Voipfuture developed a new intelligent alarming technology for Qrystal Intelligence, which only raises alarms that are significant, pointing to unusual events in a VoIP service. Conventional alarming compares current data to static thresholds. This approach is good for use cases such as SLA monitoring or policy control. Here, hard thresholds need to be met, because they are part of contractual agreements or enshrined in management goals.
Such hard thresholds do not help to detect anomalies. Voice service quality often depends on the network load, the route, destination and the hour of day and day of week. Setting static thresholds in such dynamic environments is almost impossible – either thresholds are too tight, raising too many alarms, or thresholds are too loose, making them effectively useless. In practice, VoIP service operation teams are often flooded with alarms preventing them from identifying the issues that truly matter.
Intelligent Alarming is designed to overcome the limitations of conventional alarming. Intelligent Alarming uses machine learning technology to predict future KPI values based on historical data. This adaptive thresholding detects events that are outside the norm and help operations teams focus on real incidents.