ReleaseBytes
gcp Google Cloud Blog ·

AlloyDB boosts pgvector HNSW vector search speed by 4x

blogaigcppreviewengineer
feature patch

AlloyDB introduces columnar engine accelerated HNSW for pgvector, achieving up to 4x higher queries per second (QPS) for vector search compared to standard PostgreSQL. This enhancement is crucial for enterprise AI applications that require maximizing QPS without sacrificing search accuracy. The feature is now available in preview for demanding workloads and requires no application code changes.

  • →Columnar engine accelerated HNSW for pgvector in AlloyDB
  • →Improved performance through in-memory columnar cache and vectorized access
  • →Columnar engine accelerated HNSW offers cost and quality benefits
  • →AlloyDB provides flexibility with multiple vector index options
  • →Steps to enable and use columnar engine accelerated HNSW
Features (1) ›
  • Columnar engine accelerated HNSW for pgvector in AlloyDB

    AlloyDB now offers columnar engine accelerated HNSW, a feature designed to significantly improve vector search performance for pgvector. This enhancement provides up to 4x higher QPS compared to standard PostgreSQL HNSW, while maintaining or improving search recall.

Enhancements (1) ›
  • Improved performance through in-memory columnar cache and vectorized access

    The columnar engine stores frequently accessed index data in an optimized, in-memory columnar format. This bypasses standard PostgreSQL buffer manager overhead and uses a vectorized memory layout for faster traversals, leading to substantial QPS and recall gains.

Notes (3) ›
  • Columnar engine accelerated HNSW offers cost and quality benefits

    This feature allows for reduced infrastructure costs by achieving higher performance with fewer compute resources. It also enables AI applications to attain better search accuracy at speeds previously only possible for lower-quality searches, with no application code modifications needed.

  • AlloyDB provides flexibility with multiple vector index options

    Beyond HNSW, AlloyDB also supports ScaNN, a Google Research-backed index, and standard k-nearest neighbor (KNN) search for use cases requiring absolute precision and 100% recall.

  • Steps to enable and use columnar engine accelerated HNSW

    To utilize this feature, users need to enable the google_columnar_engine.enabled and google_columnar_engine.enable_index_caching flags. After creating an HNSW index via pgvector, a specific SQL command caches the index within the columnar engine.

Read the original announcement →

https://cloud.google.com/blog/products/databases/supercharge-pgvector-4x-faster-hnsw-with-alloydb/

Related releases