Adaptive Multi-Cloud Container Orchestration for Optimal Workload Portability and Resource Utilization using ML-Driven Predictive Scaling

Tadisetti, Shanmukha Sai Teja (2025) Adaptive Multi-Cloud Container Orchestration for Optimal Workload Portability and Resource Utilization using ML-Driven Predictive Scaling. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (2MB) | Preview
Preview	PDF (Configuration Manual) Download (588kB) | Preview

Abstract

Businesses with multi-cloud infrastructure are overprovisioned by 35-45 % of resources leading to the organization wasting thousands of dollar a month in cloud operation expenditure due to global cloud expenditure surpassing 270 billion every year. This study offers an adaptive orchestrator framework that uses Long Short-Term Memory ( LSTM ) predictive scaling in conjunction with intelligent hybrid AWS Lambda-Fargate workload classification to resolve the inefficiency related to reactive scaling. The framework utilizes multi-horizon LSTM models where a level of 42.24% Mean Absolute Percentage Error (MAPE) can be attained over 5, 10, and 15-minute forecast windows, which allows its implementation with proactive resource allocation even when the error levels are high. The accuracy of classification of the proposed hybrid orchestration engine is satisfactory, and it could redirect the burst workload to Lambda and sustained workload to Fargate accurately with under 100ms decision latency. Experimental validation against baseline demonstrates 28 percent savings in organizational costs of maintaining a traditional Kubernetes Horizontal Pod Autoscaler (HPA), 70 percent faster scaling responses (45 seconds versus 150 seconds baseline) and 80 percent Fargate-deployment resource capacity utilization. Its AWS-native conception allows brokering deployable directly, to resolve nagging multi-cloud orchestration problems via predictive intelligence and workload-measured resource distribution.

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:	31 Mar 2026 10:36
Last Modified:	31 Mar 2026 10:36
URI:	https://norma.ncirl.ie/id/eprint/9272

Actions (login required)

View Item