TigerData

What TigerData actually is

TigerData is the company formerly known as Timescale Inc. — the team that built TimescaleDB, the open-source PostgreSQL extension for time-series and event data. The rebrand landed on 17 June 2025 and the naming now splits cleanly: TimescaleDB stays the open-source extension, Tiger Cloud is the managed cloud platform, and Tiger Postgres is their customized PostgreSQL build with newer pieces like Tiger Lake (an Iceberg/lakehouse sync layer, introduced July 2025). The company is New York-based, Series C funded, and reports roughly 2,000 customers including Mistral, Nvidia, Toyota, NASA and JP Morgan Chase.

Under the branding churn, the technical proposition is stable and specific. TimescaleDB turns PostgreSQL tables into hypertables — automatic time-based partitioning into chunks, with optional space partitioning — and adds continuous aggregates (incrementally maintained materialized views), columnar compression that routinely reaches 90%+ on time-series data, and hundreds of native time-series functions. Everything stays SQL. You get the full PostgreSQL extension ecosystem: PostGIS, pgvector/pgvectorscale, logical replication, the drivers and ORMs you already use. That is the core argument for picking it over a proprietary time-series engine: no new query language, no data island.

The managed service, Tiger Cloud, runs on AWS and Azure with disaggregated compute and storage: compute metered hourly and resizable up to 64 CPU / 256 GB per service, high-performance storage up to 64 TB compressed per service, and bottomless tiered storage on S3 for older data. Read replicas, multi-AZ HA, connection pooling and point-in-time recovery are productized features, not runbooks. Self-hosting remains fully supported — the extension is open source (the community edition), and that escape hatch matters when you evaluate long-term cost.

Pricing: what you really pay

Tiger Cloud pricing is consumption-based and published. The Performance plan starts at $30 per month for compute with storage at $0.177 per GB-month (their stated effective price assuming 5x compression); it includes up to 4 database services, up to 8 CPU and 32 GB memory per service, up to 16 TB disk per service, single-node HA replicas, 3-day point-in-time recovery and 9-5 ET support. The Scale plan starts at $36 per month with storage at $0.212 per GB-month, unlimited services, up to 32 CPU / 128 GB memory, up to 64 TB disk, 40K IOPS / 1,500 MB/s class storage, unlimited tiered storage on S3, read replicas, multi-node HA, 14-day PITR, SOC 2 report and private networking (AWS Transit Gateway, private endpoints). Enterprise is custom-priced with 24/7 support and contractual SLAs.

Billing mechanics matter more than the sticker here. Compute is metered per hour — you are billed for a full hour even if usage is less, and a mid-hour config change bills at the config in force at the end of that hour. Storage is metered on average GB consumed per quarter-hour and grows/shrinks with your data. Add-ons (tiered storage, HA replicas, connection pooling) incur additional charges. Compute prices vary by region (AWS infrastructure cost pass-through); storage pricing is uniform across regions. Their docs also describe a free tier — up to 2 free services with shared CPU and a 750 MB storage cap — while the marketing pages advertise a 30-day free trial with a $1,000 credit; both are real onboarding paths, and we list the trial as the headline because that is what the pricing page sells today.

Two honest observations. First, the pricing page's "effective storage price based on 5x compression" framing is a derived number — compression ratios are workload-dependent (they also quote up to 98% in docs), so treat $0.177/GB-month as best case, not default. Second, the self-hosted option is genuine price pressure: if you have the ops capacity, TimescaleDB on your own VMs costs only the VMs. We would price both before signing.

Performance & infrastructure

The infrastructure claims are concrete: 110,000+ IOPS single-volume read throughput on their "Fluid Storage" with 1.4 GB/s bandwidth and synchronous replication on every write; per-plan IOPS floors of 5K (Performance) and 40K (Scale). Multi-AZ deployments, continuous incremental backup, cross-region backup and restore, configurable backup retention up to 180 days, and regular database and disk snapshots for fast restore are listed as included platform features on both AWS and Azure.

Region coverage is broad for a managed database: 13 AWS regions (Frankfurt, Zurich, Ireland, London, N. Virginia, Ohio, Oregon, Mumbai, Singapore, Sydney, Tokyo, Central Canada, São Paulo) plus 4 Azure regions (Virginia, Amsterdam, Frankfurt, Singapore). EU, US and APAC data residency are all selectable at service creation, and their enterprise pitch includes regional data isolation. Terraform provider, REST API, CLI and an MCP interface exist for automation; metric export goes to CloudWatch, Prometheus and Datadog on higher plans.

