Search tools, AI agents, MCP servers
Submit a tool

Search iThub

Search tools, categories and collections. Type at least two characters.

Explore all tools
↑ ↓ to explore resultsEnter to open · Esc to close

Prometheus

CNCF metrics monitoring system with a pull-based scraper, time series database, PromQL and alerting rules

Monitoring & ObservabilityOpen sourcev3.16.0

Prometheus at a glance

Pricing
Open source
Deployment
Self-hosted
License
Open source
Version
v3.16.0November 5, 2026
Platforms
WindowsmacOSLinux
Language
Go
Data location
Self-hosted: metrics are stored on the local disk of each Prometheus server, or in external systems the operator configures through remote write.
Maker
prometheus
Founded
2012

What is Prometheus?

Prometheus is an open-source systems monitoring and alerting toolkit. According to the project overview, it was originally built at SoundCloud starting in 2012 and joined the Cloud Native Computing Foundation in 2016 as the second hosted project after Kubernetes. It is maintained independently of any company and is developed in the prometheus/prometheus repository under the Apache-2.0 license. The server is written in Go and ships as static binaries.

Prometheus stores metrics as time series identified by a metric name and key/value labels. The server collects samples by scraping HTTP endpoints of configured targets at fixed intervals, keeps them in a local time series database, runs recording and alerting rules, and exposes the data to PromQL queries through its web UI and HTTP API. Short-lived batch jobs can push metrics through the separate Pushgateway, and alerts are forwarded to Alertmanager, which handles grouping, silencing and notification. The wider ecosystem includes client libraries for instrumenting code and exporters for systems such as HAProxy, StatsD and Graphite. Grafana and other API consumers are commonly used for dashboards.

Precompiled binaries are published on the download page and as GitHub release assets for Linux, macOS, Windows, FreeBSD, NetBSD, DragonFly BSD, illumos and AIX on architectures from amd64 and arm64 to ppc64le, s390x and riscv64. Docker images are available as prom/prometheus on Docker Hub and on Quay.io. Besides regular minor releases, the project marks selected releases as long-term support; the download page lists 3.13.5 from October 21, 2026 as the current LTS release.

The latest release is v3.16.0, published on November 5, 2026. It makes smoothed and anchored vector selectors stable, adds experimental query cost estimation and execution limits (--enable-feature=query-cost), an experimental integral() function, a per-query header for start timestamp processing and metadata WAL records in Agent mode. It also improves XOR2 chunk compression, reduces OpenMetrics 2.0 parser allocations, and fixes several TSDB issues around out-of-order samples and WAL replay. Prometheus 3.0, released in November 2024, introduced a new web UI, UTF-8 metric and label names, Remote Write 2.0 and native OTLP ingestion.

The storage documentation states that local storage is not clustered or replicated, and points to remote write and remote read integrations for long-term or scalable storage. Other monitoring tools on iThub include Grafana, Zabbix and Netdata.

Key capabilities

  • Multi-dimensional data model Every time series is identified by a metric name and a set of key/value labels, so the same metric can be broken down by instance, job, path, status code or any other dimension. Counters, gauges, histograms (including native histograms) and summaries are the core metric types.
  • Pull-based scraping The server fetches metrics from HTTP endpoints of targets at a configured interval. Targets expose metrics in the Prometheus text format or OpenMetrics; release 3.16.0 continues work on an OpenMetrics 2.0 parser and offers Zstandard scrape compression behind a feature flag.
  • Service discovery Targets can be listed statically or discovered automatically. The configuration reference documents more than 30 discovery mechanisms, including Kubernetes, Consul, DNS, file-based, HTTP, AWS EC2, Azure, GCE, Docker, Docker Swarm, Nomad, OpenStack, Hetzner and DigitalOcean, with relabeling to filter and rewrite targets.
  • PromQL query language PromQL selects and aggregates time series by label, computes rates over counters, joins series with vector matching and evaluates functions over time ranges. Experimental functions and features are enabled individually with --enable-feature, and 3.16.0 adds optional query cost reporting and execution limits.
  • Local time series database Samples are stored on local disk in two-hour blocks with a write-ahead log and later compacted. Retention is set by time (default 15 days) or by size with --storage.tsdb.retention.time and --storage.tsdb.retention.size. The project states that local storage is not clustered or replicated.
  • Recording and alerting rules Recording rules precompute expensive expressions into new series, and alerting rules fire when a PromQL expression is true for a set duration. Firing alerts are sent to one or more Alertmanager instances, which route them to notification channels.
  • Web UI The built-in UI at port 9090 runs queries as tables or graphs, explains queries as a tree with per-node result counts, and shows targets, rules, alerts, service discovery state and configuration. The UI introduced in Prometheus 3.0 includes light and dark themes; the previous UI is still available behind the old-ui feature flag.
  • HTTP API The HTTP API under /api/v1 serves instant and range queries, series and label metadata, targets, rules, alerts and TSDB status. Admin endpoints for snapshots and deletion, and lifecycle endpoints for reload and shutdown, are disabled unless enabled by flag.
  • Remote write and remote read Prometheus can forward samples to external storage systems through remote write, including the Remote Write 2.0 protocol, and query them back through remote read. It can also act as a remote write receiver when started with --web.enable-remote-write-receiver.
  • OpenTelemetry ingestion With --web.enable-otlp-receiver, Prometheus accepts OTLP metrics over HTTP. Feature flags cover delta-to-cumulative conversion and native delta ingestion, and 3.16.0 fixes several OTLP exemplar handling issues.
  • Agent mode Started with --agent, Prometheus scrapes targets and forwards samples through remote write without local querying or rule evaluation. This mode is intended for clusters that send data to a central Prometheus-compatible store, and 3.16.0 adds metadata WAL records support to it.
  • Federation A Prometheus server can scrape selected series from other Prometheus servers through the /federate endpoint, which supports hierarchical and cross-service setups. Release 3.16.0 fixes the federation of float native histograms.
  • promtool The bundled promtool checks configuration and rule files, runs unit tests for rules, queries a running server and works with TSDB blocks, including backfilling. Version 3.16.0 prints test file names and allows custom maximum block durations.
  • Security options TLS and basic authentication for the web endpoints are configured through a web configuration file passed with --web.config.file. Since 3.16.0, the default external URL switches to https when that file enables TLS.
  • Binary installation Tarballs and zip files for many operating systems and CPU architectures are published for each release with SHA256 checksums. The server starts with prometheus --config.file=prometheus.yml and stores data in data/ by default.
  • Docker images docker run --name prometheus -d -p 127.0.0.1:9090:9090 prom/prometheus starts a container from the official image, which is published on Docker Hub and Quay.io.

