Vector Search FAQs

Vector search is widely used for queries where relevance is important. For example, searching a large corpus of documents for documents related to a specific use case or generating recommended items for a shopping cart. Increasingly, vector stores are also being used as a persistent cache for generative AI models to prevent the need for recomputation.

Rockset vector operations support float and int type vectors.

Vectors are syntactically identical to arrays and there is no limit on array size, so there is no limit on vector size.

Vectors can be bulk ingested and stream ingested as part of incoming documents the same as arrays. Just be sure to use VECTOR_ENFORCE on the vector field in your collection's Ingest Transformation.

Yes, you manipulate vectors in all of the same ways you would manipulate arrays. Rockset sits on top of an LSM tree based RocksdDB storage engine, which means random mutations are fast. Updates, inserts, and deletes are immediately visible in any ANN index associated with the vector.

There is no general limit on the number of vectors that can be stored in Rockset. Rockset's disaggregated Storage Architecture allows your storage tier to scale independently of your compute needs.

Rockset is already built for low-latency complex analytics on real-time data which perfectly complements vector search use cases.