# TimescaleDB

Hypertables, columnar compression and continuous aggregates bolted onto plain Postgres.

*https://stockmarketstack.com/tools/timescaledb · Tick Data Storage & Time-Series Databases*

## Facts

### At a glance

| Field | Value |
| --- | --- |
| Vendor | Tiger Data |
| Category | Tick Data Storage & Time-Series Databases |
| Job | tick_storage |
| Website | https://www.tigerdata.com/timescaledb |
| Pricing model | usage |
| Free tier | true |
| Open source | true |
| Licence | Apache-2.0 core; Timescale License (source-available, not OSI) for compression, continuous aggregates, policies and gapfill |
| Self-hosted | true |
| Tested hands-on | false |
| Last updated | 2026-09-13 |

### Pricing

| Tier | USD | Period |
| --- | --- | --- |
| Self-hosted | 0 USD | month |
| Tiger Cloud Performance | 30 USD | month |
| Tiger Cloud Scale | 36 USD | month |
| Tiger Cloud Enterprise | on request | month |
| TimescaleDB Enterprise | on request | month |

### Coverage

| Field | Value |
| --- | --- |
| Asset classes | none |
| Markets | none |
| Works outside the US | false |
| Data latency | none |
| Platforms | web, library |
| AI features | none |

### Interfaces

| Field | Value |
| --- | --- |
| API | false |
| Webhooks | false |
| Scripting | SQL |
| Python | false |
| Spreadsheet add-in | false |
| MCP server | false |
| Export | csv, json |

### Capabilities

Yes: none

No: charting, screening, scanning, backtesting, automation, live_trading, paper_trading, portfolio_tracking, broker_import, tax_reporting, alerts, news, options_analysis

*Verified: pricing 2026-09-13; capabilities 2026-09-13; coverage 2026-09-13.*

## What it is

TimescaleDB is a PostgreSQL extension. You run `CREATE EXTENSION timescaledb`, convert a table
into a hypertable, and Postgres starts partitioning it into time-ordered chunks behind the same
table name. Nothing changes on the client side — same wire protocol, same drivers, same joins
against your reference data — and that is the whole argument for it over a purpose-built tick
store. On top of partitioning it adds a columnstore (branded Hypercore) that converts older
chunks to columnar form, continuous aggregates (incrementally refreshed materialised views, which
is how you keep one-minute and daily bars current without recomputing them), retention and
compaction policies run by a job scheduler, and time-series SQL — `time_bucket`,
`time_bucket_gapfill`, `locf`, `interpolate`, `first` and `last`.

The publisher renamed itself. Timescale, Inc. has traded as Tiger Data since 17 June 2025 and the
managed service is now Tiger Cloud; the extension kept the name TimescaleDB. Development is
unambiguously live: 2.30.0 shipped on 8 September 2026, the last commit on `main` was 11
September 2026, and the repository is not archived. PostgreSQL 16, 17 and 18 are supported —
2.29.0 dropped 15 in July 2026.

## Pricing

Running it yourself costs nothing in either edition. Tiger Cloud is metered rather than
subscribed — compute hourly, storage on average GB-hours — so the plan headlines are floors, not
prices. Performance starts at $30 a month of compute with storage at $0.177/GB-month; Scale
starts at $36 with storage at $0.212/GB-month and tiered object storage at $0.021/GB-month;
Enterprise is a contact form. New accounts get a 30-day trial carrying $1,000 of credit with no
card, and every plan includes two free services, currently in beta. TimescaleDB Enterprise, a
commercially licensed self-managed build announced on 20 April 2026, is still an early-access
waitlist with no public price.

## Data & coverage

None, in the sense this catalogue usually means. TimescaleDB ships no market data — no feed, no
history, no symbology. You load ticks or bars from a provider and it stores them.

What it stores them *with* depends on the licence, and the split is not where most readers assume
it is. Apache-2.0 covers hypertables and chunk management, `time_bucket`, `first`, `last`,
`histogram` and the informational views. The Timescale License covers most of what people install
it for: compression and the columnstore, continuous aggregates, retention policies, the job
scheduler, SkipScan, chunk reorder, move and split, and gapfill with `locf` and `interpolate` —
source-available, free to self-host and modify, closed to anyone reselling it as a service.
Vendor compression claims run from "90%+ typical" in the README to "up to 98%" in the billing
docs; neither is tested here, and both turn on how you set `segmentby` and `orderby`.