Where Prometheus fits

Infrastructure and host monitoring

Exporters such as the Node exporter expose CPU, memory, disk and network metrics from Linux hosts, and Prometheus scrapes them on a schedule. PromQL expressions turn the raw counters into rates and utilization values, and recording rules keep frequently used aggregations precomputed for dashboards.

Kubernetes cluster monitoring

Kubernetes service discovery finds nodes, pods, services, endpoints and ingresses through the Kubernetes API, and relabeling selects which of them to scrape. This lets the set of monitored targets follow deployments and scaling events without editing static target lists.

Application instrumentation

Services are instrumented with official or community client libraries that expose counters, gauges and histograms on a metrics endpoint. Labels such as route, method and status code allow request rates, error ratios and latency percentiles to be calculated per endpoint with PromQL.

Alerting on conditions

Alerting rules evaluate PromQL expressions at regular intervals and send firing alerts to Alertmanager, which deduplicates, groups, silences and routes them to email, chat or paging services. promtool test rules runs unit tests against the rule files.

Long-term and central storage

Because local storage is a single-node database, setups that need long retention or a global view send data through remote write to compatible systems, or run Prometheus in Agent mode inside each cluster and forward samples to a central store. Federation provides a pull-based alternative for aggregating selected series from several servers.

Ingesting OpenTelemetry metrics

Applications that already emit OpenTelemetry metrics can push them to Prometheus over OTLP once the receiver is enabled. Prometheus 3.0 and later accept UTF-8 metric and label names, so OpenTelemetry names with dots can be stored and queried with quoted selectors. Dashboards are commonly built on top in Grafana.

Install

docker run --name prometheus -d -p 127.0.0.1:9090:9090 prom/prometheus

Reviews of Prometheus

No reviews yet

Be the first to share your experience with Prometheus.

Loading review form…

Found Prometheus useful? Upvote it todayYour vote helps other IT teams find it in the daily ranking.

Related guides

All guides

Similar to Prometheus

Compare:Prometheus vs GrafanaPrometheus vs ZabbixPrometheus vs Netdata

  1. ZabbixAGPL-licensed monitoring for networks, servers, VMs, applications and cloud services, with agents, proxies and an APIMonitoring & Observability72 views
  2. NetdataOpen-source, real-time infrastructure monitoringMonitoring & Observability22 views
  3. OpenObserveRust observability platform for logs, metrics, traces, RUM and LLM monitoring on Parquet and object storageMonitoring & Observability51 views
  4. SigNozOpen-source observability platform built on OpenTelemetryMonitoring & Observability27 views
  5. GrafanaAGPL-3.0 dashboards, Explore and alerting for metrics, logs and traces from Prometheus, Loki and other sourcesMonitoring & Observability62 views
  6. SentryError monitoring, tracing, logs, session replay and profiling with SDKs for web, mobile, backend and game enginesMonitoring & Observability58 views
  7. New RelicObservability platform for APM, infrastructure, logs, browser, mobile and synthetic monitoringMonitoring & Observability41 views
  8. Highlight.ioOpen-source session replay, errors, logs and traces; hosted service moved to LaunchDarkly in 2026Monitoring & Observability38 views