What's changed: Created Professional Cloud DevOps Engineer Chapter 3 (Domain 3 "SRE": SLI/SLO/SLA, error budgets, nines vs cost/opportunity cost; service lifecycle, capacity planning = quotas/limits/reservations/Dynamic Workload Scheduler, autoscaling = MIG/Cloud Run/GKE, incident mitigation = drain/redirect/add capacity/rollback).
3.1SLIs, SLOs, SLAs, and error budgets
Understand SLIs (availability, latency, etc.) that measure reliability, SLOs (targets), SLAs (external contracts), error budgets that balance change velocity and reliability, and the cost of reliability (number of "nines").
SRE (site reliability engineering) turns reliability into measurable targets and balances it against change velocity with data. It starts from SLIs/SLOs/SLAs and error budgets.
3.1.1SLIs, SLOs, SLAs
An SLI (Service Level Indicator) measures reliability (availability, latency, error rate). An SLO (Service Level Objective) is the internal target for that SLI (e.g., 99.9% availability). An SLA (Service Level Agreement) is the external contract with customers, with remedies on breach. Set SLOs stricter than SLAs so you notice internally before a contractual breach. Map "metric = SLI," "internal target = SLO," "external contract = SLA."
3.1.2Error budgets and the cost of reliability
An error budget is "100% − SLO," the allowed amount of unreliability (e.g., SLO 99.9% → 0.1% budget). With budget remaining, ship features aggressively; when exhausted, slow changes and restore reliability. This decides the balance of change velocity and reliability with data. Each added nine of reliability costs sharply more, so over-strict targets are an opportunity cost. Map "arbiter of velocity vs reliability = error budget."
Common: requirement → means. E.g., "metric of reliability" = SLI; "internal target" = SLO; "customer contract" = SLA; "arbitrate too-many features vs reliability" = error budget; "budget exhausted" = freeze changes and restore reliability. Set SLOs stricter than SLAs.
Watch the mix-ups: (1) Do not confuse SLO (internal target) with SLA (external contract)—keep SLO stricter than SLA. (2) Do not target 100% reliability (cost and opportunity cost spike). (3) The error budget is a decision tool, not punishment (a license for velocity).
3.1.3Section summary
- SLI = metric, SLO = internal target, SLA = external contract; keep SLO stricter than SLA
- Error budget = 100% − SLO arbitrates velocity vs reliability; on exhaustion, slow changes and recover
- Each nine costs much more—do not target 100%; mind opportunity cost
Sign in to track progress — Log in.
Quick check
(just a quick review)Q1. What is the indicator that measures service reliability such as availability or latency called?
Q2. What is the internal target value for an SLI (e.g., 99.9% availability) called?
Q3. What is the external customer contract with remedies on breach called?
Q4. If the SLO is 99.9%, what is the allowed unreliability (error budget)?
Q5. When the error budget is exhausted, what is the most appropriate SRE response?