## Integrations

Anything that speaks PostgreSQL, which is most things: Grafana, dbt, psycopg and SQLAlchemy,
JDBC, every BI tool. The vendor also publishes Docker images, a Ruby gem, a Terraform provider
and Tiger CLI, which bundles an MCP server for managing Tiger Cloud services — the extension
itself ships none. Self-hosted builds are supported on Linux, Windows and macOS.

## Limitations

- The licence, not the code, is what to read first. The Community Edition cannot be offered as a
  service, so a product that hosts TimescaleDB for its own customers needs a conversation with
  the vendor.
- The genuinely open edition is thin. Remove compression, continuous aggregates, retention and the
  scheduler and what is left is partitioning plus a bucketing function. Since 2.26.4 in April 2026
  an Apache-only build errors outright when you query a compressed chunk.
- One machine. Distributed hypertables were sunsetted in 2.14 and never replaced; scaling out
  means read replicas, not sharding. Against a columnar engine on full-tick multi-venue feeds a
  single Postgres write path is the ceiling.
- Self-hosting is a Postgres operations job — vacuum, WAL, backups, failover, major-version
  upgrades — plus chunk intervals and compression keys to tune.
- Tiering old chunks to object storage is a Tiger Cloud feature, not part of the self-hosted
  extension. Self-hosted, cold data sits on the same disk as hot.

## Alternatives

[QuestDB](https://stockmarketstack.com/tools/questdb) and [ClickHouse](https://stockmarketstack.com/tools/clickhouse) are the columnar answers: faster on wide scans over billions of rows, and
neither gives you Postgres joins, constraints or the surrounding ecosystem. [ArcticDB](https://stockmarketstack.com/tools/arcticdb) is the
opposite trade — versioned DataFrames straight onto S3, no server to run and no SQL at all. [kdb+](https://stockmarketstack.com/tools/kdb)
still wins on raw tick throughput and still costs what it costs.

## FAQ

### Is TimescaleDB open source?

Half of it. Code outside the repository's `tsl` directory is Apache-2.0 and unrestricted; code inside it is under the Timescale License, which is source-available rather than OSI-approved. Self-hosting either is free, and modifying either is allowed, but you may not offer the licensed half as a hosted service.

### Which features do I lose on the Apache 2 edition?

Columnar compression, continuous aggregates, retention policies, the job scheduler, SkipScan, chunk reorder/move/split, and the gapfill family — time_bucket_gapfill, locf and interpolate. What stays is hypertables, time_bucket, first, last, histogram and the informational views. Since 2.26.4 in April 2026, querying a compressed chunk on the Apache build raises an error outright.

### Was TimescaleDB renamed?

The company was, not the extension. Timescale, Inc. began trading as Tiger Data on 17 June 2025 and the managed service became Tiger Cloud, but the Postgres extension is still called TimescaleDB and still installs as `timescaledb`.

### Does it come with market data?

No. It is storage and query machinery only — no quotes, no history, no symbology, no vendor feed. You load the ticks yourself from a data provider and TimescaleDB stores them.

## Also worth comparing

- [ClickHouse](https://stockmarketstack.com/tools/clickhouse.md) — Columnar OLAP database that keeps years of ticks on disk cheaply and scans them fast.
- [Apache Druid](https://stockmarketstack.com/tools/druid.md) — Real-time OLAP cluster built for high-concurrency dashboards, not for research joins.
- [ArcticDB](https://stockmarketstack.com/tools/arcticdb.md) — Versioned Pandas frames written straight onto S3, with no server to run.
- [DolphinDB](https://stockmarketstack.com/tools/dolphindb.md) — Closed-source tick database with a vector language, as-of joins and streaming engines.
- [DuckDB](https://stockmarketstack.com/tools/duckdb.md) — In-process analytical SQL over Parquet tick files, with no server to run.
- [InfluxDB 3](https://stockmarketstack.com/tools/influxdb.md) — Line-protocol time-series database rewritten on Arrow, DataFusion and Parquet.