Error Budget Language
5 exercises — Practice the English for error budget policy enforcement, burn rate interpretation, and linking reliability to deployment velocity.
0 / 5 completed
Quick reference: Error Budget
- Error budget — (1 − SLO) × window length = allowed unreliability
- Burn rate — how fast the budget is consumed relative to the window
- 14.4× burn rate — Google's threshold for an immediate page alert (99.9% SLO)
- Error budget policy — agreed action to take when budget thresholds are crossed
1 / 5
An SRE team has an SLO of 99.9% availability. After a P1 incident, the team reviews their status page: "We've burned 68% of our monthly error budget in the first two weeks." What would be the most appropriate response language in the next engineering review?
Burning >50% of the budget mid-month is the trigger for error budget policy enforcement.
An error budget policy defines what happens when the budget is consumed beyond a threshold. Common policy tiers:
• Yellow (50% burned): Increase monitoring, slow release cadence
• Orange (80% burned): Freeze non-critical deployments
• Red (100% burned/SLO breached): Full deployment freeze, all hands on reliability
Lowering the SLO to create more room is considered a last resort and requires stakeholder agreement — it's not a fix, it's an acceptance of lower reliability.
Key vocabulary:
• Error budget policy — agreed rules for what changes when budget is depleted
• Budget burn — the rate at which error budget is being consumed
• Deployment freeze — halting all releases to protect remaining budget
• Reliability work — engineering effort directed toward fixing the root causes of budget burn
An error budget policy defines what happens when the budget is consumed beyond a threshold. Common policy tiers:
• Yellow (50% burned): Increase monitoring, slow release cadence
• Orange (80% burned): Freeze non-critical deployments
• Red (100% burned/SLO breached): Full deployment freeze, all hands on reliability
Lowering the SLO to create more room is considered a last resort and requires stakeholder agreement — it's not a fix, it's an acceptance of lower reliability.
Key vocabulary:
• Error budget policy — agreed rules for what changes when budget is depleted
• Budget burn — the rate at which error budget is being consumed
• Deployment freeze — halting all releases to protect remaining budget
• Reliability work — engineering effort directed toward fixing the root causes of budget burn