● generator operational EN FR EN
nines.arewel.com
Service objectives, written like a manual.
Menu

SRE glossary

Plain-language definitions of the reliability vocabulary used across nines.

SLI (service level indicator)
A measurement of how well a service behaves from the user's point of view, usually a ratio: good events divided by total events. Read more →
SLO (service level objective)
A target for an SLI over a time window, for example 99.9% of requests succeed over 28 days. Read more →
SLA (service level agreement)
A contractual commitment to customers, with consequences if it is missed. It should be looser than the SLO, so you are warned before you breach it. Read more →
Error budget
The amount of failure an SLO allows: 100% minus the objective. A 99.9% SLO over 30 days allows about 43 minutes of full outage. Read more →
Burn rate
How fast the error budget is being consumed compared with the pace that would use it up exactly at the end of the window. At 14.4 times that pace, 2% of a 30-day budget is gone in one hour. Read more →
Multi-window, multi-burn-rate alert
An alert that fires only when both a long and a short window exceed a burn-rate threshold. The long window shows the problem is significant; the short one shows it is still happening. Read more →
Availability
The share of time, or of requests, in which a service works correctly. It is the most common SLI. Read more →
MTTR (mean time to restore)
The average time from the start of an incident to recovery. Read more →
MTTF (mean time to failure)
The average time a service runs between failures. Read more →
Recording rule
A Prometheus rule that pre-computes an expression and stores the result as a new time series, so dashboards and alerts stay fast.
Cardinality
The number of distinct time series a metric produces, driven by its label combinations. High cardinality increases memory use in Prometheus. Read more →
Business-hours SLI
An SLI measured only during working hours. nines generates a business_hours:active recording rule for it, because hour() and day_of_week() are evaluated at query time, not over history.