Company Overview

Thanos is a cloud-native open-source project that provides highly available Prometheus setups with long-term storage capabilities. Originally started in 2017 by Fabian Reinartz and Bartek Plotka while at Improbable, Thanos addresses the challenges of Prometheus high availability and data persistence at scale. Using a sidecar architecture, Thanos uploads metric data to object storage without modifying Prometheus configurations, while providing a unified query layer across multiple Prometheus instances. Thanos became a CNCF graduated project in 2021 and is used by thousands of organizations worldwide for production observability.

Key Milestones

Year Event
2017 Project initiated by Fabian Reinartz and Bartek Plotka at Improbable to solve Prometheus HA and long-term storage
2018 Open-sourced on GitHub; released Thanos Sidecar, Thanos Store, and Thanos Query core components
2019 Added Thanos Compactor (compaction/downsampling) and Thanos Ruler; joined CNCF Sandbox
2020 Introduced Thanos Receive (Remote Write receiver); advanced to CNCF Incubating
2021 Graduated as a CNCF project; adopted by thousands of enterprises for production monitoring
2023 Released Thanos Query Frontend for improved query performance and multi-tenant support
2026 Continues active development; remains one of the most popular long-term storage solutions in the Prometheus ecosystem

Core Products

Component Description
Thanos Sidecar Sidecar agent deployed alongside Prometheus instances, uploading time-series data to object storage and exposing a Querier API
Thanos Store Provides read access to historical data in object storage with efficient label-based and time-range retrieval
Thanos Compactor Handles data compaction, downsampling, and deduplication to reduce storage costs and accelerate large-range queries
Thanos Ruler Standalone rule evaluation engine supporting cross-cluster Prometheus recording and alerting rules
Thanos Query Unified query layer aggregating data from multiple Prometheus instances, Sidecars, Store gateways, and other sources
Thanos Receive Implements the Remote Write receiver protocol, accepting metrics from external systems such as Grafana Agent and OpenTelemetry

Competitive Advantages

  • Global Query View: Thanos Query aggregates data across multiple Prometheus instances and object storage backends, providing a unified view across clusters and data centers.
  • Infinite Long-Term Storage: Leverages object storage (AWS S3, GCS, MinIO, etc.) to retain historical metrics indefinitely at very low cost, free from local disk constraints.
  • High Availability Architecture: Designed with no single point of failure; every component can be horizontally scaled, and individual Prometheus instance failures do not affect historical data availability.
  • Seamless Prometheus Integration: Works via sidecar mode without modifying Prometheus configuration or code, allowing smooth upgrades from existing deployments.
  • Automatic Downsampling: Thanos Compactor automatically downsamples historical data, significantly improving query performance across large time ranges.
  • CNCF Graduated Project: Mature governance, broad vendor support, and an active contributor ecosystem ensure long-term stability and growth.

Competitors

Key competitors in the Prometheus long-term storage and high availability space include:

Competitor Description
Grafana Mimir Prometheus-compatible long-term storage from Grafana Labs with built-in horizontal sharding and multi-tenancy
VictoriaMetrics High-performance metrics storage and query system with Prometheus-compatible API, capable of millions of samples/sec per node
Cortex Early Prometheus long-term storage solution (now deprecated), functionality evolved into Grafana Mimir
InfluxDB Time-series database usable for long-term metrics storage, though not natively Prometheus-compatible
Prometheus (standalone) Native Prometheus only, suitable for small-scale or single-cluster deployments lacking global view and long-term storage
Splunk Enterprise observability and log analytics platform with metrics storage capabilities for large organizations