We have not benchmarked Tiger Cloud ourselves — it is not in our 2026-04-01 VM test set, and cross-engine database comparisons are not what that run measured. What we can point at: the vendor publishes a public SLA (99.9% monthly uptime for HA-replicated services on Enterprise, 99.8% for non-replicated) with a real service-credit table, and a public status page with per-region history. Third-party RTA benchmark claims exist on their site; we treat those as vendor figures. The Postgres-native design also means you can replicate their benchmark methodology yourself on a $30 service inside an hour — a transparency point we value.

Where it falls short

The hourly compute meter is unforgiving at the edges: a full hour is billed even for partial use, and a burst scale-up that you forget to scale back down pays full price at the higher config. Consumption billing is fair when you watch it and expensive when you do not; budget alarms are on you.

Second, the feature gates are real. On the Performance plan you get 3-day PITR, single-node HA, 1 VPC, 1 IP allow-list, and 9-5 ET support only; the serious reliability and security features (multi-node HA, 14-day PITR, SOC 2 report, private endpoints, unlimited services) sit behind the Scale plan or Enterprise. If your compliance regime needs the SOC 2 report, the $30 headline is not your price.

Third, this is still a specialist tool. Continuous aggregates, chunk intervals, compression policies and retention policies are not hard, but they are real concepts your team must learn, and PostgreSQL expertise is assumed. Their own cons list — and ours — include that point: a zero-ops experience it is not, and self-hosted deployments carry full operational load. The company also carries the usual single-vendor risk of a Series C database business, mitigated substantially by the open-source extension being portable out.

Finally, one legacy wrinkle: an Aiven-partnered "Managed Service for TimescaleDB" exists with a 99.99% availability commitment, separate from Tiger Cloud. If a reseller quotes you that, know it is a different product with a different SLA.

Who should pick this

Our pick for teams running observability platforms, IoT/industrial telemetry, financial tick or event data, and real-time analytics who want SQL and the PostgreSQL ecosystem rather than a proprietary query language. The compression economics are the strongest argument at scale — 90%+ storage reduction on typical time-series data changes the cost curve versus plain Postgres or a document store, and continuous aggregates keep dashboard queries off the raw data.

We would also pick it over a bespoke metrics stack (custom partitioned Postgres plus cron-maintained rollups) for most teams below the scale where that engineering pays for itself. We would look elsewhere — plain managed PostgreSQL, or self-hosted TimescaleDB on cheap VMs from /category/database-hosting — for small steady-state datasets where $30/month of metered compute is a bad trade, and at dedicated vector or search engines for workloads that are not time-shaped at all. Compare VM economics on /benchmarks if you go the self-hosted route.

Key takeaways

• TimescaleDB (open source, since 2017) plus Tiger Cloud (managed, AWS + Azure) from the New York company rebranded from Timescale to TigerData on 17 June 2025.
• Hypertables, continuous aggregates and 90%+ compression inside standard PostgreSQL — no new query language, full extension ecosystem.
• Consumption pricing: compute from $30/month (Performance) or $36/month (Scale), storage $0.177-0.212/GB-month effective at 5x compression, hourly compute metering, full-hour billing.
• 13 AWS + 4 Azure regions including Frankfurt, Ireland, Amsterdam and London — EU, US and APAC residency all selectable.
• Real reliability productization: multi-AZ HA, PITR up to 14 days (Scale/Enterprise), backup retention to 180 days, public SLA of 99.9% for HA services (Enterprise) and a public status page.
• Feature gates (SOC 2, private endpoints, 24/7 support) require Scale or Enterprise; self-hosted TimescaleDB keeps you price-flexible.

Sources

• https://www.tigerdata.com/pricing
• https://docs.timescale.com/about/latest/pricing-and-account-management
• https://www.tigerdata.com/docs/deploy/tiger-cloud/tiger-cloud-aws/pricing-and-account-management
• https://www.tigerdata.com/docs/get-started/supported-regions
• https://www.tigerdata.com/docs/get-started/feature-comparison/index.md
• https://tigerdata.com/legal/service-level-agreement
• https://tigerdata.com/cloud
• https://tigerdata.com/blog/timescale-becomes-tigerdata
• https://tigerdata.com/newsroom/tiger-data-unveils-tiger-lake-to-bridge-the-gap-between-postgres-and-lakehouse
• https://grokipedia.com/page/TimescaleDB
• https://grokipedia.com/page/TigerData
• https://status.timescale.com/