PlanetScale recently announced the intention to fork MySQL adding vector search. While PostgreSQL has been the default open-source choice for vector search, the company behind the Vitess database wants to release a version of MySQL and PlanetScale with vector support.
PlanetScale will implement the Hierarchical Navigable Small World (HNSW) algorithm, an option that constructs optimized graph structures and makes searching vector similarity in large datasets efficient. Nick Van Wiggeren, CTO at PlanetScale, writes:
AI/ML apps that want to harness the power, stability, reliability, and scalability of MySQL. Instead of adopting a second database just for vectors, you'll be able to do the same storage and retrieval right in PlanetScale, reducing cost and operational burden significantly.
Thanks to the pgvector extension, PostgreSQL is the most popular open-source database for projects requiring vector similarity search. While Oracle recently announced enhancements to MySQL HeatWave, including support for vector store, the MySQL Community Edition does not support yet it. Peter Zaitsev, founder at Percona and open source advocate, comments:
Oracle's focus on MySQL Proprietary Cloud Noticed Innovation has become increasingly noticed by the community. Great to see PlanetScale to step up to introduce Vector support in MySQL.
Vector support is critical for embedding, the technique using machine learning to transform arbitrary data into a vector and create a uniform numerical representation that can be analyzed for similarity. The feature targets use cases such as semantic search, recommendation systems, anomaly detection, and image recognition workloads. John Hwang, generative AI architect at AWS, comments:
Vector database is not a separate database category. In fact, every database will offer vector search sooner or later. In the near future, (...) the boundary between what’s a vector DB and what’s not will blur.
PlanetScale is not the only database that introduced vector support in the last few months: Cloudflare announced the beta release of Vectorize, Timescale Vector enhances pgvector with faster search, and Google Cloud integrated vector search in managed databases. Furthermore, the Vector Engine for Amazon OpenSearch Serverless is now in preview, and Aurora PostgreSQL supports pgvector, as separately reported on InfoQ. MongoDB Atlas Vector Search launched recently and Oracle announced last month the plan to support AI vector similarity search in Oracle Database 23c. AnalyticDB for MySQL, a managed database on Alibaba Cloud, provides a vector search feature.
The intention to fork MySQL has had positive reactions. Van Wiggeren adds:
PlanetScale already maintains a fork of MySQL and we'll be adding vector types and indexes to it. When released, we'll run that MySQL fork in PlanetScale as we do today. We will publish packages and containers for our PlanetScale-flavored MySQL that will allow users to test and develop locally.
The feature is currently in private beta and developers can sign up to be notified of upcoming releases.