Anomaly detection is a technique from the field of machine learning that makes it possible to detect deviations (anomalies) or irregularities in large amounts of data. The idea behind anomaly detection is to develop a model that defines a “normal” data range and then identifies all data points that lie outside this normal range. Different methods can be used, such as statistical methods, clustering or neural networks. Anomaly detection is often used to detect patterns in large data sets that may indicate certain events or processes. Examples include monitoring computer systems for possible threats or monitoring financial transactions for signs of fraud. Our advanced analytics solution CIO COCKPIT uses aspects of anomaly detection to identify and avert risks at an early stage. Thus, a deviation on the cost axis (e.g. due to storage growth) can be quickly followed up.