Skip to main content

🧩 SLO / SRE (Site Reliability Engineering)

✅ Overview​

Practical system to achieve reliability by "Engineering". Design availability with SLO (Objective) and Error Budget.

✅ Problems Addressed​

  • Develop and operate with ambiguous definition of availability.
  • Conflict of "Stability vs Development Speed" does not converge.
  • Response at failure is personalized.
  • Culture trying to realize high availability with "guts".

✅ Basic Philosophy & Rules​

  • Three stages: SLI (Indicator) → SLO (Objective) → SLA (Agreement).
  • Balance control by "Error Budget".
    → If reliability is sufficient, increase development speed.
    If used up, focus on stabilization.
  • Standardization of incident response (On-call, Runbook).
  • Post-mortem culture (Improvement rather than blame).

✅ Suitable Applications​

  • Internet-scale services.
  • Microservices and distributed systems.
  • Strong availability requirements (99.9% - 99.999%).

❌ Unsuitable Cases​

  • Small-scale apps with low availability requirements.
  • Organization where operation culture is not nurtured (Organization training is needed).

✅ History​

  • Systematized starting from Google SRE.
  • SLO / Error Budget became standard and adopted as metrics in cloud era.

✅ Related Styles​

✅ Summary​

SRE is an approach to "guarantee reliability with code",
and availability management by SLO + Error Budget becomes central.