What is this?
Thanos is a set of components that turn existing Prometheus deployments into a highly available metric system with unlimited storage capacity. It is a CNCF Incubating project written in Go.
Thanos is a set of components that turn existing Prometheus deployments into a highly available metric system with unlimited storage capacity. It is a CNCF Incubating project written in Go.Is Thanos a single program or a set of separate components?A set of composable components shipped as one Go binary with subcommands — each subcommand does one thing well, e.g. 'thanos query' proxies calls to Store API endpoints and merges results.
A set of composable components shipped as one Go binary with subcommands — each subcommand does one thing well, e.g. 'thanos query' proxies calls to Store API endpoints and merges results.How do I get it running?Run the Sidecar next to each Prometheus in Kubernetes, or Receive to scale out ingestion. Images are at quay.io/thanos/thanos (mirrored as thanosio/thanos on Docker Hub); releases also ship tarballs for major platforms.
Run the Sidecar next to each Prometheus in Kubernetes, or Receive to scale out ingestion. Images are at quay.io/thanos/thanos (mirrored as thanosio/thanos on Docker Hub); releases also ship tarballs for major platforms.Do I need anything besides Prometheus to run it?Object storage is the only dependency, and even it is optional — history is stored in the native Prometheus 2.0 format on any object storage (S3, Google Cloud Storage and others) while keeping query latencies fast.
Object storage is the only dependency, and even it is optional — history is stored in the native Prometheus 2.0 format on any object storage (S3, Google Cloud Storage and others) while keeping query latencies fast.Will it work with my existing Prometheus setup, or do I have to reconfigure everything?No overhaul needed — Thanos integrates seamlessly with existing Prometheus setups and is simply added on top of them.
No overhaul needed — Thanos integrates seamlessly with existing Prometheus setups and is simply added on top of them.What do I actually gain over plain Prometheus?A global query view across all connected Prometheus servers, unlimited retention in object storage, deduplication and on-the-fly merging of metrics from Prometheus HA pairs, downsampling of historical data for massive query speedup, cross-cluster federation, and fault-tolerant query routing.
A global query view across all connected Prometheus servers, unlimited retention in object storage, deduplication and on-the-fly merging of metrics from Prometheus HA pairs, downsampling of historical data for massive query speedup, cross-cluster federation, and fault-tolerant query routing.Can Thanos ingest metrics from sources other than Prometheus?Yes — in Receive mode Thanos can scale out ingestion and integrate other remote-write compatible sources alongside Prometheus.
Yes — in Receive mode Thanos can scale out ingestion and integrate other remote-write compatible sources alongside Prometheus.How can my own tools read metrics out of Thanos?Through a simple gRPC 'Store API' that gives unified access to all metric data, plus easy integration points for custom metric providers.
Through a simple gRPC 'Store API' that gives unified access to all metric data, plus easy integration points for custom metric providers.What are similar projects?VictoriaMetrics, prometheus, openobserve, rook, litmus, coredns, etcd, TDengine
VictoriaMetrics, prometheus, openobserve, rook, litmus, coredns, etcd, TDengineWhich topics is it in?time-series-database, google-cloud-storage, high-availability, time-series, cncf, metrics, prometheus, grpc, s3, storage, infrastructure, cloud-native
time-series-database, google-cloud-storage, high-availability, time-series, cncf, metrics, prometheus, grpc, s3, storage, infrastructure, cloud-native