TimescaleDB

Time-series Available Free tier Point-in-time recovery

PostgreSQL, tuned for time

TimescaleDB adds hypertables and time functions to PostgreSQL. Keep full SQL, joins and the Postgres ecosystem while partitioning time-series data automatically. We run the Apache-2.0 edition.

Metrics alongside relational data
IoT
Financial ticks

Create TimescaleDB Pricing Docs

psql · ts-7f3k.us-east.databasezy.com

Sample

app=> SELECT time_bucket('5 minutes', time) AS bucket,
             device_id, avg(temperature) AS avg_temp
      FROM readings
      WHERE time > now() - interval '15 minutes'
      GROUP BY bucket, device_id
      ORDER BY bucket;

Sample result of the TimescaleDB query above
bucket	device_id	avg_temp
2026-10-04 09:30:00+00	dev-17	21.4
2026-10-04 09:35:00+00	dev-17	21.9
2026-10-04 09:40:00+00	dev-17	22.3

3 rows · 5.5 ms Illustrative data and timing

Features

TimescaleDB on Databasezy

Hypertables

Tables partition themselves by time automatically as data arrives.
time_bucket and friends

Bucket and aggregate time-series with time_bucket and ordinary SQL.
It is PostgreSQL

Joins, indexes, drivers and ORMs work exactly as they do on Postgres.
Point-in-time recovery

The same barman-cloud base backups and WAL archiving as our PostgreSQL engine.

Quick start

Connect to TimescaleDB

Hosts, ports and the CA bundle come from the portal or the CLI after you create an instance. Credentials are shown once and never stored by our control plane.

From the CLI

1. Create the instance

zb instances create --engine timescaledb --name timescaledb-7f3k --wait

2. Download the CA bundle

zb ca > databasezy-ca.pem

3. Open a shell with a one-time credential

zb connect timescaledb-7f3k

At a glance

Category: Time-series
Versions: 18, 17, 16
Wire protocol: postgres · port 5432
Runtime: Operator: cloudnative-pg
Backup: barman-cloud base backups + continuous WAL archiving to object storage · point-in-time recovery
Pause: Scale to zero
Free tier: Yes
Licence: Apache-2.0 (Apache edition)

Connection strings and drivers

Connection URI

postgresql://app:<password>@ts-7f3k.us-east.databasezy.com:5432/app?sslmode=verify-full&sslrootcert=databasezy-ca.pem

psql

psql "host=ts-7f3k.us-east.databasezy.com port=5432 dbname=app user=app sslmode=verify-full sslrootcert=databasezy-ca.pem"

Node.js (pg)

import pg from "pg";
import { readFileSync } from "node:fs";

const pool = new pg.Pool({
  connectionString: process.env.DATABASE_URL,
  ssl: { ca: readFileSync("databasezy-ca.pem", "utf8") },
});
const { rows } = await pool.query("select now()");

Python (psycopg)

import os, psycopg

with psycopg.connect(os.environ["DATABASE_URL"]) as conn:
    print(conn.execute("select now()").fetchone())

Deployment

Choose where it runs

Placement is chosen per database, so a regulated workload and a side project can live in the same organisation.

Standard hosting

Shared cells with namespace isolation, default-deny network policy and an IP allow-list. Available from the free plan up.
- Usage-based compute, billed per running hour
- Spend cap to stop surprises
Secure hosting (HIPAA)

This database runs on dedicated nodes in a separate HIPAA cell, under a signed Business Associate Agreement. Available on Team and Enterprise.
- $0.10 per instance-hour on top of compute
- How secure hosting works
From the CLI or API

Script instances in CI with the zb CLI or the REST API. Pick the placement per instance.
secure placement
```
zb instances create --engine timescaledb --placement secure --wait
```

Plans

TimescaleDB on each plan

Usage-based pricing with a spend cap: you pay a plan fee plus what your instances use.

Free tier

Yes. One f0 instance with a shared vCPU (bursts to 1), 512 MiB of memory and a 500 MB database, no card required. It sleeps after 15 minutes idle and wakes on connect.

