Alopex DB¶

CLI Defaults¶

On a TTY, the CLI launches the TUI by default; use --batch or --output to force batch output. In the results TUI, press a to open the admin console when available. The status bar emphasizes Connection/Focus/Action and Ops so you can see the current target and primary actions at a glance. The admin console uses a rainfrog-style layout (left resource tree, right detail input, right-bottom status/preview) with h/l focus switching.

Current Status¶

# Rust
cargo add alopex-embedded alopex-sql alopex-server alopex-cluster

# Python
pip install alopex

The Problem¶

Modern AI applications require multiple database technologies—creating complexity, inconsistency, and operational overhead.

Key Features¶

Any Scale, One Engine¶

Start small, scale infinitely—without changing your data model or application code.

Mode	Use Case	Architecture
WASM Viewer	Browser Data Exploration	Read-only viewer with IndexedDB caching
Embedded	Mobile Apps, Local RAG, Edge Devices	Single Binary / Library (like SQLite)
Single-Node	Microservices, Dev/Test Environments	Standalone Server (HTTP/gRPC)
Replicated	High Availability, Read-heavy Workloads	Primary-Replica with automatic failover
Distributed	Large-Scale Production	Multi-Raft Cluster (Range Sharding)

Learn more about deployment modes

SQL + Vector in Action¶

-- Create a table with vector column
CREATE TABLE knowledge_chunks (
    id INTEGER PRIMARY KEY,
    content TEXT,
    embedding VECTOR(1536)
);

-- Hybrid Search: SQL Filter + Vector Similarity
SELECT id, content,
       vector_similarity(embedding, ?) AS score
FROM knowledge_chunks
ORDER BY score DESC
LIMIT 5;

-- Create HNSW index for fast similarity search
CREATE INDEX idx_embedding ON knowledge_chunks
USING HNSW (embedding)
WITH (m = 16, ef_construction = 200);

-- Search with HNSW acceleration
SELECT id, content
FROM knowledge_chunks
ORDER BY vector_similarity(embedding, ?) DESC
LIMIT 10;

use alopex_embedded::Database;

let db = Database::open("./my_data")?;

// Execute SQL
let results = db.execute_sql(
    "SELECT * FROM docs WHERE vector_similarity(embedding, ?) > 0.8",
    &[query_vector]
)?;

// HNSW search
let similar = db.search_hnsw("docs", &query_vector, 10)?;

View SQL + Vector guide

Roadmap¶

gantt
    title Alopex DB Development Timeline
    dateFormat  YYYY-MM
    axisFormat  %Y-%m

    section Foundation
    v0.1-v0.2 Core          :done, 2025-01, 2025-10
    v0.3 SQL + HNSW         :done, 2025-10, 2025-12

    section Python & Server
    v0.3.3 Python Wrapper   :done, 2025-12, 2025-12
    v0.4 Server + DataFrame :done, 2026-01, 2026-01
    v0.5 GROUP BY + DataFrame P1 :done, 2026-01, 2026-01

    section Distributed
    v0.6 SQL JOIN + Nim Parser :done, 2026-07, 2026-07
    v0.7 Cluster-aware      :done, 2026-07, 2026-07
    v0.8 Metadata Raft      :2026-10, 2026-12
    v1.0 GA                 :milestone, 2027-03, 0d

What's Complete¶

Version	Features	Status
v0.1-v0.2	Embedded KV, WAL, MVCC, Vector (Flat), Columnar	Complete
v0.3	SQL Frontend, HNSW Index, Embedded Integration	crates.io Published
v0.3.3	Python Wrapper (alopex-py), CLI	PyPI Published
v0.4.0	Server Mode, DataFrame API, Async/Stream	crates.io Published
v0.5.0	GROUP BY/Aggregation, DataFrame P1 (JOIN/sort/null)	crates.io Published
v0.6.0	SQL JOIN & Subqueries, Nim FFI SQL Parser	crates.io Published
v0.7.x	alopex-cluster, Mode-Parity Suite, gRPC Cluster Admin	v0.7.6 Published
Chirps v0.5	Gossip, SWIM, Membership, Raft Consensus API	Complete

