The software monitoring service of SIG involves source code analysis on a regular basis of millions of lines of code. The stream of metrics data that is generated by this analysis is used by the SIG consultants to discover trends and anomalies in the software production process.

To support the monitoring consultants in this task, we have developed detection algorithms that discover significant change patterns as they occur. Upon detection, an alert is sent to the responsible consultant who follows up with root-cause analysis, interpretation, and possibly reporting towards the customer.

Our event detection mechanisms are based on principles taken from the field of statistical process control, but adapted to the particularities of source code metric streams. The event detection and ensuing alerts allow our consultants to react swiftly and accurately, even in the face of large and detailed streams of measurement data.

Related: