ASSA: Optimizing SLA Violation and VM Failure Rate for Task Scheduling in Cloud Environment

Abstract

Efficient task scheduling in cloud computing plays a crucial role in enhancing resource utilization, minimizing service disruptions, and meeting user expectations defined through Service Level Agreements (SLAs). This paper proposes a novel scheduling approach using the Artificial Satellite Search Algorithm (ASSA), inspired by the orbital dynamics and surveillance capabilities of artificial satellites. The algorithm effectively balances exploration and exploitation to optimize task-to-VM mapping in dynamic cloud environments. Key performance parameters such as SLA violation, Virtual Machine (VM) failure rate, and throughput are systematically addressed. By continuously observing the system state and adapting to resource failures or overloads, ASSA ensures minimal SLA violations and significantly reduces task delays caused by VM failures. Experimental results demonstrate that the proposed ASSA-based scheduler improves system throughput by 23.4%, lowers SLA violation rates by 19.7%, and decreases VM failure-induced task rescheduling by 21.1% compared to conventional metaheuristic algorithms. These results confirm that ASSA provides a reliable and scalable solution for task scheduling in modern cloud infrastructures, especially under fluctuating workloads and stringent QoS requirements.