NORMA eResearch @NCI Library

Configuring serverless functions using Q learning and Deep Q learning algorithms

Thomas, Johns (2024) Configuring serverless functions using Q learning and Deep Q learning algorithms. Masters thesis, Dublin, National College of Ireland.

[thumbnail of Master of Science]
Preview
PDF (Master of Science)
Download (1MB) | Preview
[thumbnail of Configuration Manual]
Preview
PDF (Configuration Manual)
Download (433kB) | Preview

Abstract

Serverless functions attains great attention in recent years particularly because of its advantages like low administrative overhead, automatic scaling and fined grained control over billing. Developers can easily deploy service to cloud environments using serverless functions within seconds and are provided with few configuration options. Configuration like memory and timeout in commercial platforms like AWS Lambda directly affects performance and cost. Therefore, it is crucial to configure serverless functions with optimal parameters. Serverless function configuration becomes easy if the underlying relationship between configurations and cost is known. This research is an attempt to study the relationship between performance and configuration of a serverless function using reinforcement learning techniques such as Q learning and Deep Q learning. The Q learning and Deep Q learning agents have been developed and trained to learn the optimal configuration on serverless functions. This includes defining state and action space, reward function, and collecting the execution details of function execution after each invocation. Image processing functions deployed on AWS Lambda platform is used as the environment for the agents to interact with. Both Q learning and Deep Q learning agents have provided positive results in learning the relationship between performance and configuration of serverless function. The analysis of results show that Deep Q learning has edge in learning the relationship compared with Q learning. However, both agents can improve the performance by increasing state space by considering parameters like concurrency.

Item Type: Thesis (Masters)
Supervisors:
Name
Email
Emani, Sai
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: 04 Jul 2025 11:36
Last Modified: 04 Jul 2025 11:36
URI: https://norma.ncirl.ie/id/eprint/8061

Actions (login required)

View Item View Item