REDUCING THE DELAY OF E-LEARNING TRANSACTIONS IN COMPUTER NETWORKS OF HYPERCONVERGENT ARCHITECTURE

Abstract

The relevance of research. The infrastructure created on a convergent platform involves combining memory, computing and network resources into a single pool. But in the context of a hyperconvergent infrastructure, computing power, storage, servers, networks are integrated by software. This contributes to reducing operating costs, which is especially important for e-learning support systems. The subject of the research is processing e-learning transactions in computer networks of hyperconvergent architecture. The goal of the article is to reduce the delay of e-learning transactions in computer networks of hyperconvergent architecture. The following methods were used to reduce the delay of e-learning transactions – the methods of set theory; optimization by penalty functions, the method of potentials. The following results were achieved: the method for minimizing the average delay of transactions in computer networks of a hyperconvergent architecture was proposed, which enabled assigning the distributed computing resources for processing a set of e-learning transactions equally to the quanta of a given time interval. The method also suggests partitioning the set of transactions into subsets and distributing them to network nodes while distributed processing so that the average delay of the data packet in the network assumes the minimum value and ensures a uniform network loading for a large number of subscribers. In the proposed method, the objective function of the task of finding a rational partitioning of a set of e-learning transactions processed in the computer network into subsets and their assignment to the nodes of the hyperconvergent network is determined by using the penalty function when each unit of the computing resource is assignedto the current time quantum. Conclusions. The approach for reducing the delay of e-learning transactions in computer networks of a hyperconvergent architecture is proposed. The approach is based on the proposed method for minimizing the average delay, taking into account the features of a hyperconvergent architecture. The application of the approach enables balancing the network load with a large number of transactions and meeting the requirements for the speed of e-learning transaction processing.