LOAD DISTRIBUTION METHOD IN GENERAL-PURPOSE MULTI-COMPUTER SYSTEMS ACCORDING TO NODE CLUSTERING BY RESOURCE UTILIZATION LEVEL

Abstract

The study examines methods for distributing computational load in general-purpose multicomputer systems, with a focus on clustering nodes based on their load levels. The research aims to improve the efficiency of resource utilization through dynamic load balancing across nodes, according to their current operational state. It involves an analysis of contemporary approaches to load balancing in multi-node systems, the development of a node clustering model based on load levels, and the creation of an algorithm that enables efficient load distribution among clusters to minimize delays and maximize resource utilization. The proposed approach is subject to experimental testing under simulation conditions, allowing for a comparative assessment of its effectiveness against baseline load distribution strategies.

The scientific novelty of the study lies in the implementation of a new load balancing method that involves clustering nodes according to their load levels, which not only ensures a more even distribution of tasks across the system but also contributes to reduced energy consumption and lower latency. The practical significance of the findings lies in the potential application of the proposed method for optimizing computational processes in environments such as cloud services, data centers, supercomputers, and corporate computing networks, where efficient resource management is of critical importance. To validate the effectiveness of the proposed method, the study utilizes performance metrics such as task completion time, system throughput, and energy efficiency, comparing them across various simulated scenarios.

To carry out the research, clustering methods that take into account the current load of computing nodes are employed, enabling adaptive load distribution based on the dynamic state of the system.