asopi tech OSS Developer
asopi tech
asopi tech OSS Developer

Unified Database

Alopex DB

1 File, 5 Modes, Infinite Scale

Same data file scales seamlessly from embedded → server → cluster. No data conversion or migration pain when scaling up—a rare capability only ultra-versatile products like Redis can offer. Raft-based KVS as core with SQL, vector search, and graph ETL in one engine.

Same data file from embedded to cluster. Raft-based KVS + SQL + Vector + Graph ETL in one engine.

Pre-Alpha OSS
Deployment Modes
5
Data Models
SQL+Vector+Graph

Key points

  • Same data file across all 5 modes, no migration required
  • Raft-based KVS with SQL, Vector, and Graph ETL integrated
  • Zero code changes to scale, 4 systems consolidated into 1
database vector rust

Challenge

The pain of ever-growing DB products for different needs and scales

Application development requires various data models and scale forms depending on use cases and project size—embedded DBs like SQLite, vector search engines, RDBs, database clusters, and more.

  • Separate products, separate connection methods, separate development
  • Separate setup and infrastructure management for each product
  • Architecture redesign and data conversion when scaling—an excruciating ordeal

Alopex DB uses the same data file to scale seamlessly from embedded to cluster. Unlike typical DB products that require painful data conversion and migration when switching, Alopex DB eliminates this entirely. Raft-based KVS as core with SQL, vector search, and graph ETL in one engine.

Currently in Pre-Alpha development. Bug reports and feature requests are welcome via GitHub Issues.

Raft-based KVS + SQL + Vector + Graph ETL = One Engine

Alopex DB uses the same data file from embedded → server → cluster with seamless scale-up. Enables 'scale without data conversion'—a capability only ultra-versatile products like Redis offer. Raft-based KVS as core with SQL, vector search, and graph ETL in one engine.

  • Same data file across all 5 deployment modes
  • No data conversion or migration when scaling
  • Raft consensus for strongly consistent distributed KVS
Data conversion
None
Code changes to scale
0

Capabilities

Five Modes, One Engine

Same data file across all modes. Switch topologies based on operational needs without data conversion.

WASM & Embedded

Browser apps, mobile, edge devices, local RAG

  • WASM Viewer: Read-only in-browser, IndexedDB/OPFS
  • Embedded: SQLite-like single-file, full ACID
  • Zero-config, same SQL interface everywhere

Single-Node & Replicated

Microservices, dev/test, read-heavy workloads

  • Single-Node: Standalone server, Postgres wire protocol*
  • Replicated: Leader + read replicas, async replication
  • Chirps messaging layer with SWIM + QUIC

Distributed Mode

High-availability, global production workloads

  • Shared-nothing cluster with Range Sharding
  • Raft consensus for strong consistency
  • Automatic shard splitting and rebalancing

Features

Key Features

Features designed for AI applications with hybrid SQL + vector + graph queries.

Native Vector Search

VECTOR(N) data type with HNSW, IVF, and Flat indexes. Hybrid SQL + vector queries in single statement.

Graph-Ready Storage

Optimized for Knowledge Graphs with nodes, edges, and embeddings. Graph traversal with SQL syntax.

Chirps Cluster Messaging

SWIM protocol over QUIC transport. 3-layer architecture: Gossip, Raft, and Application messaging.

Lake-Link Architecture

Zero-ETL Parquet import. Cold/hot data tiering with instant re-hydration.

Pure Rust Engine

Custom LSM-Tree storage, MVCC transactions, memory-safe. WebAssembly ready.

Tech Stack

Technology Stack

A next-generation database engine built from the ground up in Rust.

Storage & Transactions

LSM-Tree, MVCC, Raft Consensus

Custom LSM-Tree with vector-optimized layout MVCC for read isolation, OCC for writes Raft consensus for distributed consistency

Vector & Graph

HNSW, IVF, Flat indexes

VECTOR(N) with configurable dimensions Hybrid SQL + vector in single query Graph nodes/edges as first-class citizens

Cluster Messaging

Chirps: SWIM + QUIC

Gossip layer for membership & failure detection Raft layer for consensus coordination Application layer for client queries

Status

Development Status

Alopex DB is currently in Pre-Alpha. Star the GitHub repo to follow progress.

公開リソース

最新の資料・コード・コミュニティ導線をまとめています。

Get Involved

Join the Development

Alopex DB is being developed as open source. Star the GitHub repo to track progress.