🤖 Meet OnCall AI, our observability copilot that makes troubleshooting easy. Read announcement.

Skip to content
Back to Videos

Anomaly Detection and Troubleshooting

May 1, 2023 / 2:50
Detect every issue without having to predict them in advance. Edge Delta automatically surfaces anomalies and provides everything you need to troubleshoot. So, you can see every issue and know right away how to fix it.


One of the items our customers struggle with the most is the idea of identifying where in their log data they should focus their time and effort.

Edge Delta surfaces these anomalies for you automatically in real-time, anytime we recognize something meaningful.

In this example here, we're deployed to a Kubernetes cluster and have identified an anomaly in the search service. With one click, we can drill directly into the findings for this particular environment and for the search service see exactly what happened at that given moment in time.

Here we have an anomaly surfaced automatically with Edge Delta on this Kubernetes cluster, specifically for the search service in the production namespace.

This example anomaly has been identified by automatically understanding all the events in any given environment, and creating patterns on this data using the known historical baseline.

As you can see in this particular example, there are several brand-new events that have never occurred before. These events all have a negative sentiment, meaning Edge Delta has scored each and every event as a negative connotation event using our proprietary scoring mechanisms.

An example of a negative sentiment event would be something like a fatal or severe air level, or maybe something like an invalid access key occurring in the message.

The best part about this analysis is that it's run directly on your cluster at the edge and is completely autonomous with no configuration required.

Now, if we wanna know a little bit more about these negative sentiment events that Edge Delta has identified, we can, of course, drill in one layer deeper and analyze these messages in context.

Across the entire service, we know exactly what's occurring at any given moment in time, including this period where we've identified an anomaly in action.

Let's take a look at that known historical baseline. There's only 200 or so events occurring that are negative in connotation in nature.

Edge Delta's identified this variance with a significant deviation during this time window, which was hence why we identified the anomaly in the first place.

If I wanna know more about one of these specific patterns that Edge Delta has identified, we can quickly and easily drill into each of these, and understand what was happening at that given moment in time for this brand new pattern Edge Delta identified.

In this example here, we see that there's fatal errors occurring in our search service, indicating the anomaly in the first place.

Explore anomaly detection like this with Edge Delta for free at edgedelta.com.

(lively music)

(screen whooshing)

Related Videos