Google has added inline filtering to its AlloyDB for PostgreSQL database. Another addition is observability and management tooling for vector indexes, including a new recall evaluator and Google is also introducing vector index distribution statistics.

AlloyDB was developed by taking the PostgreSQL API and extending it with Google's own machine learning, AI and storage features. It automates administrative tasks such as backups, replication, patching, and capacity management and uses adaptive algorithms and machine learning for PostgreSQL vacuum management, storage and memory management, data tiering, and analytics acceleration.

AlloyDB for PostgreSQL uses Google Research's vector search index, ScaNN, for end-to-end retrieval of data with a single SQL statement. Google has now added more features to make vector search faster.

The first addition is inline filtering, which improves the performance of filtered vector search in AlloyDB. AlloyDB can perform filtered vector search directly in the database, instead of post-processing on the application side. Google says Inline filtering combines the best of vector indexes and traditional indexes on metadata columns to achieve better query performance.

Inline filtering, is a new query optimization technique that allows the AlloyDB query optimizer to evaluate both the metadata filtering conditions and the vector search in tandem, leveraging both vector indexes and indexes on the metadata columns. Inline filtering is now available for the ScaNN index in AlloyDB, a search technology based on over a decade of Google research into semantic search algorithms.

AlloyDB already had pre-filtering that is used when filters are very selective so that only a small number of rows match, so pre-filtering uses an index to find the small subset of rows that match the filter, then performs a nearest-neighbor search on only those rows. AlloyDB also has post-filtering for use where a large percentage of rows match the filtered condition. In this case the query planner starts with the vector index to come up with a list of relevant candidates, and then removes results that do not match the predicates on the metadata columns.

Inline filtering offers an advantage for queries that fall between these two extremes. As AlloyDB searches through the vector index, it only computes distances for vectors that match the metadata filtering conditions. 

The second addition is observability and management tooling for vector indexes, including a new recall evaluator. This means users no longer have to build your their measurement pipelines and processes for their applications to deliver good results.

Google is also introducing vector index distribution statistics. This is aimed at customers with rapidly changing real-time data and Google says it will help achieve more stable, consistent performance.

The updated AlloyDB is available now. 

 

More Information

AlloyDB

Related Articles

Google Announces Downloadable AlloyDB

AlloyDB For PostgreSQL Now Generally Available

Google Announces AlloyDB To Free Users From Legacy Databases

To be informed about new articles on I Programmer, sign up for our weekly newsletter, subscribe to the RSS feed and follow us on Twitter, Facebook or Linkedin.

Comments

Make a Comment or View Existing Comments Using Disqus

or email your comment to: comments@i-programmer.info