Sillo, Oritsejolomi Christian (2024) Investigating Multi-Agent Reinforcement Learning for adaptive cost-optimized storage allocations within cloud environments. Masters thesis, Dublin, National College of Ireland.
Preview |
PDF (Master of Science)
Download (1MB) | Preview |
Preview |
PDF (Configuration Manual)
Download (252kB) | Preview |
Abstract
This research seeks to explore the applicability of Multi Agent Reinforcement learning(MARL) in developing effective strategies to reduce data cost in cloud storage environments. Existing solutions have demonstrated the effectiveness of single agent reinforcement learning(RL) algorithms in reducing storage cost over traditional methods, however, there is limited research in how effective MARL algorithms will perform in this area, especially as it pertains to the scalability and fault tolerance of the system. In this study, we design and build a custom Gymnasium and cloudsim cloud storage environment in order to simulate a real world cloud storage solution, where MARL agents can be trained using Proximal Policy Optimization (PPO) algorithm. The study was able to demonstrates MARLs ability to significantly reduce cloud storage cost compared to baseline strategies while also being more fault tolerant than existing single RL solutions. Our study also highlights the potential for MARL to address the challenge of scalability in comparable systems.
Item Type: | Thesis (Masters) |
---|---|
Supervisors: | Name Email Kazmi, Aqeel 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 |
Divisions: | School of Computing > Master of Science in Cloud Computing |
Depositing User: | Ciara O'Brien |
Date Deposited: | 04 Jul 2025 11:09 |
Last Modified: | 04 Jul 2025 11:09 |
URI: | https://norma.ncirl.ie/id/eprint/8057 |
Actions (login required)
![]() |
View Item |