PlanetScale has announced that vector support is now available in the public. This feature allows you to store vector data along with relational MySQL data, eliminating the need for a separate vector database. The implementation is based on MySQL forks and includes advanced features such as Euclidean (L2), index vectors by dot product or cosine distance, supporting vectors with dimensions up to 16,383.
“We’re committed to creating a new and exciting new environment,” said Patrick Reynolds, software engineer at PlanetScale.
“Since the start of open beta, we’ve doubled query performance, improved memory efficiency eight times, and focused on robustness, making sure it’s as solid as all other data types MySQL support.”
The architecture supports transactional operations. This means that vector data insertions, updates, and deletions are reflected immediately in the index. This feature is integrated into MySQL’s default storage engine and allows for seamless interaction with queries such as Join and Where Clauses.
For more information, see the official PlanetScale announcement and the hacker news discussion.
At the next 2025 meeting, Google Cloud announced various extensions across database products, particularly focusing on AI capabilities within AlloyDB. The open source PGVector extension allows users to store vector embeddings directly in the database and query them using approximate adjacency (ANN) algorithms.
With the introduction of a scalable nearest neighbor (scan) algorithm, Google Cloud has reported significant performance improvements. This is an 8x increase in the speed of creating vector indexes and a 4x increase in the serving of vector queries. According to Andi Gutmans of Google Cloud:
“We see thousands of customers working on vector processing.”
The extension also includes the new AlloyDB AI query engine that allows you to embed natural language queries directly within SQL.
For more insights, see the official Google Cloud blog and the Scann whitepaper.
AMD has released Gaia, an open source framework designed to run large-scale language models (LLMs) locally on consumer hardware. Gaia aims to support searched generation (RAG) and provide solutions to environments that are sensitive to latency and privacy concerns.
/filters:NO_UPSCALE()/News/2025/04/AMD-Gaia-llm/en/resources/13274150-technical-blog-gaia-1-1743878291733.png)
GAIA includes APIs that are compatible with local open AI and tools for indexing local data sources. Inference can be generated on-device, resulting in significant latency reductions. This framework can handle a variety of content sources and vectorize them for context queries.
Developers can access Gaia’s source code on GitHub to further explore its functionality.
Future MySQL and HeatWave Summit will feature key speakers from Oracle and the broader database community. Notable speakers include Wim Coekaerts, Executive Vice President, Oracle, and Nipun Agarwal, Senior Vice President, MySQL Database and Heatwave Development.
The Summit is designed to discuss advances in MySQL and HeatWave, focusing on how organizations can leverage these innovations to improve database management.
For more information, please see the MySQL Events page.
In the area of authentication, businesses are increasingly looking for efficient and secure solutions. Mojoauth specializes in providing passwordless authentication solutions, including PassKeys, Telephone OTP, Email OTP, and a comprehensive passwordless experience. This migration not only enhances security, but also simplifies the user’s login process.
By integrating passwordless authentication into web and mobile applications, organizations can significantly reduce the risk of breach while providing a seamless experience for their users. Explore Mojoauth’s products to improve your certification strategy.
***This is Mojoauth’s Security Blogger Network Syndicate Blog. You will be redirected to the passwordless created by DeveshPatel. Read the original post at https://mojoauth.com/blog/planetscale-vectors-ga-a-game-changer-for-mysql-databases/
