Methods for network monitoring to detect anomalies in time series

Abstract

The open source RRDtool software solves the challenge of collecting and storing time series data from service networks. However, even an experienced network technician will find it difficult to monitor all relevant service network time series at the same time. The approach is to incorporate a mathematical model into the monitoring application that can automatically detect deviations in the time series. And when we talk about the growing number of models, the choice is difficult, because the chosen model must be compatible with real-time monitoring, in this paper we adopt a new approach which is to integrate the theory of exponential smoothing and intelligent properties in Holt Winters algorithm by using RRDtool as a tool for data collection and representation With real-time schemes, although this technique is not perfect, it is adaptable, effective and successful as a tool for automatic identification of deviant behavior after being trained to identify anomalies through prediction based on past and future steps.