AWS Cost Anomaly Detection
AWS Cost Anomaly Detection is an Amazon Web Services feature that uses machine learning to continuously monitor cost and usage for unusual cloud spending.
Key points
- Uses machine learning models to learn normal spending patterns and flag unusual cost and usage activity.
- Continuously monitors spend rather than relying on static, manually configured thresholds.
- Complements AWS Budgets and AWS Cost Explorer as part of the AWS Cost Management toolset.
- Offered at no additional charge — there’s no extra cost to set up or run Cost Anomaly Detection.
- Supports both AWS managed and customer managed monitors across four dimensions:
- AWS services monitor — tracks all AWS services used by the account.
- Linked account monitor — tracks spending across member accounts.
- Cost allocation tag monitor — tracks all (or selected) values for a specified tag key.
- Cost category monitor — tracks all (or selected) values within a specified cost category.
- Sends a notification when actual spend deviates from the ML-predicted normal pattern by more than a user-set threshold — e.g., a 100 baseline triggers an alert once spend exceeds $110.