Performance analysis and prediction of data networks by operational analysis

Abstract

We investigate applicability of operational analysis for performance evaluation and prediction of public data networks. For public data networks, mathematical queueing network models and simulation models have been used. However, they have several weaknesses for modeling and analyzing complex data networks, including restrictive, unverifiable assumptions, unavailable or not measurable parameters, oversimplifying network operations, and excessive computation time and memory. The conventional operational analysis approaches have been developed for computer systems. We apply the operational analysis for data networks with some modifications. From testable as-sumptions, we derive operational laws and performance quantities. We report experi-mental performance of the proposed method for a simulated public data network.