asopi tech
asopi techIndie Developer

Alopex DB · Core features

Alopex DB features

An overview of Alopex DB: an embedded database that keeps everything in one file, with SQL, vector search, KV, and a server.

Updated: Oct 11, 2026

Alopex DB is an open-source database written in Rust that you embed in your application. Your data lives in a single file, and SQL, vector search, and key-value access run in one engine. Update table data and embedding vectors in the same transaction, with no second database to stitch in.

Architecture

Rust embedded Python CLI HTTP server gRPC server DataFrame SQL Direct APIs Core Transactions Storage engine Vector search Columnar Single file (.alopex) Cluster (Chirps)

Six ways to use Alopex DB (embedded Rust, Python, the CLI, an HTTP server, a gRPC server, and DataFrames) go through SQL and the Direct APIs to the same core and the same single file. Whichever way you use it, your data is the same .alopex file. Transactions, the storage engine, vector search, and columnar data come together in one core.

When to use it

  • Put a database with SQL and vector search inside an app or tool, with no server to run.
  • Keep vectors and attribute data in one file for RAG and similarity search.
  • Run embedded while you develop, then stand up an HTTP and gRPC server when you need one.

Your data in one file

A database is a single .alopex file. Backing it up or handing it over is a file copy.

Main features

Storage. Writes are taken as appends to a log and, by default, synced to disk in batches. The database recovers after a crash. Row tables, key-value pairs, vectors, and columnar data share one storage layer. For columnar tables, you choose per table between Zstd compression and no compression.

Transactions. ACID transactions, with reads that never block writes. Control them with BEGIN, COMMIT, ROLLBACK, and SAVEPOINT.

SQL. JOIN, recursive CTEs, window functions, aggregation, views, constraints, EXPLAIN, and full-text search.

Vector search. Store embeddings in vector columns and search by cosine, L2, or inner product. Use exact search over every row, or fast approximate search with an HNSW index. Combine either with SQL conditions to find the nearest rows among those that match.

Embedded and server. Use it as a library from Rust and Python, and work with SQL, KV, and vectors from the command line. It also runs as an HTTP and gRPC server with TLS, audit logs, and metrics. You keep using the same database from embedded use to a server.

Cluster foundation. Cluster state management and routing diagnostics. Node-to-node messaging uses the related product Chirps.