stream-aggregation-helper

VictoriaMetrics stream aggregation reduces cardinality, ingestion load, and query cost by processing metrics at vmagent before storage. It groups data by selected labels and applies output functions like rate_sum or quantiles.

VictoriaMetrics stream aggregation supports by/without label configurations and outputs including rate_sum, total, min/max, quantiles, and histogram_bucket. These options allow flexible metric grouping for cardinality reduction.

Rolling out stream aggregation in vmagent follows a structured phase-guided process spanning planning, deployment, and verification. Cross-check queries confirm data equivalence between aggregated and raw metrics.

VictoriaMetrics stream aggregation supports high availability and sharded vmagent deployments. The design phase includes topology considerations to ensure aggregation rules function correctly across distributed architectures.

Use stream aggregation when you need to reduce cardinality and sample frequency at ingestion time via vmagent, before data reaches storage. Recording rules query existing stored data, whereas stream aggregation pre-processes incoming streams.

Designing stream aggregation rules follows five phases: gate, intake, decide, config, and verify. This structured approach ensures safe aggregation planning by evaluating requirements for labels, intervals, outputs, and rollout steps.