AI-Driven Load Balancing in Cloud Computing with LSTM Forecasting and DDPG on Kubernetes

Ali, Zeeshan (2025) AI-Driven Load Balancing in Cloud Computing with LSTM Forecasting and DDPG on Kubernetes. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (3MB) | Preview
Preview	PDF (Configuration Manual) Download (851kB) | Preview

Abstract

Load balancing plays a vital role in cloud computing, especially since it has a direct influence on the throughput of the system, the use of resources, and the use of energy. Round-robin and threshold-based autoscaling are traditional controls that are reactive; that is, they are not predictive of the best way to allocate resources to sudden workload spikes, with a result of under-utilization or over-utilization. This paper presents a new AI-based intelligent load balancer that uses the Long Short-Term Memory (LSTM) networks to predict the workload and Deep Deterministic Policy Gradient (DDPG) reinforcement learning to schedule the resources dynamically in the Kubernetes clusters. The system does latency-driven optimization of the low latency, high throughput and energy-saving by making use of predictive resource requests with the help of the Google Cluster traces. Extensive experiments show that the hybrid LSTM+DDPG system has a high compliance of 98.2% SLA compliance in contrast to 89.5% of Kubernetes HPA, low average response time by 31% (145ms vs 210ms), lower energy consumption by 22.8 percent (0.112 vs 0.145 CPU-hours per 1000 requests), and less scaling oscillations by 51.9 percent. All the improvements are statistically validated (p < 0.001). The hybrid way is better than the currently achieved state-of-the-practice strategies and shows improved availability and sustainability of cloud-based services.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Gupta, Shaguna UNSPECIFIED
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
Divisions:	School of Computing > Master of Science in Cloud Computing
Depositing User:	Ciara O'Brien
Date Deposited:	31 Aug 2026 11:58
Last Modified:	31 Aug 2026 11:58
URI:	https://norma.ncirl.ie/id/eprint/9688

Actions (login required)

View Item