FAQ
What is SrvDB?
Section titled “What is SrvDB?”An open-source embedded vector database for Python and Rust. It stores embeddings on local disk and finds the most similar ones to a query, inside your own process: semantic search, RAG and AI memory without a server.
Is SrvDB the same as SurrealDB?
Section titled “Is SrvDB the same as SurrealDB?”No. SrvDB (S-R-V-D-B) is an embedded vector database library. SurrealDB is an
unrelated multi-model database server. Install SrvDB with pip install srvdb
or cargo add srvdb.
Does SrvDB need a server or internet access?
Section titled “Does SrvDB need a server or internet access?”No. It is a library inside your process and never opens a network connection.
Which embedding models work?
Section titled “Which embedding models work?”Any model with a dimension between 128 and 4096. Use the model’s own dimension; no padding is needed.
Do I have to normalise vectors?
Section titled “Do I have to normalise vectors?”No. SrvDB ranks by cosine similarity, which ignores length.
Why don’t my latest vectors survive a crash?
Section titled “Why don’t my latest vectors survive a crash?”Only persist() makes writes durable. See Durability.
Why do I get BlockingIOError when opening?
Section titled “Why do I get BlockingIOError when opening?”The database is already open, in this process or another. Share one
Database object between threads instead.
Can I use SrvDB in a commercial product?
Section titled “Can I use SrvDB in a commercial product?”Yes. SrvDB is Apache-2.0 licensed. Commercial support is available: srinivasvarma764@gmail.com.
How fast is it?
Section titled “How fast is it?”We publish numbers only from reproducible benchmarks; see Benchmarks.