What's Next¶

View detailed roadmap

Python Support (Available Now)¶

import alopex

# Open database
db = alopex.Database.open("./my_data")

# Execute SQL
results = db.execute_sql(
    "SELECT * FROM docs WHERE category = ?",
    ["science"]
)

# Vector search
similar = db.search_hnsw("docs", query_embedding, k=10)

import alopex

# Polars-compatible DataFrame API
df = alopex.read_parquet("data.parquet")

result = (
    df.lazy()
    .filter(alopex.col("score") > 0.5)
    .select(["id", "content", "embedding"])
    .collect()
)

Python Guide

Chirps — Cluster Foundation¶

Alopex Chirps is the control plane for distributed Alopex DB clusters.

Learn about Chirps architecture

Skulk — Time-Series Companion¶

Alopex Skulk is the time-series member of the family — an embedded storage and ingest core for monitoring, IoT, and observability workloads. v0.3.0 replaced the storage engine with Arrow + Parquet columnar, in a 3.7 MB pure-Rust binary.

Query execution (PromQL, SQL-TS) arrives in v0.4; HTTP serving in v0.6.

Learn about Skulk · Quick start

Trail — For Logs, Audit Trails, and Events¶

Write first. The schema binds when you read.

Application logs, audit trails, webhook payloads — the things your system emits when something happens. They arrive with fields nobody declared, sometimes without a usable timestamp, and occasionally with a field that was an integer last week and a string today.

Nobody updates a log line. Alopex DB is transactional — built for rows you change, and strict about types because of it. Skulk is append-only like Trail, but built for numeric series on a schedule. Logs are neither, and that gap is what Trail is for — the newest design in the family.

Trail is being designed in the open. The API is not frozen, so what you need can still change it.

See the preview · Read the concept

Alopex OTel — Storage and Dashboards, One Product¶

OpenTelemetry from embedded to cluster.

Observing one application normally means deploying a Collector, Prometheus, Tempo, Loki, and Grafana — five systems, three storage engines, three query languages. Alopex OTel is one system: ingestion, storage, and the dashboard ship together, and that whole thing runs from a single embedded process to a distributed cluster.

Alopex OTel is a published design. The metrics path builds on what Skulk ships today; traces and logs follow Trail.

Read the concept

Alopex Data UI — One Interface for Data Apps¶

One interface, wherever the data moves. Alopex Data UI is the newest member of the family: a consistent UI model for data applications. Adapters translate schema, queries, filters, selections, and editable records into one DataSource contract, so forms, grids, charts, graphs, and maps evolve as one coherent system — whether data arrives from Alopex DB, DuckDB, an API, or a file.

Currently in early development — the concept and component model are published on the project site.

Visit Alopex Data UI View on GitHub

Alopex Enterprise — Beyond OSS¶

The Alopex platform is not OSS-only. Alopex Enterprise — commercial middleware built on the Alopex OSS foundation — is currently in design:

• Security Suite — physical & logical storage encryption (AES-GCM, XChaCha20), identity with JWT / OIDC / RBAC
• Unified Query Model — one engine for SQL / AQL / PromQL, with Enterprise Search (full-text, vector, aggregation)
• Observe (SRE) — SLO engine, error budgets, and automated actions
• Infrastructure Packages — distributed config store (etcd/ZK compatible), service discovery + internal DNS

Design phase — no fixed delivery dates yet. Interested in early conversations? Get in touch.

Join the Pack¶

Alopex DB is open-source under the Apache 2.0 License.

We welcome contributions from engineers passionate about Rust, Distributed Systems, and Vector Search.

Contributing Guide GitHub Discussions