Server monitoring statistical analysis – anomaly detection

We are excited to announce a new feature for our server monitoring service, Server Density – statistical analysis. Using the data we store based on the historical behaviour of your servers, we are able to make predictions about what we expect values to look like in the future. Where the actual data deviates, we highlight the anomalies so you can investigate further.

We start displaying anomalies 48 hours after you add a server but then analysis is performed every hour. If there’s no label on your charts then there are no anomalies detected.

This is available right now for trial and paid users.

Why is it useful?

Sometimes this will be nothing to worry about but often it will help detect situations that might not otherwise be discovered right away. For example, a sudden decrease in network traffic transmitted might indicate that your nightly backup job has failed. Or, an unusual increase in Apache requests could signify a robot downloading all your content.

Where can I see it?

Anomalies are highlighted on the graphs and you can click through to the server snapshot to see the actual values, and a few hints about what to do with that information. This is the first step towards providing you with more proactive information to help you manage your servers, and we have many other ideas about what we can do with the data.

Every server is different so our hints are quite generic right now. We hope to be able to offer more tailored suggestions as we work on improving the functionality in the future.

How does it work?

Using a modified Holt-Winters algorithm, Server Density uses past behaviour to predict future behaviour, future variability, and measure whether current behaviour is within the expected bounds, given the behaviour that could be predicted from the past.

The algorithm trains itself as it goes along – it will get more accurate at detecting anomalies the more history it has to work with. So after 48 hours it will start trying to detect anomalies, but it’ll carry on getting better after that.

You can find out more about how it works in our documentation.

Any questions or problems, just let us know.