BlockMax WAND: How Weaviate Achieved 10x Faster Keyword Search

Weaviate's hybrid search utilizes keyword and vector search to enhance data retrieval, with the BlockMax WAND algorithm offering significant improvements in document scoring speed. By optimizing the inverted index and employing advanced compression methods like varenc and delta encoding, BlockMax WAND reduces the number of documents inspected during keyword searches, leading to faster query times and a decrease in disk space usage by 50-90%. The process involves dividing posting lists into blocks with local max impact, which allows for more efficient data skipping and loading, while the compression techniques applied to term frequencies and doc IDs further enhance storage efficiency. As a result, p50 query times are reduced to 10-20% of the original, and Queries Per Second (QPS) are significantly increased, marking a crucial step towards making Weaviate's hybrid search scalable to billions of documents. However, BlockMax WAND is currently available as a technical preview in version 1.29, with potential changes expected in future updates, and it is not yet recommended for production environments.

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.