Back to blog
6 min read
Mar 16, 2026
Table of contents
01 Performance and scalability02 Platform reach and enterprise access03 Developer experience04 Infrastructure as Code05 Try Tiger Cloud for freeTL;DR: Over the last few months, we shipped a lot. TimescaleDB 2.25 brings up to 289x faster queries on compressed data. Postgres 18 is now the default. Full-text search got dramatically faster and leaner. Azure customers get Tiered Storage and self-serve Marketplace signup. Tiger Console has a new floating SQL editor, a jobs timeline view, and built-in CLI/MCP setup. And Terraform now manages S3 source connectors. Here is everything, and why it matters.
It has been a busy start to 2026 at Tiger Data. We have shipped improvements across every layer of Tiger Cloud: from deep query engine optimizations in TimescaleDB to new cloud regions, Azure feature parity, and developer experience upgrades in Tiger Console.
Rather than cover each release in isolation, this post groups everything into four themes so you can see the bigger picture: Tiger Cloud is getting faster at the core, reaching more customers where they are, making day-to-day work smoother, and becoming easier to manage programmatically.
TimescaleDB 2.25 landed on Tiger Cloud on March 5, and it is a significant release for anyone running analytical queries on compressed data.
Fast queries on compressed data, without the tradeoffs: The headline feature is a new ColumnarIndexScan execution path that uses columnar metadata to accelerate common aggregate functions. Instead of scanning individual rows or decompressing batches, queries using MIN, MAX, FIRST, or LAST on compressed data can now resolve directly from chunk-level summaries. In internal testing, this class of queries ran up to 289x faster. Similarly, COUNT(*) with time filters can skip reading the time column entirely, with speedups of up to 50x in tested workloads. The feature is currently opt-in via the enable_columnarindexscan GUC and will be on by default in 2.26.
Continuous Aggregates (i.e. incremental materialized views in TimescaleDB) also got more efficient. Direct compress on refresh (experimental feature, opt-in via GUC timescaledb.enable_direct_compress_on_cagg_refresh) now writes results straight to the columnstore, skipping the intermediate rowstore step that could cause policy contention and extra WAL activity.
The "direct batch delete" optimization (which deletes whole batches on the columnstore instead of row-by-row) was previously disabled for tables with Continuous Aggregates because it didn't emit the necessary invalidation logs. This is now fully supported, meaning columnstore hypertables with Continuous Aggregates will benefit from much faster, lower-I/O deletes, keeping refresh transactions smaller and steadier.
On the operational side, 2.25 adds the ability to estimate original size for columnstore chunks (useful for billing and Console reporting), an option to auto-add new chunks to a publication for logical replication, and configurable work_mem for background worker jobs. Example:
SELECT _timescaledb_functions.estimate_uncompressed_size('<schema>.<chunk_name>');For the full architectural context on why these changes matter, read the TimescaleDB 2.25 release notes and our related blog post, Start on Postgres, Scale on Postgres.
Tiger Cloud now runs Postgres 18 by default for all new services. Existing services can upgrade over the coming weeks.
Postgres 18 is a meaningful release. Asynchronous I/O with io_uring on Linux significantly speeds up read-heavy workloads. Major upgrades are faster and less disruptive thanks to improved pg_upgrade and the ability to preserve planner statistics across versions. Virtual generated columns are now the default, and the new uuidv7() function improves UUID indexing behavior. Query performance gets a boost from expanded index usage, including skip-scan on multicolumn B-tree indexes, and parallel GIN index builds. On the security side, OAuth 2.0 authentication and page checksums enabled by default round out the release.
For more, see the official Postgres 18 announcement.
pg_textsearch has seen three releases on Tiger Cloud as it moves toward GA, and the cumulative improvements are substantial.
v0.3.0 introduced the Block MAX-WAND algorithm for ranked keyword search. Query performance is now competitive with the fastest Postgres-based search solutions, including ParadeDB.
v0.4.0 added posting-list compression, shrinking index sizes by 40% or more. In many cases, pg_textsearch indexes are now smaller than ParadeDB equivalents. This release also includes stability improvements for indexes on tables with large partition counts.
v0.5.0 brings parallel index builds, so CREATE INDEX now uses multiple workers for faster indexing of large tables, along with improvements for BM25 indexes on hypertables and additional stability fixes.
Learn more: pg_textsearch v0.3.0 release notes | pg_textsearch v0.4.0 release notes | Optimize full text search with BM25
Microsoft customers can now sign up for Tiger Cloud through Azure Marketplace with fully self-serve procurement and consolidated billing. Previously, this required manual intervention from our team. Now you can find Tiger Cloud under both annual commit and pay as you go pricing options and get started immediately.
Tiered Storage is now available for Tiger Cloud services running on Azure. You can automatically move rarely accessed data to low-cost Azure Blob Storage while continuing to query it with standard SQL. The experience mirrors what Tiger Cloud AWS customers already have: enable Azure Tiered Storage in Tiger Console, set tiering policies on your hypertables, and Tiger Cloud handles the rest. Customers typically see 2-5x reductions in storage costs depending on compression rates.
Tiger Cloud is now available in the AWS Europe (Zurich) region, enabling low-latency access for applications with European data residency requirements. For a complete list of available regions, see available regions. To learn more, see Get started with Tiger Data.
The SQL editor in Tiger Console is now a floating panel instead of a dedicated tab. It follows you as you navigate around the UI within a service, so you can reference your schema through Explorer while writing queries without switching contexts. This is a small change that makes a real difference in daily workflow.
We brought back the toggle between Ops view and Data view at the service level. You can now switch to Data view from inside an individual service without navigating away. The toggle sits above the top navigation in the UI.
The Jobs page now includes a timeline view that shows the status of all recent job runs at a glance. Hover over any run for details, or click into a specific job for the deep dive. This makes it much easier to spot failures, delays, or patterns without scrolling through log tables. See Monitor your Tiger Cloud services for more.
Tiger Console now includes a built-in panel that walks you through setting up the Tiger CLI (for managing your project from the terminal) and the Tiger MCP server (for adding Tiger Cloud functionality to your AI-assisted development tools). Walkthroughs are available for macOS, Linux, and Windows. Find them in the project view under the CLI/MCP tab.
The Tiger Data (TimescaleDB) Terraform provider now includes a timescale_connector_s3 resource for full Infrastructure as Code management of S3 source connectors. You can declaratively create, update, and manage connectors that support CSV and Parquet files, multiple auth methods, and configurable sync options alongside the rest of your TimescaleDB infrastructure.
Available in Terraform provider v2.7.0+.
All of these features are live on Tiger Cloud. If you are already a customer, check out TimescaleDB 2.25 on your compressed hypertables, try the new floating SQL editor, or explore the Jobs timeline view.
If you are new to Tiger Cloud, start a free trial and see what a purpose-built Postgres cloud looks like.
pg_textsearch 1.0: How We Built a BM25 Search Engine on Postgres Pages
pg_textsearch 1.0 brings native BM25 search to Postgres. No Elasticsearch sidecar needed. Learn how it works and see benchmarks vs. ParadeDB at 138M documents.
Read more
Receive the latest technical articles and release notes in your inbox.
