Watch Everything, Watch Anything - Anomaly Detection
Five years ago Ian Malpass posted his “Measure Anything, Measure Everything” article that introduced StatsD to the world. Since then DevOps has grown and defined itself around the ideal to measure everything.
Now it’s time to take it further. “Watch Everything, Watch Anything”. At any DevOps shop you quickly realize that you can’t watch everything even if you put it on big screens throughout the office where everyone can see. You need something to watch the data for you and to tell you what is important and what is not. The open source Kapacitor project, the last piece of InfluxData’s TICK stack, is just that tool. Kapacitor can watch all your metrics looking for anomalies and notify you of interesting or important events in your metrics. Todays presentation will walk you through the basics of anomaly detection with Kapacitor, how it works and how to know which algorithms to use for your various metrics.