ArangoDB Multi-Model Database

ArangoDB is a free and open-source native multi-model database that combines the capabilities of document, graph, and key-value stores in a single database engine with a unified query language. Developed by ArangoDB Inc. (founded in 2012 by Claudius Weinberger and Frank Celler, headquartered in San Francisco and Cologne, Germany), ArangoDB eliminates the need for multiple specialized databases by providing all three data models in one product. Key features: multi-model: documents are stored as JSON-like objects (VelocyPack binary format) in collections, the same as document databases like MongoDB. Relationships between documents are stored as edges in graph collections, enabling graph traversals. Key-value access is provided as a subset of document access. AQL (ArangoDB Query Language): a declarative query language that unifies document, graph, and key-value operations in a single query. AQL supports graph traversal (FOR v, e, p IN 1..3 OUTBOUND startNode graphName), joins, aggregations, full-text search, and geospatial queries. Sharding: horizontal scaling via automatic or manual sharding of collections across a cluster of DB-Servers, with consistent hashing for even distribution. Replication: synchronous replication within a cluster for high availability, and asynchronous replication for cross-datacenter replication. SmartGraphs and EnterpriseGraphs: optimize graph traversals in sharded clusters by co-locating connected graph data on the same server. Transactions: supports multi-document ACID transactions across collections. Indexes: primary, edge, hash (exact match), skiplist (sorted range), TTL, persistent (rocksdb), full-text (BM25 via ArangoSearch), and geo-spatial indexes. Foxx: run JavaScript microservices directly inside the database, similar to stored procedures. RocksDB storage engine for persistence. C++ codebase. Cross-platform. Apache-2.0.