Cloud Engineering as Search-Based Software Engineering: An Optimisation Framework for Cloud Workload Scheduling
Main Article Content
Abstract
Cloud computing has risen to become an essential part of today's software systems, providing the infrastructure for resource-intensive applications with scalability and flexibility. Nevertheless, resource distribution and workload scheduling in dynamic cloud environments are important issues to optimise. The paper discusses how Search-Based Software Engineering (SBSE) methodologies, namely Genetic Algorithm (GA), Particle Swarm Optimisation (PSO) and Simulated Annealing (SA), can be used to optimise cloud workload scheduling and resource allocation. With the Cloud Workload Dataset, the paper shows that these SBSE algorithms are applicable to improve performance metrics such as latency, CPU utilisation, cost, and throughput in various cloud workloads, such as data processing, machine learning, signal processing, and web server tasks. The findings suggest that GA performs best with low-latency applications, whereas PSO is the best in resource usage and throughput, which is the best in data-intensive tasks. Though computationally intensive, SA is useful in difficult problems in optimisation when an exhaustive search is needed. The paper presents the benefits of SBSE methods in enhancing cloud resources and scaling of systems, which can be valuable to cloud engineers and researchers. Future work discusses the hybrid methods based on SBSE and machine learning models and examines the multi-objective optimisation in order to achieve the optimal balance between latency, cost, and throughput in cloud-based environments.