Skip to content

Apache HugeGraph

Full-stack graph platform

Query relationships in real time, analyze whole graphs offline, and use Graph AI to build LLM memory and ontology graphs

Get Started GitHub
Graph AI Graph Toolchain Graph Computing Graph Database

Explore the layers

Start with one vertex

Follow an edge to a neighboring vertex. Use embedded storage on a single machine or Raft replication for highly available distributed storage

g.V().has('name', 'marko').out('knows') Explore database deployment

Follow an edge

Load vertices and edges with Loader, then explore the query results in Hubble, one relationship at a time

Explore the graph toolchain

Analyze the whole graph to discover communities

Graph computing runs OLAP over the whole dataset, with Vermeer in memory and Computer on a BSP cluster

Explore graph computing

Explore graphs with AI

Our graph extraction SDK, Memory, and ontology platform are evolving quickly. Join us in building Graph AI

Explore Graph AI

A full-stack graph system that carries your data from storage and real-time querying to offline analysis and AI

100s of billions
vertices and edges
Milliseconds
multi-hop query latency
30+
built-in graph algorithms
First
Apache top‑level graph system

Key features

Efficient

Batch import for massive datasets, millisecond-level retrieval and concurrent online operations for thousands of users

AI-Ready

GraphRAG, automated knowledge graph construction and 20+ built-in graph machine learning algorithms

Convenient

Gremlin, Cypher and a RESTful API, plus a complete toolchain for import, visualization and operations

Adaptable

Property graph modeling on Apache TinkerPop, covering OLTP and OLAP, with Hadoop and Spark integration

Scalable

Horizontal scaling and distributed deployment, from a standalone server to PB-level clusters

Open Ecosystem

TinkerPop standards, multi-language clients and an active, evolving Apache community

Works seamlessly with your existing stack

  • Apache Spark
  • Apache Flink
  • Apache Hadoop
  • Docker
  • Kubernetes
  • Helm
  • RocksDB
  • Python
  • Go
  • OpenJDK
ALL IN ONE

Deploy the database on one node or across a cluster. Add tools, graph computing and AI as needed

Graph Database

Server Start here
Core graph engine with REST API, Gremlin and Cypher
PD
Placement Driver for metadata, partitions and cluster scheduling
Store
Distributed storage with Raft consensus

Toolchain

Hubble
Web platform for graph visualization
Loader
Data import tool for bulk loading
Tools
Command-line utilities for backup, restore and migration

Computing and AI

Vermeer
In-memory graph algorithms for PageRank and community detection
AI
GraphRAG and knowledge graph construction with LLMs

Quick start

Docker starts HugeGraph Server in standalone mode by default on port 8080 of localhost. Standalone mode supports in-memory or disk storage

docker run -d -p 127.0.0.1:8080:8080 hugegraph/hugegraph

Other install options Verify setup and run your first query

Welcome to the HugeGraph open source community

New users and contributors are always welcome