.avif)
Today, we’re announcing OnCall AI – our new observability copilot. OnCall AI summarizes anomalies in conversational text and provides recommendations on how to remediate them.
You can try OnCall for yourself by signing up for Edge Delta.
Customer-facing services and infrastructure generate massive amounts of logs and metrics. Historically, it’s been solely the user’s responsibility to make sense of this data and take action on it.
When monitoring, this means understanding what behaviors to alert against and taking the time to build monitors at scale. When troubleshooting, this means deciphering massive amounts of data to determine the root cause – akin to looking for a “needle in the haystack.”
‘At Edge Delta, we believe in a future where artificial intelligence makes observability easy and self-sustaining. That’s why we’ve built AI anomaly detection to automate your monitoring processes.
Today, we’re thrilled to further that mission with OnCall AI – the observability copilot that helps you get to the root cause of issues faster.
AI Recommendations for Faster Resolution
When Edge Delta detects an anomaly, it presents the findings within the Anomalies screen. Up to this point, we’ve surfaced:
- The time window of the anomaly
- The data that contributed to the anomaly
- The affected systems, components, or services
These findings have helped companies like Super League Gaming dramatically improve resolution times. As Justin Head, VP of DevOps at Super League explains: “Edge Delta helped us find things hours faster than we would have. It allows our developers to see – for the first time – what was making the most logs, what was giving the most errors. When things did go bump in the night, what changed specifically.”
OnCall AI streamlines this process further. When Edge Delta identifies an anomaly, OnCall AI:
- Analyzes the contents of the logs contributing to the anomaly
- Communicates the severity of the issue and what it’s impacting
- Summarizes the negative behavior in conversational text
- Provides a recommendation on how to resolve the issue
- In other words, OnCall AI helps move you along the troubleshooting process faster.
OnCall AI in a Real-World Scenario
Last Wednesday, the Edge Delta platform detected an anomaly in our own environment. The alert fired due to seven unique, negative sentiment patterns appearing at the same time.
The seven patterns were made up of 226 error and exception messages. These log events communicated an issue when saving changes to an SQL database.
Without Edge Delta, we would have interpreted the 226 messages as part of the troubleshooting process. Patterns made this task easier by bucketing similar messages together. OnCall AI further simplified the process by recommending potential resolutions, speeding up the overall troubleshooting process. /
In this scenario, OnCall AI recommended three potential paths to resolution. A quick glance at the summaries showed a problem with the database call. This helped us zero in on the problem as quickly as possible.
Review the database configuration and ensure that the identity insert is set to ON for the Trades table. Check for any missing or incorrect mappings between the entity and the database table.
This first suggestion might work if we need to solve the problem at the database level. However, in this scenario, we can see that we’re inserting data into the database without an ID. This resolution wouldn’t help us prevent calling the database with a bad input.
Check the network connection and the parameters being passed to the DbCommand. Ensure that the data types and values are correct.
The third suggestion helps us ensure we’re passing the correct parameters to the database, isolating the specific call that went wrong. This message helps us work backward and determine where to investigate next.
Investigate the inner exception details to identify the specific issue with the entity changes. Check for any constraints or triggers that might be causing the error.
The second suggestion is from the application itself. Here OnCall AI is highlighting the specific function that didn’t execute correctly. This suggestion guides us closer to the resolution.
A call from the network layer could add checks to prevent bad calls from occurring in the future.
These recommendations allowed us to resolve the problem in a matter of minutes.
This example is just a glimpse into what’s possible with OnCall AI. In the coming weeks, we’ll be expanding its functionality to apply AI to more use cases. Stay tuned!
Start Using OnCall AI Today
OnCall AI is generally available as of today. It is currently an opt-in feature. To enable OnCall AI, visit the My Organization, click edit, and select Yes under AI Enabled? .
From there, you can access the feature directly from the Anomalies screen. Try it for yourself by signing up for Edge Delta.
