Dynamic load balancing in AI-enabled cloud infrastructures using reinforcement learning and algorithmic optimization
Independent Researcher, University of the Cumberlands, USA.
Research Article
World Journal of Advanced Research and Reviews, 2023, 20(01), 1327-1335
Article DOI: 10.30574/wjarr.2023.20.1.2045
Publication history:
Received on 19 August 2023; revised on 21 October 2023; accepted on 24 October 2023
Abstract:
Dynamic load balancing is a key challenge in AI-enabled cloud infrastructures with volatile resource demand. This results in resource utilization drifting away from balance and creating performance loss, so the infrastructure starts to operate inefficiently. In this paper, we introduce a principled approach based on reinforcement learning and algorithmic optimization to dynamically allocate the load across the infrastructure. Our approach is based on reinforcement learning, providing instructions on what the ideal actions for load balancing in an ever-changing environment are. It takes advantage of a deep neural network to capture the complex interactions from historical states and associated load-balancing actions. The best actions are selected by maximizing the sum of rewards, taking into account short-term and long-term objectives. To increase the efficiency of the load balancing even further, we then apply algorithmic optimization approaches like genetic algorithms and ant colony optimization. Smart load-balancing strategies: These are done using an introduction of deep Q-learning algorithms, which helps in the optimization of the decision-making process of such reinforcement learning agent targeting for highly intelligent and efficient load-balancing act aggregate. Experimental results based on simulations and real-world experiments show that our framework can help network programs highly efficiently balance workloads and significantly improve the performance of the infrastructure. It can adjust to changing resource demands and conditions as well, so it should prove effective against such a dynamic environment. Overall, we present a new paradigm for implementing dynamic load balancing for AI cloud infrastructures. By combining the best of reinforcement learning and algorithmic optimization, it can improve resource utilization, delivering high-performance servers.
Keywords:
Load Balancing; Dynamic Environment; Reinforcement Learning; Flexibility; Efficient Resource
Full text article in PDF:
Copyright information:
Copyright © 2023 Author(s) retain the copyright of this article. This article is published under the terms of the Creative Commons Attribution Liscense 4.0