December 2025 Summaries

Postgres 18 has been released on PlanetScale, now serving as the default version for new databases created on the platform, with an option to choose prior versions via a dropdown menu. This release coincides with PlanetScale's recent introduction of affordable $5 single-node and $50 Metal databases, enhancing its appeal for Postgres-backed applications. Key updates in Postgres 18 include a new asynchronous I/O system for improved query performance, built-in support for UUIDv7, and the Skip Scan optimization, which enhances the usage of multi-column indexes. While there is no automated upgrade path from Postgres 17, users can transition by creating a new Postgres 18 database and migrating data using PlanetScale's import guides. Benchmark comparisons showcase the performance advancements from version 17 to 18.

DuckDB has emerged as a powerful tool for OLAP workloads, offering a rich SQL dialect suitable for analytics in conjunction with OLTP-oriented relational databases. The announcement of the pg_duckdb extension for Postgres databases on PlanetScale, in partnership with MotherDuck, marks a significant development in this space. This extension allows DuckDB's column-store analytics engine to be integrated within Postgres, enabling seamless execution of OLTP and OLAP queries. Users can create tables in either the standard Postgres or DuckDB formats and leverage DuckDB to query external data sources like Apache Parquet and Iceberg. The integration with MotherDuck, a cloud data warehouse, further enhances this setup by allowing analytical queries to be offloaded to the cloud, thus preventing performance impacts on OLTP operations. This setup facilitates efficient data movement and unifies analytics with relational data in a seamless manner.

PlanetScale has announced the general availability of its Metal offering for Postgres, introducing smaller and more affordable options starting at $50 per month, with configurations as low as 1GiB of RAM and storage options ranging from 10GB to 1.2TB. This update marks a significant step toward decoupling CPU and RAM from storage capacity, allowing customers to tailor their database configurations to specific workloads while maintaining low latency and reliability through locally attached NVMe drives. The new offering provides flexibility for users to choose optimal storage without necessarily increasing CPU and RAM, addressing customer demand for lower starting prices and customizable storage options. The platform is now available in AWS regions with both Intel and ARM CPUs, and support for GCP and Vitess is anticipated soon, enabling customers to either create new databases or resize existing ones with enhanced I/O capacity.