SRE defined
SRE (Site Reliability Engineering) is a discipline proposed and practiced by Google in 2003 that applies software-engineering methods to IT operations. Its central question: how to use automation and engineering to maintain service reliability while the business grows rapidly.
Core SRE practices
Key SRE practices include:
- SLO (Service Level Objective): define an acceptable unavailability budget per service — e.g. 99.9% availability means at most ~43 minutes of downtime per month
- Error budget: when the error budget is exhausted, all non-emergency changes are frozen until reliability recovers
- Automation: turn repetitive operations into code to reduce human error
- Blameless Postmortem: after incidents, focus on system defects rather than individual mistakes
- Capacity planning on demand: procure based on real traffic curves, not peak estimates
Where physical infrastructure fits in SRE
Traditional SRE discussions focus on software — code deployment, configuration management, capacity planning — but physical-infrastructure failures often account for 30–50% of data center downtime. Server power failures, PDU overloads, room cooling outages, and failed switch optical modules burn error budget too. CloudSino DCOS feeds physical-layer telemetry (BMC events, PDU power, temperature/humidity) into the SRE monitoring pipeline, making SRE teams accountable for end-to-end availability from chip to rack.