Promscale was an observability backend built by Timescale that used PostgreSQL with the TimescaleDB extension as the underlying store for Prometheus metrics and OpenTelemetry traces. It implemented the Prometheus remote_write and remote_read APIs so that a standard Prometheus server could stream samples into a Postgres hypertable, and it provided a PromQL query engine that translated PromQL into SQL against that schema.
The value proposition was full SQL access to Prometheus data: histograms, exemplars, and labels were all queryable with joins against business tables living in the same Postgres instance, and long-term retention and downsampling were handled by TimescaleDB’s continuous aggregates and compression policies. Promscale also accepted OTLP traces and stored them alongside metrics so that trace-to-metric correlation could be expressed as a SQL join instead of a cross-system link.
Timescale announced Promscale’s discontinuation in February 2023 and the repository is archived. Teams that evaluated Promscale have generally moved to Mimir, Thanos, VictoriaMetrics, or Cortex for metrics and to Tempo or Jaeger for traces. The project remains useful as a reference for how a SQL database can back a PromQL engine, but it should not be used for new deployments.
