pgvector

“hybrid search ( plpgsql_bm25 + pgvector ) with Reciprocal Rank Fusion”

“I used this in a different project (halfvec in pgvector) without affecting results”

“you can use quantized vector indexes[0] (handled by pgvector) which will get past the limits imposed by PostgreSQL.”

“a knowledge graph and pgvector which accounts for most of the performance”

“I've been doing some RAG prototypes with hybrid search using pg_textsearch plus pgvector and have been very pleased with the results.”

“I was genuinely surprised just how easy it was to get a fully working RAG set up with Postgres.”

“if you're already running postgres with pgvector, you're in a good spot”

“pgvector is great if you're already on postgres and only need text search”

“Pgvector is definitely the easiest way to get started.”

“surely the easiest solution is to run `CREATE EXTENSION vector`”

“pgvector could handle their use case with less operational overhead”

“pgvector's HNSW indexes handle the approximate nearest neighbor search well enough for workplace search”

“it's certainly simpler than bolting another solution for it”

“PostgreSQL with pgvector now supports HNSW indexing, meaning you can keep your vectors right next to your relational data.”

“have seen very fast lookups with pgvector (HNSW index) and sqlite-vec on 20k vectors”

“pg_vector's economics can be much worse, especially in multi-tenant scenarios”

“You can use pgvector for the vector lookup and paradedb for bm25.”

“something pgvector itself wasn't able to handle in a reasonable amount of time”

“We regularly do tens of thousands of QPS on pgvector fine on massive data stores.”

“pgvector just doesn't really scale to the kinds of workloads we ran here”

“pgvector + tsvector (for hybrid search) seems like the most logical MVP”

“pgvector HNSW indexes (m=64, ef_construction=200) give you sub-100ms semantic search without leaving Postgres”

“Fast approximate nearest neighbor search directly in Postgres”

“you can just add the pgvector extension determine what embeddings that you want to use and try it out without committing to much”

“We actively use pgvector in a production setting for maintaining and querying HNSW vector indexes”

“Vacuuming is grief enough. Rebuilding the index sounds like more of a nightmare than with Solr/Lucene.”

“IVFFlat indexes have to be periodically rebuilt, that HNSW has high overhead”

“Postgres isn't built from the ground-up for high-velocity vector insertions”

“the limitations of pgvector are very real when you're talking indices that grow beyond millions of things”

“found Postgres to be unsuitable at several orders of magnitude less than we currently have”

“He doesn't point to a single thing that pg_vector doesn't do or doesn't do well.”

“Even without fancy indexing, a full scan (1 search vector -> 5M stored vectors), takes less than 40ms.”

“If you need vector search against 100,000 vectors and you already have PostgreSQL then pgvector is a great YAGNI solution.”

“But they’re all workarounds for the fact that pgvector wasn’t really designed for high-velocity real-time ingestion.”

“HNSW index on a few million vectors can consume 10+ GB of RAM or more”

“it's hard to predict when postgres will decide to use pre- vs post-filtering”

“as the author noted is not without some flaws and are typically in the smallish range”

“all the mentioned issues are resolved, I will favour in using the PGVector”

“can support 15x more updates per second comparing to pgvector's HNSW”

“We do at Discourse, in thousands of databases, and it's leveraged in most of the billions of page views we serve.”