// the find
ArcadeData/arcadedb
ArcadeDB Multi-Model Database, one DBMS that supports SQL, Cypher, Gremlin, HTTP/JSON, MongoDB and Redis. ArcadeDB is a conceptual fork of OrientDB, the first Multi-Model DBMS. ArcadeDB supports Vector Embeddings.
ArcadeDB is a multi-model database engine that handles graphs, documents, key-value, time-series, and vector data in a single process, written by the original OrientDB author. It's aimed at teams who need multiple data models without running multiple databases, and who are willing to bet on a small-team project over Postgres+Neo4j+Pinecone. Active daily commits and ~940 stars suggest it's past proof-of-concept but well short of widely-adopted.
70+ built-in graph algorithms (PageRank, Louvain, betweenness, etc.) available via SQL without a separate analytics layer — that's genuinely rare. The wire protocol compatibility story is real: you can point a Postgres driver, a Redis client, or a Neo4j Bolt client at it and it works, which lowers migration friction significantly. Low-level Java with explicit GC pressure management means the performance-per-core numbers hold up on cheap hardware. The HA resilience test suite (network partitions, packet loss, split-brain, rolling restart) is unusually thorough for a project this size.
OrientDB's history of breaking changes and eventually going effectively abandoned should make anyone cautious about a spiritual successor with one primary author. The 'vector support' is there but the docs are thin on HNSW vs. flat index tradeoffs and production-scale benchmarks are self-reported. MongoDB and Redis compatibility is subset-only — the README says so — which is fine until a dependency uses a command that isn't implemented. At 940 stars it has a small ecosystem: limited third-party tooling, sparse Stack Overflow coverage, and if the core team has a bad quarter you're reading Java source to debug your production issue.