Archera’s Anomaly system is built to reduce overhead on your team. Other systems require you to generate forecasts and set budgets. Once enabled Archera’s anomaly service will pull cost trends related to a specific service and look for cost deviations off of a trailing 30-90 day average trend. Since the size of the anomaly can vary depending on the workload type you can adjust the Minimal Threshold for anomalies as well to reduce over alarming.
To track anomaly corrections as well as improve our algorithm you can mark Anomalies in a number of Stages. They start as New, Reviewed, Resolved , and No Issue. No Issue our team will review for algorithm improvements. If you make notes and provide feedback we will tune or adopt our system appropriately.
Anomalies - Setup
Segment Settings - Enable Anomalies
You can enable anomalies for any segment by heading the the Segment Settings page and toggling them on. We recommend setting up Anomalies on Primary segments that are assigned to Groups. You can also adjust the minimal threshold by entering a monetary value and hitting enter. For Dev and Staging environments we recommend $50 and for established production environments we recommend around $100. This will look for immediate spikes that your team should remediate.
Notifications & Reviewing Anomalies
You can signup for Daily updates for Email and Slack in the Notifications Settings section under Settings for a specific segment. If you are the owner of the system we recommend signing up for the daily alerts. If you are a FinOps user you can look at the Events Page to do a quick verification of Anomalies to see what is new.
From both email and the notification page, links will take you to specific Anomalies in the Segment Explorer. The initial filter is by Service and is broken out by Usage type.
Example of an AWS Glue Service cost spike with Usage type breakdown.
Hovering on the Anomaly Row will show you the relative date with the cost spike