Future-proof your data strategy: AlloyDB adds PostgreSQL 18 and new Extended Support

As you look out at your 2026 infrastructure roadmap, your goal is to balance the need for rapid innovation with operational stability. You shouldn’t have to choose between adopting the latest database features and maintaining a secure, supported environment for your workloads. Today, we are announcing the general availability of PostgreSQL 18 in AlloyDB and the introduction of Extended Support for older major versions. These updates give you the flexibility to build with the most advanced open-source tools while providing a reliable, long-term safety net for your existing applications.

Maintain stability with Extended Support

Moving production workloads to a new major version is a significant undertaking that requires careful planning and testing. To provide you with the flexibility to upgrade on your own schedule without compromising security, we are introducing Extended Support for AlloyDB.

This new offering bridges the gap between community end-of-life (EOL) dates and your upgrade timelines, ensuring business continuity for your most critical applications.

Key timelines

AlloyDB clusters are eligible for three years of Extended Support and are automatically enrolled according to the following timeline:

  • PostgreSQL 14: February 1, 2027 through February 1, 2030.

  • PostgreSQL 15: February 1, 2028 through February 1, 2031.

  • PostgreSQL 16: February 1, 2029 through February 1, 2032.

  • PostgreSQL 17: February 1, 2030 through February 1, 2033.

  • PostgreSQL 18: We will announce the Extended Support timeline at a later date.

What’s included

During the Extended Support period, Google Cloud provides:

  • Critical security patches: Protection against all High and Critical severity common vulnerabilities and exposures (CVEs)

  • Proactive bug fixes: Resolution of issues within AlloyDB-maintained code

  • SLA coverage: Continued availability protection for clusters that meet eligibility criteria.

  • New cluster creation: The ability to provision new clusters on Extended Support versions

Managing your transition

Extended Support will be available for an additional fee. You’ll be able to opt out at any time by simply upgrading your clusters to a major version in regular support. We’ll announce extended support with updated pricing later this year.

Speed up development with PostgreSQL 18 on AlloyDB

PostgreSQL 18 (PG18) introduces features designed to make your applications faster and your development process more intuitive. When you choose PG18 for your AlloyDB clusters, you gain immediate access to performance improvements and modern data handling:

  • Faster queries with B-tree skip scans: The engine can now bypass index entries that don’t match your query, accelerating data retrieval.

  • Parallel GIN index usage: You can speed up full-text and JSON searches by utilizing multiple CPU cores.

  • Virtual generated columns: You can use columns that are computed on-the-fly, providing the same easy API as stored columns without using extra disk space.

  • Native UUIDv7 support: You can build distributed applications with UUIDv7, which offers better sortability and indexing efficiency than traditional random UUIDs.

Proven reliability: UKG modernizes the data foundation

Upgrading a production system at scale can be complex, but Google Cloud makes the process simple for PostgreSQL in AlloyDB, automating many of the pre- and post-upgrade tasks. UKG, a provider of HR and payroll solutions, recently upgraded its AlloyDB clusters to PostgreSQL 17 to power new features for their near-real-time data foundation, People Fabric. Managing a high-density, multi-tenant architecture with a massive number of database objects presented a significant challenge. By using in-place major version upgrades, UKG minimized risk and avoided impact to their users.

“Upgrading a multi-tenant environment with thousands of objects usually introduces significant risk, but AlloyDB’s in-place upgrade path allowed us to modernize our fleet without the typical downtime or performance regressions,” said Rajiv Jain, Sr Director, Engineering, Data Platform, UKG. “This enabled us to hit our targets for our latest release of People Fabric and put the power of new PostgreSQL features to work for our customers.”

In-place major version upgrades allow you to modernize your database without moving data or changing connection strings, reducing upgrade time to minutes. This streamlined path applies to all target versions, including PostgreSQL 18, so that even massive, multi-tenant fleets can adopt the latest features with minimal downtime. When paired with query plan management, this process provides a fast, predictable, and low-risk transition to the newest PostgreSQL releases.

Maximize performance with database-aware storage

AlloyDB builds upon the innovations of PG18, extending its capabilities with a specialized architecture. By separating compute from storage, we offload heavy database operations to a dedicated, intelligent layer. This has the following advantages:

  • Database-aware offloading: AlloyDB delegates logging and maintenance tasks to a dedicated service. This frees your primary database instance to focus entirely on processing transactions.

  • Horizontal scaling without data duplication: You can add read-only replicas in seconds. Because every replica attaches to the same distributed storage, you avoid the cost and lag of managing multiple copies of your data.

  • Better price-performance: Recent benchmarks show that AlloyDB provides up to 2x better price-performance than self-managed PostgreSQL. Even with half the compute resources of a self-managed environment, AlloyDB delivers higher transactions per minute by using its kernel more efficiently.

  • Pay only for what you use: AlloyDB storage is elastic. You don’t need to specify a storage size; the system grows and shrinks automatically based on your data.

  • Predictable throughput: Many cloud databases require you to buy more storage just to get higher performance. AlloyDB does not limit speed based on storage size — you get full performance from day one, paying only for the data you store.

Get started

Related Posts