service time distributions, thus providing a comprehensive analysis of scheduling efficiency. © (2024), (International Association of Engineers). All rights reserved.