Aggregation Operators

Aggregation reduces many series into fewer series. PromQL provides built-in operators to combine values across dimensions.

sum() — total across all matching series.

sum(rate(http_requests_total{job="demo"}[5m]))

This returns a single value: the total request rate across all instances, methods, and paths.

Other common aggregation operators:

• avg() — arithmetic average
• min() / max() — smallest / largest value
• count() — number of series
• stddev() / stdvar() — standard deviation / variance
• quantile(φ, ...) — the φ-quantile (e.g., quantile(0.95, ...) for the 95th percentile)
• topk(k, ...) / bottomk(k, ...) — largest / smallest k elements
• count_values("label", ...) — count distinct values

All aggregation operators support by() and without() to control which dimensions are preserved.