中文
devexamples
← Workflows

阅读中文版本

SLO and Error Budget Review Workflow

Review SLO attainment and error budget consumption, decide whether to slow releases or invest in reliability, and adjust SLOs that no longer reflect users.

intermediate~90 minreliabilitysreplanning

Objective

Use SLO and error budget data to make explicit release and reliability investment decisions.

Environment

  • SLO definitions and attainment data
  • A release and reliability planning cadence

Steps

1. Review attainment and budget consumption

Review attainment and budget consumption for each SLO.

Expected state: The reliability picture is current.

2. Decide release pace from budget: normal,

Decide release pace from budget: normal, cautious, or freeze.

Expected state: The decision is tied to data.

3. Invest in reliability when budget

Invest in reliability when budget is repeatedly exceeded.

Expected state: Root causes get attention.

4. Adjust SLOs that no longer match

Adjust SLOs that no longer match user expectations.

Expected state: SLOs stay meaningful.

Decision points

Freeze releases when budget is exhausted?

  • Slow and invest
  • Hard freeze

Slow releases and invest in reliability; a hard freeze can be counterproductive when the fix is a release.

Tighten or relax SLOs?

  • Match user experience
  • Always tighten

Set SLOs to match what users actually need; over-tight SLOs waste effort and cause freeze fatigue.

Expected outcome

Explicit decisions on release pace and reliability investment based on current SLO and error budget data.

Verification

Confirm the decision follows the agreed policy for the budget level and that any SLO change is justified by user impact data.

Cleanup

Record the decisions and the SLO definitions reviewed.

Overview

Use SLO and error budget data to make explicit release and reliability investment decisions. Confirm the decision follows the agreed policy for the budget level and that any SLO change is justified by user impact data.

Steps

  1. Review attainment and budget consumption for each SLO. Expected state: The reliability picture is current.
  2. Decide release pace from budget: normal, cautious, or freeze. Expected state: The decision is tied to data.
  3. Invest in reliability when budget is repeatedly exceeded. Expected state: Root causes get attention.
  4. Adjust SLOs that no longer match user expectations. Expected state: SLOs stay meaningful.
# budget consumed this period
budget_used=$(awk '$1=="budget" {print $2}' slo-report.txt)
echo "error budget used: ${budget_used}%"
# decision follows the policy threshold

Verification

  • Explicit decisions on release pace and reliability investment based on current SLO and error budget data.

Decision points

  • Freeze releases when budget is exhausted? Options: Slow and invest; Hard freeze. Recommended: Slow releases and invest in reliability; a hard freeze can be counterproductive when the fix is a release.
  • Tighten or relax SLOs? Options: Match user experience; Always tighten. Recommended: Set SLOs to match what users actually need; over-tight SLOs waste effort and cause freeze fatigue.