JanusGraph/janusgraph
Janusgraph
JanusGraph is a distributed, elastically scalable graph database designed to store and query highly connected data across a cluster of machines. It supports the property graph data model with ACID consistency and integrates multi-model search capabilities including geo, numeric range, and full-text queries. The database also includes a Graph OLAP engine for running batch analytics and global graph computations on large datasets using the Hadoop framework.
The project distinguishes itself through a masterless cluster architecture that eliminates single points of failure, allowing every node to handle reads and writes without a central coordinator. It provides elastic storage scaling by adding or removing machines without downtime, and features a pluggable backend storage layer that decouples the graph engine from the underlying store. Multi-modal index integration combines graph traversal with external search indexes such as Elasticsearch, Solr, and Lucene, while vertex-centric indexing addresses the super-node problem by indexing edges at the vertex level for fast local lookups.
The system supports real-time traversal of large graphs with horizontally scalable transactional capacity, and includes an elastic cache layer that distributes frequently accessed vertices and edges across cluster memory. It offers interactive graph visualization through built-in web tools and third-party integrations, and provides multi-criteria graph search across geo, numeric range, and full-text dimensions. The documentation covers installation, configuration, and operational guidance for deploying and managing the database across distributed environments.
Features
- Graph Databases - An open-source graph database that stores and queries highly connected data across a cluster of machines with ACID consistency.
- Graph Batch Analytics - Runs full-graph processing jobs as MapReduce or Spark tasks on a Hadoop cluster for offline computation.
- Pluggable Storage Backends - Decouples the graph engine from the underlying store so any key-value or document database can serve as the persistence layer.
- Distributed Real-Time Traversals - Answering complex traversal queries on huge graphs in milliseconds with horizontally scalable transactional capacity.
