HYBRID DYNAMIC LOAD BALANCING ALGORITHM FOR CLOUD PERFORMANCE IMPROVEMENT

Cloud computing is a modern robust approach, enabling individuals and businesses to buy the services they need per their demands over the internet. It offers a wide range of amenities such as easy access to online applications and services, storage, deployment platforms, and much more. Load balancing is a crucial component of cloud computing, and it prevents the overburdening of nodes while others are idle or underutilized. Maintaining the Quality of Service (QoS) parameters can be difficult for cloud providers when equal workload distribution across servers is a challenge. An effective Load Balancing (LB) approach should enhance and provide a high level of customer satisfaction by effectively utilizing Virtual Machines across servers. Even though load balancing algorithms (LBA) have been the subject of much research, efforts to decrease runtime, makespan, and boost throughput have not yielded satisfactory results. Through the hybridization of a dynamic load balancing algorithm and a machine learning algorithm, this research intends to decrease the runtime of load balancing activities, decrease makespan, and boost task throughput in a cloud computing environment. This study combines the Q-learning algorithm with the Honeybee Foraging Load Balancing Algorithm (HBF-LBA). The proposed Honeybee Foraging Q-Learning algorithm (HBFQL) was implemented in the CloudSim simulation environment. The suggested solution successfully decreased runtime by 13.1% and makespan by 8.95% while enhancing throughput by 8.37% during routing operations compared to the Shortest Job First (SJF) algorithm. Compared with the Ant Colony Optimization (ACO), the proposed algorithm reduced runtime by 14.57% and makespan by 13.71% while increasing throughput by 3.43%. This research improved task execution speed by continually monitoring the virtual machine usage history to route tasks to the best available virtual machine and ensure effective task distribution