Lexicon
Error budget
An error budget is the unreliability an SLO permits: 100 percent minus the target. How budgets and burn rates turn reliability into a resource teams spend.
Definition
An error budget is the amount of unreliability an SLO permits: one hundred percent minus the target. A 99.9 percent monthly objective leaves a budget of 0.1 percent, roughly 43 minutes of full downtime or its equivalent spread across partial failures.
What it means in observability
The error budget reframes reliability as a resource to spend rather than a virtue to maximize. Budget remaining is permission: ship faster, run the risky migration, test in production. Budget exhausted is a signal with teeth: slow down, harden, pay the debt. That framing dissolves the classic standoff between teams paid to ship and teams paid to keep things up, because both are now managing the same number, and arguments about whether we can afford this risk become arithmetic instead of politics.
How it works in practice
The budget is computed from the SLI over the SLO window, and the operational tool is burn rate: the speed at which budget is being consumed relative to plan. A burn rate of one exhausts the budget exactly at window's end; a burn rate of ten is an emergency. Multi window burn alerts catch both shapes of trouble, a fast burn over minutes for incidents and a slow burn over days for degradations, with far less noise than static thresholds. Budget policies then define what actually changes when thresholds are crossed.
Where it gets hard
The politics arrive at enforcement. A budget policy that freezes releases is only real if leadership honors it during a launch week, and a budget nobody acts on is a dashboard, not a contract. Noisy or badly defined SLIs create phantom burns that spend credibility along with budget. And error budgets stack awkwardly across dependencies, since a service can burn its entire budget through a downstream failure it does not control.
Where Tsuga fits
Tsuga's formulas and time offset functions support budget and burn rate computation, and monitors can alert across burn windows, over telemetry complete enough that the budget reflects production rather than a sample of it.
Related terms
- MTTRMTTR is mean time to resolution, the average elapsed time from a failure occurring to service being restored.
- ObservabilityObservability is the ability to understand what is happening inside a system from the data it emits: its logs, metrics, and traces.
- SLIAn SLI, service level indicator, is the measurement behind an SLO: the specific, computable signal that stands in for whether users are having a good experience.
- SLOAn SLO, service level objective, is a target for how reliable a service should be, expressed as a measured indicator held over a time window: 99.9 percent of requests succeed over 30 days, or 95 percent of checkouts complete in under two seconds.