Tiger Data

Editor’s note: This Tiger Data review is based on my hands-on test of Tiger Cloud and TimescaleDB 2.31.0 in October 2026. It weighs pricing, time-series features, forks, CLI and MCP, licensing, and rivals InfluxDB, QuestDB and ClickHouse.

Quick verdict: I recommend Tiger Data for PostgreSQL teams storing sensor, metrics, event or tick data without adopting a second database. The trade-off: columnstore compression, continuous aggregates and retention policies sit under the source-available Timescale License, so outside Tiger Cloud you self-host the Community Edition or accept the Apache build.

Timescale became Tiger Data on June 17, 2025, and TimescaleDB kept its name. In this Tiger Data review, I’ll take a closer look at Tiger Data’s pricing, features, and licensing, so you can see exactly whether time-series Postgres beats a dedicated time-series database for you.

Key Takeaways:

• Performance compute starts at $30/month and Scale at $36/month, both billed hourly.
• Hypertables are plain SQL: CREATE TABLE ... WITH (tsdb.hypertable).
• Tiger Cloud charges no ingest, egress or backup fees.
• Read replicas, tiered storage and the SOC 2 report need Scale; HIPAA and SAML SSO need Enterprise.
• Azure’s managed Postgres ships only the Apache build, without columnstore compression or continuous aggregates.

Pros and Cons

How Much Does Tiger Data Cost?

Self-hosted TimescaleDB is free, while Tiger Cloud sells three plans billed hourly, plus two free services per account in beta. The monthly “from” prices cover compute only: storage, replicas and add-ons bill on top.

• Performance (from $30/month): for new apps and internal tools, with 4 services, up to 8 CPU / 32 GB RAM and 16 TB per service, 3 days of point-in-time recovery and Basic support from 9 to 5 ET.
• Scale (from $36/month): for production workloads, with unlimited services, up to 32 CPU / 128 GB and 64 TB, read replicas, multi-node HA, 14-day recovery, the SOC 2 report, unlimited tiered storage at $0.021/GB-month and I/O Boost up to 40K IOPS.
• Enterprise (custom quote): for regulated workloads, with up to 64 CPU / 256 GB and 80K IOPS, HIPAA, SAML SSO, cross-region backups, recovery up to 180 days and 24×7 support with a 1-hour Sev 1 response.
• Free services ($0, beta): for experiments, with two services per account in us-east-1, no pooling, replication, tiering, metrics or VPC, and read-only mode once a service hits its storage limit.

Plan	Compute from	Storage, effective at 5x compression	Max per service	PITR / forks	Key gates
Performance	$30/month	$0.177/GB-month	8 CPU / 32 GB / 16 TB	3 days	4 services; no read replicas
Scale	$36/month	$0.212/GB-month	32 CPU / 128 GB / 64 TB	14 days	Read replicas, tiered storage, SOC 2
Enterprise	Custom	Custom	64 CPU / 256 GB / 64 TB	Up to 180 days	HIPAA, SAML SSO

I priced a Scale service at 2 CPU / 8 GB in us-east-1: $0.484 an hour on the time-series rate, or about $353 a month across 730 hours. Adding 500 GB at the effective $0.212 rate brings it to roughly $459 a month before add-ons, and that storage figure only holds if the data compresses 5x.

Region moves the bill: the 1 CPU / 4 GB Scale size costs $0.242/hour in us-east-1, $0.3424 in eu-west-1 and $0.492 in sa-east-1. HA replicas and read replicas bill at the primary’s compute and storage rates, so each one adds a full copy of the compute line.

Tiger Cloud charges nothing for backups, ingest, egress or per-query reads, whereas InfluxDB Cloud Serverless bills $0.09 per GB of egress. I/O Boost costs $0.41 per hour per 16K IOPS on Scale and Enterprise.

Idle services keep billing until you pause or delete them. The 30-day Performance trial needs no card and pauses any service with no queries for 7 days; after it, AWS Marketplace offers pay-as-you-go, annual-commit and private-offer billing.

Is Tiger Data Good Value for Money?

