Thanos
activeby Thanos community (CNCF) · since 2018
Sidecar-plus-object-storage layer giving Prometheus global queries, HA dedup and long retention.
Overview
Thanos extends existing Prometheus servers rather than replacing them. A sidecar uploads two-hour TSDB blocks to object storage, Store Gateways serve that history, the Querier fans PromQL out across every source and deduplicates HA pairs, and the Compactor downsamples old data. A Receive component also accepts remote write for push-based setups. A CNCF incubating project created at Improbable, it is a common choice for teams that want cheap long retention on S3 or GCS without moving to a fully centralised write path. The cost is many moving parts and query latency that depends on object-store reads.
Strengths & trade-offs
Strengths
- +Adds long retention to Prometheus without changing how scraping works
- +Global PromQL view across clusters with HA-pair deduplication
- +Object storage makes multi-year retention inexpensive
- +Downsampling speeds up long-range queries
- +Apache-2.0 and vendor-neutral under the CNCF
Trade-offs
- −Many components (sidecar, store, querier, compactor, ruler, receive) to operate
- −Recent data depends on reaching every sidecar, so a slow cluster slows global queries
- −Compactor must run as a singleton per bucket and can fall behind
- −Still pre-1.0 versioning after years of production use
Great for
- +Long-term storage for multi-cluster Prometheus
- +Global dashboards across regions
- +Cost-efficient metrics retention on S3/GCS
Think twice for
- −Single small Prometheus setups
- −Logs and traces
- −Teams wanting a single-binary backend
Capabilities
Native or official first-party support · 4 of 25 tracked
Scale & consistency
- Comfortable data size
- Massive · Multi-PB
- Horizontal scaling
- Native
- Multi-region
- No
- Transactions
- None
- Consistency
- Eventual
- Schema
- Schemaless
- Architecture
- Distributed, Shared storage
- Query languages
- PromQL
Each Prometheus stays local; global queries fan out across clusters and regions, history lives in object storage.
Deployment
Ecosystem
Alternatives to Thanos
Grafana Labs' horizontally scalable, multi-tenant long-term store for Prometheus and OTLP metrics.
thanos vs grafana-mimir →Resource-frugal, Prometheus-compatible metrics database, run as a single binary or a sharded cluster.
thanos vs victoriametrics →Uber-born distributed TSDB and aggregator for Prometheus-scale metrics with tunable replication.
thanos vs m3db →CNCF-graduated pull-based monitoring system with a local TSDB, PromQL and alerting rules.
thanos vs prometheus →CNCF multi-tenant, horizontally scalable long-term storage for Prometheus metrics on object storage.
thanos vs cortex →Label-indexed log aggregation: Prometheus-style streams, chunks on object storage, LogQL queries.
thanos vs grafana-loki →Thanos FAQ
What is Thanos?
Thanos is a time-series database by Thanos community (CNCF), first released in 2018. Thanos extends existing Prometheus servers rather than replacing them. A sidecar uploads two-hour TSDB blocks to object storage, Store Gateways serve that history, the Querier fans PromQL out across every source and deduplicates HA pairs, and the Compactor downsamples old data. A Receive component also accepts remote write for push-based setups. A CNCF incubating project created at Improbable, it is a common choice for teams that want cheap long retention on S3 or GCS without moving to a fully centralised write path. The cost is many moving parts and query latency that depends on object-store reads.
Is Thanos open source?
Yes. Thanos is open source. License: Apache License 2.0.
Does Thanos support ACID transactions?
No. Thanos does not provide ACID transactions; its consistency model is eventual.
Does Thanos scale horizontally?
Thanos scales horizontally out of the box, with built-in sharding or partitioning. It is comfortable at massive scale (Multi-PB).
How can I run Thanos?
You can self-host it.
What is Thanos best for?
Thanos is a strong fit for long-term storage for multi-cluster Prometheus, global dashboards across regions and cost-efficient metrics retention on S3/GCS. Think twice for single small Prometheus setups.
What are the best alternatives to Thanos?
Common alternatives to Thanos include Grafana Mimir, VictoriaMetrics, M3DB, Prometheus and Cortex. The right choice depends on your workload, scale and hosting; WhichDB's smart match ranks them against your requirements.
How popular is Thanos?
Thanos ranks #257 of 844 databases on the WhichDB Score (13.1/100), and #14 among time-series databases. The score combines public adoption signals such as package downloads, GitHub activity, job posts and developer discussion.