LatticeDB - Next-Gen Relational Database Management System

Why LatticeDB?

Built from scratch in modern C++17, LatticeDB combines traditional database capabilities with cutting-edge features that major RDBMS don't provide natively out-of-the-box.

🎯

Complete SQL Support

Full SQL parser with DDL, DML, DCL, TCL. JOINs (INNER, LEFT, RIGHT), GROUP BY, aggregates (COUNT, SUM, AVG, MIN, MAX), subqueries, and CTEs all fully implemented.

⏰

Time Travel Queries

Built-in temporal support with FOR SYSTEM_TIME AS OF syntax. Query historical data at any point in time with bitemporal architecture.

🤖

Native Vector Search

Fully implemented vector engine with Flat, HNSW, IVF algorithms and multiple distance metrics (L2, Cosine, Dot Product) for AI/ML workloads.

🔒

Enterprise Security

Row-level security, column encryption (AES-256-GCM, ChaCha20), authentication (password, JWT, certificates), and comprehensive audit logging.

💾

Advanced Storage

Professional buffer pool manager with LRU/Clock replacement, B+ Tree indexes, Write-Ahead Logging (WAL), and ARIES recovery.

⚡

ACID Transactions

Full MVCC implementation with multiple isolation levels, 2PL protocol, deadlock detection, and savepoints for reliable transactions.

🗜️

Adaptive Compression

Working compression engine with RLE, Dictionary, Delta, Bit-packing, LZ4, and ZSTD algorithms for optimal storage efficiency.

📊

Stream Processing

Real-time continuous queries with windowing functions for streaming analytics and materialized views with backfill support.

Feature Matrix

LatticeDB provides features that major RDBMS generally don't offer natively, all unified in a single coherent engine.

Capability	LatticeDB	PostgreSQL	MySQL	SQL Server
Full SQL with JOINs, GROUP BY	✅ Built-in	✅ Built-in	✅ Built-in	✅ Built-in
Time Travel (FOR SYSTEM_TIME)	✅ Built-in	Via extensions	Limited	Temporal tables
Native Vector Search	✅ Built-in	pgvector ext	Via plugins	Limited
Row-Level Security	✅ Built-in	✅ Built-in	Via views	✅ Built-in
Column Encryption	✅ Built-in	Via extensions	TDE only	Always Encrypted
Adaptive Compression	✅ Built-in	Basic	Basic	Advanced
Stream Processing	✅ Built-in	Via extensions	Via binlog	CDC
Custom Merge Policies (CRDTs)	✅ Built-in	Third-party	Third-party	Third-party
Built-in Web GUI	✅ Built-in	Third-party	Third-party	Third-party

System Architecture

Modular, layered architecture with clear separation of concerns for extensibility and maintainability.

flowchart TD
    subgraph "Client Layer"
        GUI[Web GUI]
        CLI[CLI Client]
        SDK[SDKs]
    end

    subgraph "Query Engine"
        PARSE[SQL Parser]
        OPT[Query Optimizer]
        EXEC[Executor]
    end

    subgraph "Storage Layer"
        MVCC[MVCC Controller]
        ULS[Universal Log Store]
        IDX[B+ Tree & Vector Indexes]
    end

    subgraph "CRDT Engine"
        LWW[LWW Register]
        GSET[G-Set]
        COUNTER[PN-Counter]
    end

    subgraph "Security & Privacy"
        RLS[Row-Level Security]
        DP[Differential Privacy]
        AUDIT[Audit Logger]
    end

    GUI --> PARSE
    CLI --> PARSE
    SDK --> PARSE
    PARSE --> OPT
    OPT --> EXEC
    EXEC --> MVCC
    MVCC --> ULS
    ULS --> IDX
    EXEC --> RLS
    EXEC --> DP
    EXEC --> AUDIT
    ULS --> LWW
    ULS --> GSET
    ULS --> COUNTER

Query Processing Pipeline

flowchart LR SQL[SQL Query] --> Parser[Parser] Parser --> AST[AST] AST --> Analyzer[Analyzer] Analyzer --> Planner[Planner] Planner --> Optimizer[Optimizer] Optimizer --> Plan[Execution Plan] Plan --> Executor[Executor] Executor --> Result[Results]

Full SQL support with recursive descent parser, cost-based optimizer, and volcano-style execution.

Transaction & Concurrency

stateDiagram-v2 [*] --> GROWING: BEGIN GROWING --> SHRINKING: Lock Acquired SHRINKING --> COMMITTED: COMMIT SHRINKING --> ABORTED: ROLLBACK COMMITTED --> [*] ABORTED --> [*]

MVCC with 2PL locking, multiple isolation levels, and automatic deadlock detection.

