Моделювання серверних систем як складних мереж масового обслуговування в умовах високого навантаження

Abstract

This work investigates the behavior of server systems characterized by nonlinear effects that manifest as disproportionate performance degradation under increasing workloads. It is shown that as resource utilization approaches critical thresholds, response times and queue lengths increase sharply. These effects are driven by internal resource constraints. This study formalizes a server system as a network of service nodes where requests transition between stages, forming queues. Ultimately, the server system is represented as a dynamic object whose state evolves over time under the influence of input request flows and internal processing mechanisms.