Self-Hosted Observability Directory

VictoriaMetrics

Prometheus-compatible TSDB optimized for ingestion rate and storage cost. Single-binary by default, cluster mode for scale.

LicenseApache-2.0
Repositoryhttps://github.com/VictoriaMetrics/VictoriaMetrics
GitHub stars (approx)17000
LanguageGo
Self-host deploymentsingle binary, docker, kubernetes helm/operator
Commercial hosted optionyes (vendor-run)
Freshness statusFresh
Freshness checked2026-05-17
Categoriesmetrics
Replaces SaaSdatadog, new-relic

VictoriaMetrics is what you reach for when Prometheus’ default storage starts hurting at >1M active series, or when retention requirements grow past local disk. PromQL-compatible (with MetricsQL extensions), reads Prometheus and Graphite line protocols, ships a single 20MB binary.

Pick it if: you want a Prometheus drop-in with substantially lower memory and disk footprint, or you need long-term retention without the operational weight of Thanos/Mimir.

Skip it if: you’re already committed to the Grafana Labs stack (Mimir integrates more naturally with Grafana Cloud tooling). Also skip if you need the most cutting-edge OTLP metric semantics — VM’s OTLP support is solid but Prometheus-first.

Deployment: single binary for <10M series; cluster mode (vmstorage + vminsert + vmselect) for horizontal scale. Helm chart and operator both first-class.

← Back to directory