• Good fit: Postgres teams facing a second time-series database next to the app database, since Tiger Cloud keeps both workloads in one service with no egress fees.
• Less convincing fit: data that compresses well below 5x, because the $0.177 and $0.212 storage rates assume that ratio.
• Budget priority: regional compute and replicas, which move a production bill far more than the $30 entry price.

Reviewer’s Notes: Start on the 30-day Performance trial, size compute from your region’s hourly time-series rates, and move to Scale the day you need read replicas, tiered storage or a SOC 2 report.

My Experience With Tiger Data

Everything I built in Tiger Cloud was plain PostgreSQL SQL: the hypertable, the hourly rollup and the retention rule are all database objects, not a separate pipeline or query language.

Signing Up and Creating a Service

I started the 30-day Performance trial without a card. The Console’s Create service flow runs in three steps with a live price estimate, so I picked the region before the compute size, since region changes the hourly rate. Trial services with no queries for 7 days are paused with their data intact and resume from the Console.

New services run PostgreSQL 18 by default, on AWS or Azure. I connected as tsdbadmin with psql, and any other Postgres client, GUI or ORM, connects the same way.

For a self-hosted equivalent, the timescale/timescaledb-ha:pg18 Docker image (or curl -sL https://tsdb.co/start-local | sh) gives the same PostgreSQL 18 starting point, minus the managed backups and upgrades.

Creating My First Hypertable

I created a sensor table with sensor_id as the segment key and the newest readings first:

CREATE TABLE sensor_data (
  time TIMESTAMPTZ NOT NULL,
  sensor_id TEXT NOT NULL,
  temperature DOUBLE PRECISION
) WITH (
  tsdb.hypertable,
  tsdb.segmentby = 'sensor_id',
  tsdb.orderby = 'time DESC'
);

From TimescaleDB 2.23, this form picks the first TIMESTAMPTZ column as the partition column and creates a columnstore policy automatically, converting chunks once they are one chunk interval old. Self-hosters on 2.20.0 to 2.22.1 get the same syntax but must call add_columnstore_policy themselves. I set segmentby and orderby explicitly because those two columns later decide how cheaply updates and deletes run against compressed data.

The Console also offers a no-SQL route: Actions > Goals > Create a hypertable > Get started, which shows the generated SQL before you click Run and proceed.

Adding a Continuous Aggregate and Retention

Next I added hourly averages per sensor on top of that table, refreshed every hour, and a 6-month retention rule on the raw table:

CREATE MATERIALIZED VIEW sensor_hourly
WITH (timescaledb.continuous) AS
SELECT
  time_bucket('1 hour', time) AS bucket,
  sensor_id,
  avg(temperature)
FROM sensor_data
GROUP BY bucket, sensor_id;

SELECT add_continuous_aggregate_policy(
  'sensor_hourly',
  start_offset => INTERVAL '1 month',
  end_offset => INTERVAL '1 hour',
  schedule_interval => INTERVAL '1 hour'
);

SELECT add_retention_policy(
  'sensor_data',
  drop_after => INTERVAL '6 months'
);

The refresh policy needs start_offset larger than end_offset: here it re-materializes the last month and skips the newest hour, while retention drops raw chunks older than 6 months on its own schedule. Both policies show up as scheduled jobs in timescaledb_information.jobs, which is where I checked them.

For querying, the Console’s Data view bundles SQL notebooks, charts, dashboards and a SQL assistant, and the Service Explorer runs queries too.

Reviewer’s Notes: Because every step is standard SQL, psql, Postgres ORMs and BI connectors connect to Tiger Cloud exactly as they do to any Postgres database, with no new query language to learn, unlike InfluxDB’s history of switching between InfluxQL and Flux.

Hypertables, Columnstore and Continuous Aggregates

TimescaleDB handles append-heavy time-series data well; frequent corrections to old, compressed data are where it still costs you, even after the 2.31.0 improvements.

Hypertables and Chunks

Hypertables split data into time-based chunks, or by UUIDv7 since 2.22. Tiger Cloud’s automated chunk tuning, generally available since June 26, 2026, adjusts the interval for new chunks only.

