● Beta — KubeEra is in active development. Profiles are live; AI-perception measurement is rolling out.
KubeEra
P

Prometheus

GraduatedObservability
Open source · CNCF · CNCF Graduated · Apache-2.0

Prometheus is a metrics-based monitoring and alerting system. It scrapes time-series data over HTTP from instrumented targets, stores it in a local TSDB, and lets you query it with PromQL.

updated 2026-07-01

Prometheus

What it is

Prometheus is a metrics-based monitoring and alerting system. It scrapes time-series data over HTTP from instrumented targets, stores it in a local TSDB, and lets you query it with PromQL. It also ships an alertmanager-compatible alerting pipeline for routing and deduplicating alerts. It is the de facto metrics standard in the Kubernetes ecosystem — kubelet, kube-state-metrics, cAdvisor, and most Kubernetes-native tooling expose Prometheus-format metrics by default.

It is not a full observability stack on its own. It handles metrics. Logs and traces are out of scope, which is why it's almost always deployed alongside Loki, Tempo, Jaeger, or a commercial backend.

Who builds it and why

Originally built at SoundCloud in 2012, modeled on Google's internal Borgmon system. Donated to CNCF in 2016 as the second project ever accepted (after Kubernetes itself), and graduated in 2018 — one of the earliest graduated projects.

Today it's maintained by a broad, decentralized group of maintainers across multiple companies (Grafana Labs, Red Hat, and independent contributors among them), not a single vendor. 1,407 contributors and continuous commit activity (last commit 2026-07-01) indicate the project isn't coasting — this is active core infrastructure, not a legacy dependency on life support.

Production readiness signal

Strong. This is one of the least risky choices you can make in the CNCF landscape:

  • Graduated CNCF project — the highest maturity tier, meaning broad adoption, documented governance, and security audit history.
  • 64,822 GitHub stars, 1,407 contributors — deep bench strength, no single-maintainer bus-factor risk.
  • Active release cadence — v3.12.0 latest, with v3.x representing a significant architectural evolution (new storage engine work, UTF-8 metric name support) rather than a stagnant v1-forever project.
  • Apache-2.0 — no licensing surprises, no relicensing risk (unlike some observability tools that moved to BSL).
  • Runs at extreme scale in production at thousands of companies; failure modes are well understood and well documented.

The main operational caveat isn't the software's maturity — it's that single-node Prometheus has real scaling limits (local disk, no long-term retention by default) that you need to plan around architecturally.

Who should use this

  • Teams running Kubernetes who want metrics in a format every other tool already understands.
  • Platform teams that need PromQL-based alerting integrated with Alertmanager.
  • Anyone who wants to avoid vendor lock-in on the metrics layer — the scrape format and query language are now an ecosystem standard (Mimir, Thanos, Cortex, VictoriaMetrics all speak Prometheus-compatible protocols).
  • Organizations comfortable operating stateful infrastructure themselves, or pairing Prometheus with a managed remote-write backend.

Who should NOT use this

  • Teams needing long-term, high-cardinality retention out of the box. Vanilla Prometheus is not built for multi-year storage or massive cardinality without bolting on Thanos, Cortex, or Mimir. If you don't want to operate that stack, look elsewhere.
  • Teams that need logs/traces/metrics unified in one product with minimal setup. Prometheus solves one of three pillars; if you want a single agent and single pane of glass, a commercial APM tool is a better fit.
  • Small teams with no ops capacity. Self-hosted Prometheus + Alertmanager + Grafana + storage backend is real operational overhead. If nobody owns it, it will rot.
  • High-cardinality event-style data (per-user, per-request labels). PromQL and TSDB were not designed for this; you'll hit cardinality explosions and cost/performance walls.

Alternatives

  • VictoriaMetrics — Prometheus-compatible, better long-term storage and cardinality handling, lower resource footprint.
  • Datadog / New Relic — commercial, fully managed, unifies metrics/logs/traces, but expensive at scale and vendor lock-in.
  • Grafana Cloud (Mimir) — managed Prometheus-compatible backend if you want the ecosystem without operating storage yourself.

Pricing

Fully open source, Apache-2.0. No license cost. Your cost is entirely operational: compute, storage, and engineering time to run and scale it (or a paid managed backend like Grafana Cloud, Amazon Managed Prometheus, or Chronosphere if you don't want to self-host).

Frequently asked

What is Prometheus?+
Prometheus is a metrics-based monitoring and alerting system. It scrapes time-series data over HTTP from instrumented targets, stores it in a local TSDB, and lets you query it with PromQL. It also ships an alertmanager-compatible alerting pipeline for routing and deduplicating alerts.
Who builds Prometheus?+
Originally built at SoundCloud in 2012, modeled on Google's internal Borgmon system. Donated to CNCF in 2016 as the second project ever accepted (after Kubernetes itself), and graduated in 2018 — one of the earliest graduated projects.
Is Prometheus production ready?+
Strong. This is one of the least risky choices you can make in the CNCF landscape: Graduated CNCF project — the highest maturity tier, meaning broad adoption, documented governance, and security audit history. 64,822 GitHub stars, 1,407 contributors — deep bench strength, no single-maintainer bus-factor risk.
Who should use Prometheus?+
Teams running Kubernetes who want metrics in a format every other tool already understands. Platform teams that need PromQL-based alerting integrated with Alertmanager.
Who should not use Prometheus?+
Teams needing long-term, high-cardinality retention out of the box. Vanilla Prometheus is not built for multi-year storage or massive cardinality without bolting on Thanos, Cortex, or Mimir. If you don't want to operate that stack, look elsewhere.
What are the alternatives to Prometheus?+
VictoriaMetrics — Prometheus-compatible, better long-term storage and cardinality handling, lower resource footprint. Datadog / New Relic — commercial, fully managed, unifies metrics/logs/traces, but expensive at scale and vendor lock-in.
How much does Prometheus cost?+
Fully open source, Apache-2.0. No license cost. Your cost is entirely operational: compute, storage, and engineering time to run and scale it (or a paid managed backend like Grafana Cloud, Amazon Managed Prometheus, or Chronosphere if you don't want to self-host).