Attention required!

Be the first to know when your servers start to get slow.

I'm proud to announce that today we released a smart alerts system for our customers. From now on, Nazar will send email notifications every time that a server changes its behavior and in this post I'll explain how it works.

Why?

The secret to anticipate that daily performance problems will scale and become critical issues is to observe them continuously.

Everyday something new happens and the application will change its behavior in order to adapt to the new scenario.

It could be a new software release, the increasing number of users, a new application sharing the same database, etc. There is a long list of possible reasons for that and once the customer understands it, they come to the conclusion that is better and cheaper to keep monitoring and be notified whenever any change occurs.

The dashboard shows which servers are changing their behavior.

There are 2 fundamental metrics that we monitor in order to keep track of a database server performance:

Number of slow executions and Time Spent

Number of slow executions in the last 14 days

Time spent with slow executions in the last 14 days.

The number of slow executions (which takes more than 100ms to run) and the total time spent with them reflect the changes in the behavior and it clearly shows when a server is being overloaded.

When our software analyzes these two metrics together it shows some patterns and every application will present a unique behavior. Our software will learn that and when some unexpected behavior is detected your team will get notified.

Basically, we do a comparison of what happened in terms of behavior between today and the last 7 days. Based on that, our smart alerts system will appear to you in three distinct colors: yellow, orange and red.

1 — Yellow

SQL Server database’s behavior starting to change.

If any of these 2 metrics, number of slow executions or time spent with slow queries, increase more than 10% compared to the last 7 days your team will receive an email notification.

2 — Orange

You can see that there’s a continuous increase from day 6 to 7.

If the increase in the metrics are higher than 25% and less than 50%, the team will receive another email with the orange alert.

3 — Red

At this example, day 7 shows a peak.

In the worst scenario, if the difference is higher than 50%, the red alert will be triggered.

Here is how a dashboard with multiple servers looks like:

Servers that require attention because of behavior changes.

We’re sitting on a lot of data that made possible to develop our smart alerts system and our goal is to help customers addressing their performance issues as they are about to become a real problem.

If you have any feedback or thoughts, feel free to comment.