A neural network approach to system performance analysis

Neural networks are used in a wide variety of situations to solve complex problems. Some of the categories for which neural networks are used include: prediction software, classification algorithms, data association environments, data conceptualization environments, and data filtering problems. This work described in this paper implements a neural network that spans both the prediction and data association problems. The neural network approach to system performance analysis takes performance data from computer systems and uses a Kohonen based neural network to analyze the performance data and attempts to find bottlenecks in the computer system. The data performance analysis results are present as line graphs that can be interpreted by computer experts to determine bottlenecks within the computer system, and can intelligently suggest upgrades to improve any subsystem that suffers from poor performance. The aim of this work is to provide a "proof of concept" for use in IT assessments, but can also be applied to any situation involving computer performance analysis.