METHODS OF INCREASING REACTIVITY IN INTERACTIVE INTERNET SYSTEMS

Abstract

This paper addresses the problem of improving the reactivity of interactive Internet systems in the context of service delivery. Reactivity is considered as a key quality metric defined by the average system response time and the variance of response delays under given usage conditions and resource constraints. A comprehensive analysis of influencing factors is presented, including network transmission delays, queuing times, and server-side processing times. It is proposed to model the real request traffic as a fractal discrete quasi-stationary stochastic process with predictable statistical parameters (mean and variance) combined with an unpredictable component representing burst traffic. This approach enables the system to be treated as a nonlinear dynamic system operating under sudden, non-predictable excitations, requiring optimal real-time parameter control to maintain stability within a specified burst range.

A packet Internet traffic model is developed, along with a method for its processing using telecommunication protocol implementations in server hardware. The paper further introduces a structural-functional model of the interactive system represented as a queuing system (mass service system) that separates incoming traffic into predictable and unpredictable components. For the predictable component, parameter optimization is carried out to meet target reactivity values, while the unpredictable bursts are mitigated through load balancing across processing servers.

Additionally, a request traffic shaping method is proposed, capable of converting a non-stationary traffic flow into a sequence of quasi-stationary segments with limited variance. The method employs a buffer-switching mechanism and an adaptive speed control system to smooth traffic fluctuations, reduce burst impact, and maintain service stability without excessive resource allocation. The proposed models and methods can be applied to the design and operation of interactive Internet systems that require high responsiveness and reliability under conditions of irregular and bursty traffic patterns. The results are relevant for improving Quality of Service (QoS) in a wide range of networked applications, including web-based platforms, cloud services, and real-time interactive systems.