KUBERNETES CLUSTER SCALING METHODS IN CLOUD ENVIRONMENTS

Abstract

The article explores Kubernetes cluster scaling in cloud environments, a critical factor in ensuring high availability for microservice applications. It analyzes traditional techniques and modern approaches, including vertical and horizontal scaling mechanisms. Methods such as predictive autoscaling, event-driven scaling, and the use of custom metrics are also considered. The advantages and disadvantages of these approaches are discussed. Furthermore, the paper demonstrates how the configuration and optimization of scaling parameters impact cloud resource costs and application performance. Purpose of the article is to analyse modern tools and approaches to scaling Kubernetes clusters in the AWS cloud environment, to identify the advantages and disadvantages of each of the available solutions. Conclusions. Event-based scaling is especially useful for loads with irregular or impulsive nature. Promising directions include the development of dynamic mechanisms for changing container resources at runtime, the simultaneous use of horizontal and vertical scaling, the use of hybrid approaches for microservices, and the expansion of research at the infrastructural level. The choice of scaling strategy should be based on the characteristics of workloads, budgetary constraints, and the organisation's overall operational goals. Fine-tuning scaling configurations and continuously monitoring application behaviour are critical to achieving efficient resource utilisation, increased productivity and reduced costs in a cloud-native environment.