Flashcards & Memory Anchors

To enforce controlled and reviewed changes in critical branches — preventing direct pushes, force pushes, and unreviewed merges.

To reduce duplication and ensure consistency across services. One change to the template propagates to all consuming pipelines.

Create one validated artefact and promote it across environments unchanged. The same image digest/binary that passed tests in dev reaches production.

A mandatory pass/fail threshold before a pipeline can proceed. Examples: minimum code coverage, zero critical CVEs, all tests passing.

When you need progressive, risk-controlled rollout with automated health-gate progression or rollback based on live metrics.

When fast cutover and rapid rollback are required. Blue stays running after switch — rollback is instant by re-routing traffic.

Deviation between declared infrastructure (code) and actual deployed state (portal/manual changes). Terraform detects and corrects drift on next apply.

To avoid exposure in code, logs, and static pipeline definitions. Runtime retrieval from Key Vault via managed identity means no credential is ever stored.

A Service Level Indicator — a measurable indicator of service performance, e.g., request success rate, p99 latency, error rate.

A Service Level Objective — a target for an SLI over a defined time period, e.g., "99.9% request success over 28 days".

To balance feature delivery speed with reliability targets. When the error budget runs low, slow down risky releases. When it's healthy, accelerate delivery.

Change failure rate — the percentage of deployments that result in a production incident requiring hotfix, rollback, or patch. Elite: < 15%.

Mean Time To Restore — the average time from incident detection to service restoration. Elite performers: under 1 hour.

User-impacting thresholds tied to SLO risk — not raw volume or noise. Alerts that require no action cause alert fatigue and hide real incidents.

To provide traceability for audit and troubleshooting — from a production incident back to the deployment, artefact, commit, PR, and original requirement.

Prefer automated, auditable, low-risk approaches. When two answers are close, the one that is automated AND auditable AND reduces blast radius wins.

To improve reproducibility and reduce supply-chain variance. A build should produce the same output given the same source — pinned deps ensure this.

Machine-enforceable governance rules in versioned definitions — e.g., Azure Policy, OPA, Checkov rules stored in Git alongside infrastructure code.

Triggers (the metric threshold that triggers it), owners (who executes), execution steps (exactly what to do), and verification checks (how to confirm success).

Choosing manual convenience over governed automation. If an answer involves someone clicking or emailing instead of a pipeline gate or automated check, it's almost certainly wrong.