Dynamic Resource Scaling Using Predictive Analytics in Cloud

Kale, Pranal Suresh (2025) Dynamic Resource Scaling Using Predictive Analytics in Cloud. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (1MB) | Preview
Preview	PDF (Configuration Manual) Download (791kB) | Preview

Abstract

The way companies build IT infrastructure was changed due to Cloud Computing. With Cloud Computing, costs shifted from CapEx to OpEx. The biggest issue with Cloud Computing is Elasticity, the ability to adjust resources in real-time based on the needs of the business. Traditional auto-scaling systems rely on historical performance metrics and can be affected by these types of changes, causing performance drops during peak use and wasted resource costs during off-peak periods. In this report, proposed a new Dynamic Resource Scaling system that employs Predictive Analytics to proactively determine how much resource needs to be allocated at any given time. To create the Dynamic Resource Scaling system, we built machine learning models using a dataset of 3562 cloud jobs. Created three predictive analytics models based on Linear Regression, Random Forest Regressor, and Long Short Term Memory (LSTM) to predict both CPU and memory allocation. The Random Forest Regressor produced the best CPU prediction accuracy (R²=0.9968, MAE=0.1060), while Linear Regression produced the best Memory prediction accuracy (R²=0.9906). Once developed, the machine-learning models were loaded into a Python-based orchestration tool that controlled resources on the AWS platform. In a 30-minute EC2 simulation, the Predictive Analytics system successfully performed predictive scaling to allocate appropriate amounts of resource during a load spike greater than 173% of the baseline capacity.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Samarawickrama, Yasantha UNSPECIFIED
Subjects:	T Technology > T Technology (General) > Information Technology > Cloud computing T Technology > T Technology (General) > Information Technology 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 Aug 2026 15:09
Last Modified:	31 Aug 2026 15:09
URI:	https://norma.ncirl.ie/id/eprint/9709

Actions (login required)

View Item