Planning time grows with chunk count: Tiger Data’s own engineers measured about 600 ms of planning at 4,000 chunks before a fix cut it to 36 ms. Version 2.30’s DeferredChunkAppend defers chunk expansion to execution, so last-point LIMIT queries answered from the newest chunk no longer scale with chunk count.

Columnstore Compression

The columnstore organizes data by segmentby and orderby, with bloom filters since 2.20. Version 2.31 adds move_to_columnstore(), which converts rows without blocking concurrent DML, plus vectorized processing for updates and deletes. It also previews an Adaptive Integer Compressor, off by default, which Tiger Data’s release notes put at up to 20% better integer compression.

Corrections were the long-running pain: users in GitHub issue #2857 described backfills into one compressed chunk taking up to 2 hours. Tiger Data’s release notes claim up to 42x faster deletes in 2.21 and bloom-filter DML gains in 2.27, but updates and deletes that don’t filter on the segmentby or orderby columns can still rewrite large ranges through the rowstore.

Put correction-heavy traffic in your trial, not just inserts, and test the 5x compression assumption behind the storage rates against your own columns.

Continuous Aggregates

Continuous aggregates provide incremental rollups, scheduled refreshes and real-time aggregation, so hourly and daily summaries stop being recomputed from raw rows. Late data is the catch: users in GitHub issue #8132 describe one device sending old data invalidating months of an aggregate. Version 2.31 adds granular refresh APIs and refresh statistics, which narrow that problem without closing it.

Tiered Storage

Scale and Enterprise tier old chunks to object storage at $0.021/GB-month, with no transfer or compute charges. Tiering bills the original uncompressed size, so 1 TB of raw data costs about $21 a month in the tier whatever it compresses to; ClickHouse Cloud Enterprise storage, for comparison, is $25.30 per TB-month in us-east-1.

Forks, Tiger CLI and the MCP Server

Tiger Cloud’s copy-on-write forks clone a service at a point in time with independent data and configuration plus a new tsdbadmin password. Built on the Fluid Storage layer, a fork bills storage only for changed blocks, but it is still a running service, so its compute counts like any other.

How far back a fork can reach follows the plan:

• Performance: 3 days.
• Scale: 14 days.
• Enterprise: up to 180 days.

That makes a fork the safest place for a coding agent to test an index or migration: fork, test, delete, and production never sees the agent’s connection.

Tiger CLI drives the same workflow from the terminal:

curl -fsSL https://cli.tigerdata.com | sh
tiger auth login
tiger create service
tiger service fork <name-or-id>
tiger mcp install

Recent CLI releases add guardrails worth switching on before any agent gets access:

• 0.23 (September 1, 2026): protects production services from writes.
• 0.24: adds a database query command and device-code login.
• 0.26 (October 1, 2026): adds service metrics.

tiger mcp install configures Claude Code, Cursor, Windsurf, Codex, Gemini CLI, VS Code and Kiro CLI. The CLI, MCP server and forks launched together as Agentic Postgres on October 21, 2025, and they are the strongest reason to pay for Tiger Cloud over self-hosting.

Tiger Lake is still in beta: it replicates Scale and Enterprise services to Iceberg or Amazon S3 Tables, so I treat it as a separate decision from the database itself. It gained a generic Iceberg REST catalog in August 2026 and tiered-data support on October 7.

Is TimescaleDB Open Source? The License Split Explained

Only TimescaleDB’s basic hypertable layer is Apache 2; the features most buyers want ship in the free Community Edition under the source-available Timescale License.

Capability	Apache 2 edition	Community Edition, Timescale License
Hypertables and chunks	Yes	Yes
time_bucket, first/last, histogram	Yes	Yes
Informational views	Yes	Yes
Columnstore / hypercore	No	Yes
Continuous aggregates and policies	No	Yes
Retention policies and background jobs	No	Yes
SkipScan and gap-filling functions	No	Yes
Advanced hyperfunctions	No	Yes

The Community side also holds time_bucket_gapfill, locf, interpolate, percentile_agg, uddsketch and tdigest, plus the reorder, move and split chunk functions.

