AI-Driven Auto-Tuning for Serverless Performance Optimization

Aung, Min Ko (2025) AI-Driven Auto-Tuning for Serverless Performance Optimization. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (1MB) | Preview
Preview	PDF (Configuration Manual) Download (1MB) | Preview

Abstract

Serverless computing is one of the factors involved in cloud application development to help them expand dynamically and be charged under pay-per-use billing without the need for infrastructure management. Nevertheless, serverless functions continue to be challenging to optimize, especially on cloud systems like AWS Lambda, which are faced with both unpredictable loads and black-box execution patterns. This research proposes the framework for an AI-driven auto-tuning that integrates the hybrid system using machine learning (ML) and reinforcement learning (RL) to facilitate both serverless performance and cost optimization. ML is utilized for predicting the short-term invocation patterns, which are then used by RL agent for dynamically adjusting the configuration parameters, such as memory, timeout, and concurrency. The closed loop pattern allows system learning in response to the execution feedback of latency, improving the decisions, as time goes by, in terms of quality and cost. Experimental validation with synthetic workloads, including bursty, periodic, and noisy traffic, demonstrates how the framework outperforms static and single-strategy baselines. Significant improvements in response time, cost reduction, and service level agreement (SLA) compliance have resulted in and provided the simulated serverless environment.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Gupta, Shaguna UNSPECIFIED
Uncontrolled Keywords:	Serverless Computing; Auto-tuning; Machine Learning; Reinforcement Learning; AWS Lambda; Cloud Optimization
Subjects:	Q Science > QH Natural history > QH301 Biology > Methods of research. Technique. Experimental biology > Data processing. Bioinformatics > Artificial intelligence Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Artificial intelligence 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:	20 Mar 2026 11:12
Last Modified:	20 Mar 2026 11:12
URI:	https://norma.ncirl.ie/id/eprint/9198

Actions (login required)

View Item