Redis Introduces Multi‑Threaded Query Engine to Boost Vector Search Performance for Generative AI

Redis, the popular in‑memory data‑structure store, has launched an enhanced Redis Query Engine that adds multi‑threaded query processing, enabling vertical scaling and significantly higher query throughput while keeping latency low.

The new engine allows concurrent index access, letting Redis handle complex queries such as vector similarity search more efficiently, especially when the dataset grows to hundreds of millions of documents.

Benchmark tests show up to a 16× increase in query throughput compared with the previous generation, outperforming pure vector databases, general‑purpose databases with vector capabilities, and fully managed Redis cloud services in both speed and scalability.

The architecture follows a three‑step workflow: the main thread prepares the query context and queues it, worker threads pull tasks from the shared queue and execute query pipelines concurrently, and results are merged back to the main thread for final response.

Redis measured ingestion and search workloads using HNSW indexing, ANN search, and k‑NN queries on datasets such as gist‑960‑euclidean, glove‑100‑angular, deep‑image‑96‑angular, and dbpedia‑openai‑1M‑angular, employing industry‑standard vector‑db‑benchmark tools.

These improvements are crucial for generative‑AI applications that rely on real‑time Retrieval‑Augmented Generation (RAG), where meeting the “100 ms rule” for end‑to‑end latency is essential.