Research of load balancing in three-channel cluster systems with adaptive dispatching

Modern organizations face the need for effective load management in distributed computing systems to ensure high performance and fault tolerance. The article discusses an approach to load balancing in three-channel cluster systems using various adaptive dispatch algorithms. Cluster systems allow optimizing the distribution of tasks between nodes, which is important for systems with heterogeneous architecture. The purpose of the study: is to substantiate the parameters and develop simulation models for analyzing the effectiveness of various dispatch algorithms in three-channel cluster systems. Models with joining the shortest queue, switching between queues, and taking into account node performance are considered. Practical significance: due to the focus on organizations using cluster systems for processing large amounts of data, where high performance and minimal response time are required. The results of the study make it possible to optimize load distribution and increase the resilience of systems to failures. Results: the results of simulation modeling of three-channel systems with various dispatch algorithms are presented, including models with connection to the shortest queue (LCO) and to the queue with the shortest delay (LCC), with transitions between queues (PMO) and with heterogeneous nodes. The characteristics of the response time, node load, and queue length are given depending on the coefficient of server performance heterogeneity. Discussion: The key advantages of adaptive dispatch algorithms are highlighted, such as improved load balancing and reduced response time. The challenges associated with the implementation of complex algorithms are noted, including the need to take into account node performance and data transmission delays. Promising areas for further research are indicated, including optimizing node placement, accounting for stochastic uncertainty, and integration with corporate management systems.