Menu
ESC

Nhập từ khóa để tìm kiếm

↑↓ Di chuyển
Enter Mở
ESC Đóng

Đang tải...

Bài 13 — Shift-Right Testing

Test Strategy and QA Leadership Bài 13/60

Shift-Right

Định nghĩa

Test in production — observe real users, run experiments live.

Practices shift-right

  • Feature flag rollout — gradual exposure (1% → 5% → 50% → 100%).
  • Canary deployment — new version to 1 server, monitor.
  • A/B testing — variant comparison.
  • Chaos engineering — inject failure intentionally (Netflix Chaos Monkey).
  • Synthetic monitoring — bot run critical user journey 24/7.
  • Real user monitoring (RUM) — track actual user behavior.
  • Error tracking — Sentry, Crashlytics capture exception.
  • Distributed tracing — Jaeger, Honeycomb trace request flow.

Shift-right + shift-left = "test everywhere"

  • Shift-left catch bug pre-prod.
  • Shift-right catch bug prod-fast and recover.
  • Loop: prod feedback → improve test → catch earlier.

SRE collaboration

QA + SRE/DevOps share on-call. Incidents → post-mortem → test added.

Production-safe testing

  • Read-only: synthetic monitoring, RUM.
  • Write but recoverable: test transaction marked as test (skip in reports).
  • Destructive: chaos engineering chỉ trong "game day" scheduled.

VN context

Banking, payment app phải có shift-right vì regulatory không cho phép down dài. SBV expect MTTR < 1h.

Tools VN team dùng

  • Sentry (error tracking).
  • DataDog / NewRelic (APM).
  • Grafana + Prometheus (metric).
  • PagerDuty / Opsgenie (incident).