MoodengSQL

Owner: AidenHeavensXD

MoodengSQL is a blazing-fast, PostgreSQL-inspired relational database engine written in Rust. It speaks a subset of the PostgreSQL wire protocol and SQL dialect, optimized for low-latency reads and writes.

Features

• PostgreSQL-compatible SQL — CREATE TABLE, INSERT, SELECT, UPDATE, DELETE, CREATE INDEX
• PostgreSQL wire protocol — connect with the built-in CLI or any compatible client
• B-tree indexes — fast point lookups and range scans
• Binary persistence — durable on-disk storage with bincode serialization
• Metadata persistence — schema and indexes survive server restart (meta.bin)
• WAL + checkpoint — write-ahead log with automatic checkpointing (every 50 ops)
• Batch WAL fsync — groups disk sync every 10 ops for higher write throughput
• Transactions — BEGIN / COMMIT / ROLLBACK with undo log
• Index scan — SELECT ... WHERE col = value uses B-tree index when available
• ORDER BY / LIMIT / OFFSET — sorted and paginated queries
• INNER JOIN — multi-table queries with ON clause
• GROUP BY — COUNT, SUM, AVG, MIN, MAX aggregates
• Extended protocol — Parse/Bind/Execute for parameterized queries ($1, $2)
• EXPLAIN — query plan shows Index Scan vs Seq Scan
• Table-level locking — concurrent connections with read/write locks
• Constraints — PRIMARY KEY (auto-index) and NOT NULL enforcement
• Data validation — moodengsql --check verifies catalog/storage consistency
• Configuration — moodeng.toml for server, storage, and logging
• Backup / restore — gzip tar archives of the data directory
• Health check — moodengsql ping for load balancers and Docker
• Docker — Dockerfile and docker-compose for container deployment
• CI — GitHub Actions runs tests and release builds
• Concurrent catalog — lock-free table metadata via DashMap
• Rich types — INT4, INT8, FLOAT4, FLOAT8, TEXT, VARCHAR, BOOL, TIMESTAMP, JSON

Quick Start

# Build everything
cargo build --release

# Start the server (default port 5432)
./target/release/moodengsql --data-dir ./moodeng_data

# Connect with the CLI (in another terminal)
./target/release/moodeng

Example Session

CREATE TABLE users (
    id INT PRIMARY KEY,
    name TEXT NOT NULL,
    email TEXT
);

INSERT INTO users VALUES (1, 'Alice', 'alice@moodeng.dev');
INSERT INTO users VALUES (2, 'Bob', 'bob@moodeng.dev');

SELECT * FROM users WHERE id = 1;

CREATE INDEX idx_users_email ON users (email);

UPDATE users SET email = 'alice@sql.dev' WHERE id = 1;
DELETE FROM users WHERE id = 2;

Architecture

moodengsql/
├── moodeng-core/     # Storage engine, B-tree indexes, SQL executor
├── moodeng-server/   # Async TCP server (PostgreSQL wire protocol)
└── moodeng-cli/      # Interactive command-line client

Component	Technology	Purpose
SQL Parser	sqlparser (PostgreSQL dialect)	Parse incoming SQL
Storage	bincode + fsync	Durable row persistence
Indexes	BTreeMap	O(log n) lookups
Catalog	DashMap	Concurrent schema metadata
Server	tokio	Async connection handling

Server Options

MoodengSQL uses subcommands for operations. Legacy flat flags still work for serve:

# Start server (subcommand or legacy)
moodengsql serve --data-dir ./moodeng_data
moodengsql --data-dir ./moodeng_data --port 5432

# With config file
moodengsql serve --config moodeng.toml

# Validate data directory
moodengsql check --data-dir ./moodeng_data

# Backup and restore
moodengsql backup --output backup.tar.gz --data-dir ./moodeng_data
moodengsql restore --from backup.tar.gz --data-dir ./moodeng_data

# Health check
moodengsql ping --host 127.0.0.1 --port 5432

Copy moodeng.toml.example to moodeng.toml to customize host, port, data directory, and log level.

Docker

docker compose up -d --build
docker compose exec moodengsql moodeng ping --config /etc/moodengsql/moodeng.toml

Data is persisted in the moodeng_data Docker volume at /data.

Benchmark

cargo run --release --example benchmark -p moodeng-core

Prints insert and indexed point-select throughput for a quick sanity check.

CLI Options

moodeng [OPTIONS]

Options:
  -h, --host <HOST>      Server host [default: 127.0.0.1]
  -p, --port <PORT>      Server port [default: 5432]
  -c, --command <CMD>    Execute a single SQL command and exit

Performance Design

MoodengSQL prioritizes speed through:

1. In-memory row cache with lazy disk persistence
2. B-tree indexes for indexed column access instead of full scans
3. Zero-copy value serialization where possible
4. Lock-free concurrent catalog for multi-connection workloads
5. Async I/O with tokio for non-blocking connection handling

License

MIT — Copyright (c) 2026 AidenHeavensXD