Sizes

From s0 (0.25 vCPU, 1 GiB) at $10/mo up to xl32 (32 vCPU, 128 GiB). Resize without changing the connection string.

All sizes and prices

Backups

barman-cloud base backups + continuous WAL archiving to object storage. Point-in-time recovery is an add-on on Solo and Team and Enterprise ($100 per instance per month for each 7 days of retention) and included on Enterprise.

Migrate

Migrate to TimescaleDB from

Copy a database in with a connection string, a file upload or the CLI. Preflight checks the source first and verification compares the copy.

Documentation

TimescaleDB docs

TimescaleDB questions

Which TimescaleDB versions can I run?

18, 17, 16. New instances default to 18; pick another with --engine-version or in the portal. Minor and patch updates are applied in your maintenance window after a pre-change backup.

Which extensions, modules or plugins are available?

The timescaledb extension (Apache-2.0 edition) is preinstalled and created in the app database: hypertables, time_bucket and chunk management are available. Timescale License features are not: native compression and the columnstore, continuous aggregates, the job scheduler and its policies, time_bucket_gapfill and the Toolkit hyperfunctions. Run SELECT * FROM pg_available_extensions to see which other extensions the image ships.

How are backups handled?

Backups use barman-cloud base backups + continuous WAL archiving to object storage. Paid plans run scheduled backups (Solo: daily, 7-day retention; Team: daily, 14-day retention); the free plan keeps one manual snapshot. Point-in-time recovery is available as an add-on on Solo and Team and Enterprise and included on Enterprise. Restores go into a new instance, so the original stays untouched.

Is the connection encrypted?

Always. The gateway terminates TLS (1.3 where your client supports it, 1.2 at minimum) and re-encrypts traffic to the database. Plain-text connections are refused. Verify the server with our CA bundle from the portal or zb ca, and restrict access further with the IP allow-list included on every plan.

Can I pause it when I'm not using it?

Yes. Free instances sleep after 15 minutes without connections and wake on the next one; after 7 idle days they are archived with the disk kept. On paid plans you pause on demand or on a schedule. A paused instance bills storage and backups but no compute, and wakes in 10 to 40 seconds.

What are the limits?

The free plan runs one f0 instance with a shared vCPU (bursts to 1), 512 MiB of memory, 20 connections and a 500 MB database. Paid sizes run from s0 (0.25 vCPU, 1 GiB, 60 connections, $10/mo) to xl32 (32 vCPU, 128 GiB, 5,000 connections), with up to 16 TB of disk.

Can I use compression or continuous aggregates?

No. They are Timescale License features, which cannot be offered as a hosted service, and Databasezy runs the Apache-2.0 edition. Use regular materialized views refreshed by your own scheduler for rollups, and drop_chunks for retention.

What you get on each plan

Plan features as they apply to TimescaleDB.

TimescaleDB features by plan
Feature	Free	Solo	Team	Enterprise
TimescaleDB on this plan	Yes	Yes	Yes	Yes
Plan fee	$0/mo	$25/mo	$599/mo	Custom
Largest size	f0	xl32	xl32	xl32
Included per instance	500 MB	8 GB	8 GB	Negotiated
Max disk per instance	1 GB	1 TB	16 TB	16 TB
Scheduled backups	1 manual snapshot	Daily, 7-day retention	Daily, 14-day retention	Hourly, 35-day retention · cross-region copy · immutable
Point-in-time recovery	—	Add-on	Add-on	35 days included
Read replicas	0	2	5	Unlimited
Pause	Sleeps after 15 min idle, wakes on connect	On demand or on a schedule	On demand or on a schedule	On demand or on a schedule
Secure hosting (HIPAA)	—	—	Available with a BAA	Available with a BAA
Support	Community	Email	Priority	24x7 dedicated
Uptime SLA	—	—	99.9%	99.95%

Start with TimescaleDB today

The free plan needs no card. Upgrade when you need more, and keep the same endpoint.

Create TimescaleDB See pricing

TimescaleDB

Good for

Sample query

TimescaleDB on Databasezy

Hypertables

time_bucket and friends

It is PostgreSQL

Point-in-time recovery