The Timescale License lets you self-host for free but is not OSI-approved and bars offering TimescaleDB as a hosted database service. Self-hosting gives you every Community feature, with upgrades, high availability and backups as your own job, and Tiger Data names Tiger Cloud as the only managed host for the Community Edition.

Microsoft has said Azure Database for PostgreSQL Flexible Server ships only the Apache build, without the Community Edition or the Toolkit, so check Microsoft’s current extension list before you commit. Even the Apache build needs timescaledb in both shared_preload_libraries and azure.extensions there. On a managed Postgres that lists “TimescaleDB”, expect hypertables without compression or continuous aggregates.

My sensor walkthrough depends on add_retention_policy and a continuous aggregate, both Community features, so it does not run on the Apache build, including Azure’s.

The repository has 23,737 GitHub stars, but stars don’t settle license fit: pick the Apache build only when hypertables and time_bucket cover the workload and nothing on the Community list above is needed.

Updates, Upgrades and Support

TimescaleDB shipped seven minor releases in 2026, from 2.25 on January 29 to 2.31 on October 20, roughly one every 5 to 7 weeks. Current releases support PostgreSQL 16, 17 and 18; 2.29 dropped PostgreSQL 15, and 2.29.1 fixed three security vulnerabilities, so pinning an old version indefinitely is a poor default.

Self-hosters should test these areas before upgrading:

• Continuous aggregates: the 2.22.0 notes told continuous-aggregate users to wait for 2.22.1.
• Compressed-data queries: 2.29.2 fixed SkipScan dropping uncompressed rows from COUNT(DISTINCT) results.
• Configuration: 2.31.0 renames timescaledb.enable_columnarscan to debug_enable_columnarscan and changes the required extension order in shared_preload_libraries.

Tiger Cloud applies upgrades for you, though custom settings and queries still deserve a check after each one. Migrations changed too: Livesync, built on PostgreSQL logical replication, replaced the live-migration Docker image on July 24, 2026.

Support follows the plan:

• Developer: email with a 1-business-day target, free on paid plans.
• Basic: 9 to 5 ET coverage on Performance.
• Production: 24×7, an add-on on Performance and Scale and included on Enterprise.

Budget Production support separately if an outage can’t wait for business hours.

How Does Tiger Data Compare to Competitors?

Tiger Data wins when PostgreSQL compatibility and relational joins matter more than a specialist engine; each rival below wins a narrower niche.

• InfluxDB 3: wins for metrics pipelines already built on Telegraf and InfluxDB. Core is free, open source and single-node; Cloud Serverless bills $0.0025 per MB written, $0.012 per 100 queries, $0.002 per GB-hour of storage and $0.09/GB egress, its free tier keeps 30 days of data, and Enterprise is quote-only. It loses once time-series queries need to join application data that already lives in Postgres.
• QuestDB: wins on raw ingest for market data. The open-source edition is a single instance; Enterprise adds replication, failover under 100 ms, PITR, Parquet/Iceberg tiering and role-based access control, at unpublished prices, and BYOC is listed as coming soon. Budget for Enterprise if you need high availability.
• ClickHouse Cloud: wins on large analytical scans across billions of rows. The trial includes $300 credit over 30 days with no monthly minimum after it, and Enterprise in us-east-1 charges $0.39030 per compute unit-hour (8 GiB / 2 vCPU) plus $25.30 per TB-month of storage. It is not Postgres, so moving an app over means rewriting queries, not swapping a connection string.
• Apache Druid: wins for real-time OLAP dashboards that need sub-second slice-and-dice on streaming data. Tiger Data stays ahead when the same database also has to serve the application.
• Azure Database for PostgreSQL: wins only when staying inside managed Azure Postgres outweighs compression and continuous aggregates, since its Apache-only TimescaleDB stops at hypertables. Tiger Cloud runs on Azure too, so choosing Azure doesn’t force the smaller feature set.

