An approach to optimize Kubernetes resource management through its adaptive scheduling system combined with threshold-based predictive scaling

Naik, Supriya Sunil (2025) An approach to optimize Kubernetes resource management through its adaptive scheduling system combined with threshold-based predictive scaling. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (586kB) | Preview
Preview	PDF (Configuration Manual) Download (870kB) | Preview

Abstract

Kubernetes is today the de facto standard for containerized application orchestration with scalability, flexibility, and automation. While its original scheduling and autoscaling mechanisms are generally reactive in nature, they are the root of Kubernete’s delayed response under workload spikes, resource waste, and SLA breach. This thesis proposes a blended approach of resource management combining predictive scaling using threshold values with adaptive scheduling in order to improve the responsiveness and efficiency of Kubernetes clusters. The predictive component uses historical data in Alibaba Cluster Trace Program to forecast CPU and memory usage through time-series forecasting models such as Holt-Winters. Based on these forecasts, the system forecasts node scaling ahead of time before the threshold crossings are hit. Meanwhile, the adaptive scheduler schedules pod deployment for optimization in terms of forthcoming node resource availability as well as past experience. The hybrid system was proofed and executed in a Minikube environment, workload simulations constructed with Locust, and performance collection obtained through Prometheus and Grafana. A comparative analysis with the baseline Kubernetes behavior indicated significant improvements, including a reduction in pod scheduling latency. The outcomes show that the integration of predictive and adaptive methodologies can make Kubernetes an active resource manager and improve the performance in dynamic, cloud-native application environments.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Makki, Ahmed UNSPECIFIED
Subjects:	T Technology > T Technology (General) > Information Technology > Cloud computing Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning
Divisions:	School of Computing > Master of Science in Cloud Computing
Depositing User:	Ciara O'Brien
Date Deposited:	30 Mar 2026 11:11
Last Modified:	30 Mar 2026 11:11
URI:	https://norma.ncirl.ie/id/eprint/9248

Actions (login required)

View Item