Author(s): Ramasankar Molleti
This paper discusses some sophisticated autoscaling strategies in Kubernetes, and they include horizontal and vertical pod autoscaling, cluster levels autoscaling, and metrics based autoscaling. In this regard, it covers the topics of both predictive and event-triggered auto scaling approaches and their applications, the advantages and the potential issues relating to them, and more. Some of the questions answered by the study include how to improve application throughput and resource efficiency and how to achieve cost-efficient K8s clusters for practitioners while providing directions for the research on CA autoscaling in container orchestration.
View PDF