Study of Anomaly Identification Techniques in Large scale systems

Anomaly detection, also called as outlierdetection refers to identifying patterns in a given data set that deviates from usual manner. The patterns identified are called anomalies or intrusions. There are several existing solutions for Anomaly detection. However, it is difficult to successfully find anomalies from huge amount strident, high-dimensional data, because of the limitations in existing anomaly detection techniques. Since the existing systems are not scalable, they are not suitable for large scale systems. In this paper, we provide a study on traditional anomaly detection techniques and the limitations imposed by those techniques, and hence we illustrate the need of suitable anomaly detection technique for large scale systems. Finally, we discuss an automated anomaly identification methodology [37] for large scale systems.