Vector Search Engine

flowchart TD Query[Vector Query] --> Engine[Vector Engine] Engine --> Algo{Algorithm} Algo --> Flat[Flat/Brute Force] Algo --> HNSW[HNSW Graph] Algo --> IVF[IVF Index] Flat --> Results[Top-K Results] HNSW --> Results IVF --> Results

Multiple algorithms and distance metrics for efficient similarity search in high-dimensional spaces.

Quick Start

Get LatticeDB up and running in minutes

Prerequisites

C++17 toolchain (clang++/g++)
CMake ≥ 3.15
Linux/macOS/WSL2

Build Instructions

# Clone repository
git clone https://github.com/hoangsonww/LatticeDB-DBMS.git
cd LatticeDB-DBMS

# Build
mkdir build && cd build
cmake .. -DCMAKE_BUILD_TYPE=Release
make -j$(nproc)

# Launch CLI
./latticedb

Using Docker

# Build Docker image
docker build -t latticedb .

# Run container
docker run -it --rm -v $(pwd)/data:/data latticedb

# Or use docker-compose
docker-compose up -d

Start the Web GUI

# Start the database server
./build/latticedb_server

# In a new terminal, start the GUI
cd gui
npm install
npm run dev

# Open http://localhost:5173

The GUI includes SQL editor, schema browser, query history, and dark/light modes.

Example Queries

-- Create table
CREATE TABLE users (
  id INTEGER PRIMARY KEY,
  name VARCHAR(100),
  email VARCHAR(255) UNIQUE
);

-- Insert data
INSERT INTO users VALUES (1, 'Alice', 'alice@example.com');

-- Join query
SELECT u.name, COUNT(o.id) as order_count
FROM users u
INNER JOIN orders o ON u.id = o.user_id
GROUP BY u.id, u.name;

-- Time travel query
SELECT * FROM orders
FOR SYSTEM_TIME AS OF TX 5;

-- Vector search
SELECT * FROM embeddings
WHERE distance(vector, [0.1, 0.2, ...]) < 0.5
ORDER BY distance(vector, [0.1, 0.2, ...])
LIMIT 10;

Production Deployment

Deploy to major cloud providers with built-in monitoring and observability

☁️

AWS

ECS Fargate with auto-scaling, EFS storage, and CloudWatch monitoring

cd aws && ./deploy.sh

Application Load Balancer
Auto-scaling groups
EFS persistent storage
CloudWatch logs & metrics

🌐

Azure

Container Apps with serverless scaling and Azure Files integration

cd azure && ./deploy.sh

Serverless containers
Automatic HTTPS
Azure Files storage
Application Insights

🚀

Google Cloud

Cloud Run with global load balancing and Cloud SQL

cd gcp && ./deploy.sh

Fully managed serverless
Global load balancer
Cloud Storage integration
Cloud Monitoring

🏗️

HashiCorp Stack

Nomad orchestration with Consul service mesh and Vault secrets

cd hashicorp && ./deploy.sh

Service mesh (Consul)
Secret management (Vault)
Orchestration (Nomad)
On-premises ready

Built-in Monitoring & Observability

📊

Prometheus Metrics

Request rates, response times, database performance

📈

Grafana Dashboards

Real-time visualization and custom dashboards

🚨

Intelligent Alerting

PagerDuty, Slack, and email notifications

🔍

Distributed Tracing

Jaeger and OpenTelemetry integration

Documentation

Comprehensive guides and references

📖

Performance Benchmarks

Competitive performance on OLTP workloads

Operation	LatticeDB	PostgreSQL	MySQL
Q1 (Scan+Aggregate)	1.2s	1.5s	2.1s
Q3 (Join)	2.1s	2.3s	3.2s
Q5 (5-way Join)	3.2s	3.5s	4.8s
Insert 1M rows	8.5s	9.2s	11.3s
Vector Search (1M, 768D)	<100ms	~150ms	N/A

Test Environment: AWS c5.4xlarge (16 vCPU, 32GB RAM), TPC-H SF-10 dataset

Note: Performance may vary based on workload, hardware, and configuration

Join the Community

Contribute, learn, and grow with LatticeDB

💻

Contribute

Help improve LatticeDB by contributing code, documentation, or bug reports

View on GitHub →

📚

Learn

Explore tutorials, examples, and best practices

Read Documentation →

🤝

Support

Get help from the community and maintainers

Open an Issue →

Open Source

LatticeDB is released under the MIT License. Free to use in commercial and personal projects.

C++17 CMake React TypeScript Docker Terraform

Modern Database for AI & Analytics