data · open source
pgvector development
Vector search inside Postgres, no separate vector database to operate.
- Category
- data
- Vendor
- Open source
- We use it for
- 4 capabilities
The honest assessment
- What it is
- Vector search inside Postgres, no separate vector database to operate.
- Strongest at
- one database for relational and vector data, with transactions across both
- Trade-off
- at very large vector volumes a dedicated index outperforms it
- Category
- data
- Vendor
- Open source
We hold no reseller commission on pgvector. That is what makes the trade-off line above worth reading. It costs us nothing to tell you when this is the wrong choice.
Building with pgvector
pgvector is strongest at one database for relational and vector data, with transactions across both. We reach for it when that is the property a workload actually depends on, and we say so when it is not.
The strongest argument for pgvector is operational simplicity. Your embeddings live in the same database as your users, permissions and business records, so a retrieval query can join against them in one statement rather than coordinating two systems.
That also makes permission-aware retrieval far easier. Filtering results by what a user is actually entitled to see is a WHERE clause here, where a separate vector store turns it into an application-layer problem you have to get right every time.
We build the smallest thing that proves the case, put it in front of real users, and expand only what earns its keep. Six weeks to something running in production, not six quarters to a strategy document.
Alternatives in data
Questions
What is pgvector best at?
One database for relational and vector data, with transactions across both.
When would you not use pgvector?
At very large vector volumes a dedicated index outperforms it. We would look at Pinecone or Elasticsearch in that situation.
Do you have a commercial relationship with pgvector?
No. We hold no reseller commission on any technology we recommend, which is what lets the trade-off above be stated plainly.
Can you work with our existing pgvector setup?
Yes. We would rather extend and stabilise something that already works than introduce a parallel system your team has to learn.
Working with pgvector?
Tell us the workload and we will tell you whether this is the right tool for it.
Or email bd@dtrasglobal.com · call +91 74118 77878