With applied intelligence's anomaly detection, New Relic alerts your team of any unusual behavior instantly. New Relic uses applied intelligence to constantly observe your applications. We use this information to determine your application's baseline, or expected, performance. Whenever behavior deviates from the baseline, we know right away and alert your team so you can address any errors promptly and efficiently.
There are two types of anomaly detection at New Relic: custom and automatic. Learn about which anomaly detection is right for each situation your team would like to monitor and how to implement anomaly detection in your system.
Go to one.newrelic.com > Alerts & AI. Click the Anomalies tab to see the dashboard where your team can monitor any unusual behavior in your system.
How we use anomalies
At New Relic, our own developers know how important it's to monitor the health of our applications. We want our customers to have access to the data they need whenever they need it so our team needs to be alerted if there are any outliers in our system's performance. The anomaly detection of New Relic uses applied intelligence to monitor three key golden signals: throughput, error rate, and latency. With anomaly detection, our developers monitor the baseline performance for these metrics.
So, let's say that one afternoon there's a spike in response time and it's taking longer than usual for our customers to access the homepage. Anomaly detection will flag this anomalous behavior because our latency metric data has deviated from its baseline. This doesn't necessarily mean there is a problem, it just indicates that AI has registered something out of the ordinary in our system and we should take a deeper look.
We monitor this unusual behavior in a few ways. First, our team uses the anomaly dashboard so we can see what changed and when.
There are two types of anomaly detection: automatic and custom
Automatic anomalies are the most efficient way for your team to learn about unusual behavior in your APM-monitored applications. Automatic anomaly detection is a hands-off tool your team can implement to ensure that you're notified the moment behavior in your application deviates from baseline. You can use automatic anomalies to identify the source of the problem and take the appropriate steps to get your system running smoothly again.
Custom anomalies allow increased configurability for your team. Custom anomalies provide your team with the capability to alert on any NQRL condition and to adjust and optimize your thresholds. Custom anomalies also use the same advanced tuning settings as static alerting so you can ensure your team sees only the anomaly incidents important to you.
When To Use
When you need to set a single threshold for all your data.
All entities, all signals
When you want to automatically learn trends in your data but have control over the threshold
All entities, all signals
When you want a broad understanding of changes in key metrics on your applications and services with no configuration needed. Data trends and thresholds are automatically determined through our machine learning engine.
entities, golden signals