Tool	Model	Starts at	Query language	Best for
Tiger Cloud	Managed PostgreSQL + TimescaleDB	$30/month compute	PostgreSQL SQL	Time-series joined with relational data
InfluxDB 3 Cloud Serverless	Usage-priced time-series service	Free tier; paid writes $0.0025/MB	SQL / InfluxQL	Existing InfluxDB metrics pipelines
QuestDB	Open-source single instance; Enterprise	Free self-hosted; Enterprise quote	SQL	Market-data ingestion
ClickHouse Cloud	Managed columnar OLAP	Usage-based; $300 trial credit	ClickHouse SQL	Large analytical scans

Pick by query shape first, then price: InfluxDB’s usage meter, QuestDB’s Enterprise gates and Tiger Cloud’s regional hourly rates produce very different bills for the same ingest volume and retention window.

How I Reviewed Tiger Data

I tested Tiger Cloud on the 30-day Performance trial: service creation, connecting with psql, and the hypertable, columnstore, continuous aggregate and retention workflow in SQL on TimescaleDB 2.31.0. I also worked through the Console’s Data view, created a fork to check its new tsdbadmin credentials, and drove service creation from the Tiger CLI.

Beyond hands-on use, I compared hourly compute by region and plan gates, the Apache and Timescale License split, the 2026 release history, and InfluxDB, QuestDB and ClickHouse pricing, using us-east-1, eu-west-1 and sa-east-1 as reference regions. I ran no benchmarks of my own, so every speed figure here is Tiger Data’s own claim.

Prices current as of October 2026.

Should You Use Tiger Data for Time-Series Data?

Yes, for PostgreSQL teams storing metrics, IoT, event or financial tick data. Tiger Cloud is where the Community features, forks and egress-free billing come together without you running the database.

Start on Performance for evaluation and move to Scale when read replicas, tiered storage or the SOC 2 report become requirements. Budget from the regional hourly rate, not the $30 headline: my Scale example lands near $459 a month before replicas.

Look elsewhere if you are Google Cloud-only (Tiger Cloud runs on AWS and Azure), need massive ad-hoc OLAP scans (ClickHouse fits better), or require an OSI-approved license for every feature you run. If your legal team accepts the Timescale License but not a managed bill, self-hosting the Community Edition is the free fallback, with upgrades, HA and backups on you.

Your next step: start the 30-day Tiger Cloud trial with a representative table that includes historical corrections and late-arriving rows, not just clean appends.

FAQs

Is Timescale the same as Tiger Data?

Yes. Timescale renamed itself Tiger Data on June 17, 2025. Tiger Cloud is the managed platform, while TimescaleDB keeps its name as the PostgreSQL extension. Buying Tiger Cloud is a different decision from installing the free, self-hosted TimescaleDB extension on your own server.

How much does Tiger Cloud cost?

Compute starts at $30/month on Performance and $36/month on Scale, billed hourly; Enterprise is a custom quote. Storage costs an effective $0.177 or $0.212 per GB-month, assuming 5x compression, so budget compute and storage separately. The 30-day Performance trial needs no credit card.

Is TimescaleDB free?

Yes, TimescaleDB is free to self-host, but its two editions carry different licenses. Apache 2 covers basic hypertables and functions; the Timescale License covers Community features such as columnstore compression and continuous aggregates. Tiger Cloud also offers two free services per account in beta, with restricted features and a storage cap.

Which PostgreSQL versions does TimescaleDB support?

TimescaleDB 2.31.0 supports PostgreSQL 16, 17 and 18. Support for PostgreSQL 15 ended with TimescaleDB 2.29.0, and new Tiger Cloud services default to PostgreSQL 18. A Tiger Cloud service still on PostgreSQL 15 must move to 16 or later before it can adopt newer TimescaleDB releases.

Can I use TimescaleDB on Azure or AWS managed Postgres?

Azure Database for PostgreSQL Flexible Server offers only the Apache build, without compression or continuous aggregates. For the full feature set, use Tiger Cloud, which runs on AWS and Azure and bills through AWS Marketplace. For any other provider, check its current extension list and exactly which TimescaleDB edition it ships.