Dynamic Resource Allocation in Multi-Cloud Environments Using Reinforcement Learning

Varghese, Fivin (2023) Dynamic Resource Allocation in Multi-Cloud Environments Using Reinforcement Learning. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (2MB) | Preview
Preview	PDF (Configuration Manual) Download (91kB) | Preview

Abstract

This paper explores using reinforcement learning (RL) techniques for dynamic resource allocation in multi-cloud environments. The goal is to optimize performance and costs by automatically scaling cloud resources based on workload demands. Two popular RL algorithms are implemented: proximal policy optimization (PPO) and deep Q-networks (DQN). A simulation environment is created modeling key characteristics of auto-scaling Amazon EC2 instances across metrics, delays, pricing, and demand patterns. The trained RL policies are evaluated on metrics capturing the tradeoff between resource utilization, service quality, and operational expenditure. Results demonstrate both PPO and DQN successfully learn non-trivial auto-scaling strategies exceeding basic thresholds, confirming RL’s viability for cloud optimization. Further analysis illuminates an intriguing reliability efficiency spectrum contrasting their scaling behaviors. While DQN risks higher volatility in exchange for potential efficiency peaks, PPO favors gradual improvements ensuring consistent stability. The findings establish simulations as instrumental for low risk, reproducible cloud RL research while guiding real world algorithm selection tradeoffs between peak versus sustainable optimization. Ongoing directions like integrating forecasting and deploying models over live traffic would further strengthen production readiness.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Arun, Shreyas Setlur UNSPECIFIED
Subjects:	Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Cloud computing Q Science > QA Mathematics > Algebra > Algorithms > Computer algorithms
Divisions:	School of Computing > Master of Science in Cloud Computing
Depositing User:	Ciara O'Brien
Date Deposited:	11 Apr 2025 12:42
Last Modified:	11 Apr 2025 12:42
URI:	https://norma.ncirl.ie/id/eprint/7421

Actions (login required)

View Item