SLO settings and data sources - SRE Agent

This page lists the settings on an SLO and the data sources an SLI can use. For the task flow, see Set up SLOs.

Alert settings

Open an SLO and use the pencil icon to edit it. These settings control when it alerts:

Setting	What it does
Alert on burn rate	Alerts when the budget is burning faster than the thresholds.
Fast burn threshold	Compared with the burn rate over the last hour. Default 14.4.
Slow burn threshold	Compared with the burn rate over the last 6 hours. Default 1.0.
Alert on budget warning and Budget warning threshold (%)	Alerts when the remaining budget falls below the percentage. Default 20%.
Alert on tracking failure, Failure threshold (cycles) and Failure alert severity	Alerts when SRE Agent cannot compute the SLO for several cycles in a row. Default 3 cycles, severity warning.
Gate deploys on this SLO	Default, Always gate deploys or Never gate deploys. See Deploy gate.

A burn rate of 1.0 means you are spending the budget exactly as fast as the window allows. Breach alerts arrive like any other alert, so they can page people and trigger runbooks.

Data sources an SLO can use

SLI query type	Query language
Prometheus	PromQL
CloudWatch Metrics	Metric query (namespace, metric, statistic)
Loki	LogQL, wrapped in `count_over_time` or `rate` and summed
CloudWatch Logs	Logs Insights, ending in a `stats` aggregation. Set Default log group on the data source.
Datadog Metrics	Datadog metrics query, with `.as_count()` on counters
New Relic Metrics	NRQL, without `SINCE` or `UNTIL`. Set Account ID on the data source.
Synthetic	A check, not a query

An SLI is measured either as good events divided by total events, or as a single value compared with a threshold. Only Prometheus, CloudWatch Metrics and Loki sources appear in the wizard. The others you add by hand on Manage SLIs.

Connect your data: fields each data source needs.
Troubleshooting: what to check when an SLO shows degraded.