sre-engineer

SLOs (Service Level Objectives) are quantitative reliability targets like 99.9% availability; SLIs (Service Level Indicators) are the metrics that measure them. Define SLIs by instrumenting golden signals—latency, error rate, saturation—then set SLO thresholds and track burn rate to monitor how quickly you consume your error budget before breaching the target.

An error budget is the allowable downtime or errors within a measurement period derived from your SLO—for example, 99.9% availability permits 43 minutes of downtime monthly. Track burn rate to detect when incidents consume the budget too quickly, then use that signal to trigger incident response, prioritize fixes, and decide when to halt deployments.

Toil is manual, repetitive operational work that doesn't scale. Automate runbooks with remediation steps, use monitoring thresholds to trigger automatic responses, and build chaos experiments to proactively discover and fix issues before incidents occur, freeing team capacity for strategic reliability improvements.

Yes. Error budget trends reveal when capacity or performance degrades, signaling when to scale infrastructure. Burn-rate thresholds inform on-call escalation policies and help forecast when you'll miss SLOs, enabling proactive capacity and staffing decisions before reliability suffers.

Blameless postmortems focus on systems and processes, not individuals. Document what failed, why detection and response took time, and what automation or monitoring gaps existed. Use postmortem findings to reduce toil, harden runbooks, and improve monitoring to prevent recurrence.

Yes. SLOs require instrumented metrics for the golden signals you're targeting—latency, error rate, and saturation. If monitoring is absent, instrument your service first, collect baseline data, then define realistic SLOs and error budgets based on observed performance.