Imagine this: you wake up to find your Slack blowing up. Alerts are going off and users are complaining. You go straight to your Kibana dashboard, and you can see right away there's a problem. What's not clear is the reason why it happened? Since you can't tell the root cause, you realize you now have to go searching through the logs. You sink into despair... it's going to be a long night.
In this technical discussion, we'll describe an unsupervised machine learning approach that mimics the process of an experienced troubleshooter. Its goal is to identify the same root cause indicators in the logs that a human expert would have eventually found. The resultant payload is summarized using NLP and sent with relevant metrics to Elasticsearch for visualization alongside other observability data on any Kibana dashboard.
Gavin Cohen, VP of Product at Zebrium, will describe and demo the machine learning technology together with feedback from real-world users.