DEV Community

loading...

Discussion on: Using Anomaly Alerts and Triggers Is Good For Predictive Troubleshooting

Collapse
gdcohen profile image
gdcohen

We (zebrium.com) are doing some interesting things with anomaly detection in log data. Our approach is quite different to others in that we use machine learning to structure log lines (we schematize and categorize log data into "event types" and then capture parameters into typed columns). This allows us to find anomalous patterns in log data which are often indicators of an underlying problem. I'd love to get your feedback on our approach.