Skip to content

Reliability trends and quarterly SLO review

The Reliability trend report runs at the start of each quarter and can be started manually. It compares the current 90-day period with the preceding 90-day period for each configured environment:

  • availability error ratio against the 99.9% starting SLO;
  • average request latency; and
  • percentage change from the previous period.

The JSON and Markdown reports are uploaded as workflow artifacts. Missing Prometheus configuration skips an environment; configure the same STAGING_PROMETHEUS_URL, PRODUCTION_PROMETHEUS_URL, and optional PROMETHEUS_BEARER_TOKEN used by SLO burn-rate reporting.

Quarterly review checklist

  1. Download the trend reports and compare error ratio, latency, traffic, and burn-rate artifacts with the previous quarter.
  2. Overlay deployment-annotation.json events and identify releases or dependency changes near material trend shifts.
  3. Review incidents, drift issues, synthetic-monitoring issues, and rollback readiness during the period.
  4. Confirm SLO scope, exclusions, alert thresholds, and data quality. Do not raise the target solely to silence alerts.
  5. Record capacity, database, cache, email, messaging, and external-adapter risks; assign owners and due dates for follow-up work.
  6. Close the quarterly review issue only after the report, decisions, and follow-up tasks are linked.

The report status is ok when no review threshold is crossed, review when error ratio or trend deterioration is material, and critical when the change is severe enough to require incident follow-up. These are starting thresholds; service owners should tune them with real traffic and error-budget policy.