PROVIDER Β· Chinese distributed graph database
NebulaGraph
NebulaGraph is an open source distributed graph database from China whose separated storage and compute handles hundreds of billions of elements with nGQL openCypher compatibility and strong Chinese documentation for compliance-driven and massive knowledge graph projects.
3.8pts
Pros
Cons
Product Information
Detailed Notes
Overview
NebulaGraph was open sourced in 2019 by Hangzhou vesoft and written in C++ with separated storage and compute plus sharding, storing and querying hundreds of billions of vertices and edges through its nGQL language with openCypher compatibility, among the largest open source graph projects from China.
For Backend & Database work the advantages are two-fold, first extreme scale and horizontal expansion as a shared-nothing architecture with Raft consistency grows capacity and throughput by adding nodes, second domestic and Chinese-language ecosystem support with documentation, community, and local vendors responding far faster than overseas products, fitting compliance-driven and intranet deployments.
The trade-offs are ecosystem and migration, because nGQL differs from Cypher so Neo4j applications need query rewrites, client and algorithm libraries are fewer than Neo4j's, and enterprise capabilities such as visualisation and graph compute platforms need licences.
Core Strengths
- Separated storage and compute with sharding stores and queries hundreds of billions of vertices and edges
- Raft consensus keeps replicas consistent with automatic failover
- nGQL supports openCypher so common graph syntax works directly
- Complete Chinese documentation and local support meet domestic compliance and intranet requirements
Product Ecosystem
Storage and query architecture
Meta, Graph, and Storage services split duties with stateless GraphD scaling out and StorageD holding shards.
nGQL and ecosystem
openCypher-compatible syntax, LOOKUP index queries, and MATCH graph patterns with Python, Java, and Go clients.
Supporting products
NebulaGraph Studio for visualisation, Dashboard for monitoring, and Explorer for graph exploration.
Drawbacks
- nGQL differs from Cypher, so Neo4j migrations rewrite queries
- Fewer clients, algorithm libraries, and third-party tools than Neo4j
- Complex algorithms need a separate graph compute platform rather than running in the database
- Component count feels heavy for small single-machine deployments
Use Cases
- Very large knowledge graphs (β β β β β ) - Horizontal scale to hundreds of billions of edges
- Domestic compliance and intranet deployment (β β β β β ) - Open source from China with Chinese support
- Risk relationship network analysis (β β β β ) - Multi-hop queries with indexes
- Needs a rich algorithm library (β β β ) - Pair with a graph compute platform
Pricing
| Plan | Price | Scope |
|---|---|---|
| Open source | $0 | Apache 2.0, self-managed |
| Enterprise | Annual per node | Visualisation, graph compute, support |
| Managed cloud | Per instance size | Hosted with backup |
The open source edition is free with cost in servers and staff, since production wants at least three machines for the three services, while enterprise fees buy visualisation, graph compute, and support, and marketplaces offer per-size managed instances.
FAQ
-
NebulaGraph or Neo4j? NebulaGraph beyond a billion elements needing domestic sourcing and Chinese support, Neo4j for the mature Cypher ecosystem with turnkey algorithms and visualisation.
-
Does it run Neo4j Cypher directly? openCypher syntax works broadly, while APOC procedures, proprietary functions, and some aggregation semantics need rewriting, so run statement-level regression before migrating.
-
Can it run on a single machine? Yes via Docker Compose for development, though the meta, graph, and storage services mean one machine does not show the architectural benefit.