Error rate measures failed API requests within a defined time window. As a service-level metric, it reveals reliability problems and supports continuous feedback for API improvement. section 11.0 section 11.2
Use error rate effectively by:
- Tracking failures per second, minute, week, or month to expose usage and reliability patterns. section 11.2.1
- Separating HTTP 4xx user errors from 5xx server-side errors. section 11.2.1
- Combining error counts with request rate and duration in the RED method for microservice monitoring. section 11.2.1
- Adding contextual dimensions such as geography and application logs to investigate suspicious or infrastructure-related failures. section 11.2.1
- Configuring alerts when the rate exceeds a defined threshold and relating thresholds to API SLOs. section 11.2.1 section 11.4
Regional authentication failures
A large concentration of 401 Unauthorized responses in one geographic region may indicate malicious activity or bots. Application logs can then identify the infrastructure or application component behind a 5xx error. section 11.2.1
