Model of adaptive vector control of stochastic hybrid caching algorithm

The problem of increasing the cache system efficiency using the adaptive vector control of the stochastic hybrid caching algorithm is investigated. The mathematical model of the abstract one-level caching system permitting to describe the known replacement strategies, including those which use object sizes for calculating their cache rating, is considered. This model is used in the mathematical cache-system model with the controlled stochastic hybridization of several basic caching algorithms. The adaptive control scheme of the stochastic hybrid caching algorithm permitting to shorten ten times the adaptation time compared to the previously considered methods at the expense of abandoning the direct solution to the performance criterion maximization problem at the preceding route section is presented. The known replacement strategies with the description of their operation fundamentals are surveyed. The chosen investigative techniques and the performed methods structure are characterized. The comparison results of the obtained replacement strategies productivity on the non-stationary routes composed of the quasistationary sections obtained on the base of 20/80 Zipf’s distribution law are presented.