METHOD OF AVOIDING INTERLOCKS OF TASKS IN DISTRIBUTED SYSTEMS BASED ON INTERPROCESS INTERACTION PROTOCOL

Abstract

The article presents the results of the development and research of the method of avoiding mutual blocking of tasks in distributed computing systems, which is based on the integration of an event-oriented model of execution with a formalized protocol of interprocess interaction. The proposed approach forms a holistic methodological basis for managing distributed computing, in which the reactor is considered as a central coordination mechanism for coordinating structural dependencies between tasks and resource access procedures. The key result is ensuring the structural absence of waiting cycles due to the combination of algorithms for detecting potential conflicts with an adaptive coordination protocol.

In the work, a complex system of metrics was formed, which allows to quantitatively evaluate the state of nodes, flow parameters and the level of resource load, which, in turn, creates a basis for a reasonable choice of planning, balancing and scaling strategies. The experimental results, presented in graphic and tabular forms, confirm the preservation of the linear dynamics of performance with an increase in the number of nodes and the predicted increase in overhead costs without loss of system stability. Integrating a formal dependency model with a messaging protocol has been proven to guarantee the advancement of critical execution paths even in heterogeneous environments.

The obtained results indicate the practical suitability of the proposed method for building scalable and reliable distributed systems that combine the requirements of high efficiency of resource use and resistance to mutual blocking. Prospects for further research are related to the adaptation of the approach to heterogeneous node architectures and advanced configurations with a large number of components.