EDB Vector Plus (edb_vectorplus) is a Postgres extension that provides a new vector index type, ivfplus. It implements hierarchical Inverted File (IVF) clustering and RaBitQ quantization for vector search, the same general approach used by leading standalone vector databases (for example, Turbopuffer's ANN v3).
Vector Plus coexists with pgvector — it uses the same vector type and requires pgvector 0.7.0 or newer as a dependency — but delivers higher query throughput and lower latency with a reduced CPU and memory footprint, particularly at large row counts where pgvector's ivfflat index becomes prohibitively slow.
It's packaged and distributed through EDB's own repositories, the same distribution model as other proprietary EDB extensions like AIDB.
Names at a glance
| Description | Name |
|---|---|
| SQL extension | edb_vectorplus |
| Index access method | ivfplus |
| Required extension | vector (pgvector) ≥ 0.7.0 |
Requirements
EDB Vector Plus is supported on Postgres 14 through 18 and requires pgvector (vector) 0.7.0 or newer. EDB packages are available for this range. See Supported versions and platforms.
Where to start
- Overview — What Vector Plus is, how the index works, and known limitations.
- Install — Set up the EDB repository and install the
edb_vectorpluspackage. - Quickstart — Build your first
ivfplusindex and run a query end to end. - Using — Storing and querying vectors, tuning
listsandprobes, filtering and hybrid search, and measuring recall. - Troubleshooting — Common questions and problems.
- Reference — The full interface reference: operator classes, index options, session parameters, functions, and fixed limits.