Anomaly Detection
View Cost Anomalies
Cloud cost anomaly detection offers end-to-end monitoring of spikes in your Google Cloud or Amazon Web Services, seamlessly, across all your projects and services.
Cloud cost anomaly detection offers end-to-end monitoring of spikes in your Google Cloud or Amazon Web Services billing, seamlessly, across all your projects and services.
At the core of our cost anomaly detection service is our machine-learning algorithm. It works by monitoring billing data and analyzing the trend of spending in your environment. In addition, the machine learning model identifies billing patterns across many customers, forecasting your cloud spending, and constantly improving itself to provide even more accurate results.
Essentially, billing records that don't align with your anticipated spending behavior are identified for its deviation from the established patterns. You can also get insights into which resources are causing the anomalies and take corrective actions. Our anomaly detection engine is always running, however, you can opt to get email notifications, to get near-real-time email alerts on detected anomalies.

Manage anomaly notifications

To manage your notifications, select the Settings icon (a gear) from the top menu, then select Identity and access.
A screenshot showing the Identity and access screen
On this screen, select your own user account.
Then, on your user profile page, select Notifications from the left-hand menu:
A screenshot showing the Notifications screen
If you'd like to be notified of potential anomalies, check the Cloud Cost Anomalies box and then set the minimum severity threshold you want to be alerted on. There are three levels of severity:
  1. 1.
    Information
  2. 2.
    Warning
  3. 3.
    Critical
By default, the threshold for email alerts is set to Warning or higher (i.e., ">= Warning"). You can modify this value to increase or decrease the number of alerts you receive.
Please note: We begin analyzing your data as soon as you sign up. However, we need to gather at least seven full days of reference data before we have enough information to detect anomalies. If anomaly detection is critical to your operation, we strongly recommend you wait this period out before making any significant changes to your cloud spend.

Configure custom anomaly scopes

If you would like to monitor a specific subset of costs within your accounts, you can use attributions to define a custom scope.
For example, let's say you would like to create anomalies to only monitor your production infrastructure. You can accomplish this by creating an attribution defining your environment and then enabling it for monitoring.

View anomalies

We store all detected cost anomalies, so you can always access them by selecting Governance from the top menu bar, then selecting Cost anomalies (see below).

Investigate anomalies further

If you'd like to dive deeper into an anomaly, select the INVESTIGATE button in the top right-hand corner of the screen to launch a pre-configured Cloud Analytics Report. This feature allows you to zoom in on a specific anomaly and determine what any action (if any) is needed.

Fine-tune alerts with direct feedback

To reduce the number of false alerts, you can provide a thumbs-up or a thumbs-down for each alert. We use this information to fine-tune our anomaly detection algorithm and improve results.

Anomalies Widget

View the bite-sized video below for a closer look at anomaly detection: