GraphForge vs. ArcadeDB
Do you need multimodel storage with bundled graph analytics, or an investigation-oriented graph workflow?
ArcadeDB is an Apache-2.0 multimodel database. Graphs, documents, key-value, time-series, vector, and geospatial data share one engine, readable through SQL, openCypher, Gremlin, GraphQL, MongoDB, and Redis protocols. It bundles about 70 graph algorithms as Cypher procedures. GraphForge is narrower and runs inside your process: a property graph with 94 native algorithms, Arrow results, and an investigation history of checkpoints, provenance, and hypotheses.
On this page
Should I choose ArcadeDB or GraphForge?
ArcadeDB for many data models and wire protocols on the JVM; GraphForge for Python, Node.js, or Rust graph analysis.
Choose GraphForge when
- You work in Python, Node.js, or Rust rather than on the JVM.
- You want algorithm and query results as Apache Arrow tables for dataframes and notebooks.
- Your analysis needs checkpoints, provenance, and recorded competing hypotheses.
Choose ArcadeDB when
- One database must serve graph, document, key-value, time-series, and geospatial data together.
- Clients already speak PostgreSQL wire, Bolt, MongoDB, or Redis protocols and should not change.
- You run on the JVM and want either embedded or server deployment with Raft-based high availability.
What can each one do?
18 of 26 capabilities differ. For 7, ArcadeDB's docs did not confirm support, so they are marked Not established rather than No.
Each capability has a status and a source. “Not established” means we could not confirm the feature from ArcadeDB's primary documentation. It does not mean ArcadeDB lacks it.
| Capability | GraphForge | ArcadeDB |
|---|---|---|
| Storage | ||
| Property graph model | Shipped Labeled property graph with typed properties. 1 | Yes Vertices and edges as first-class records with properties; lightweight property-less edges available. 18 |
| Document model | Not offered No document-store model; properties accept nested, heterogeneous values. 1 | Yes Native document model sharing storage and transactions with graph data. 18 |
| Relational tables | Not offered Not a relational database; results arrive as Arrow tables. 1 | Partial SQL dialect and PostgreSQL wire protocol, but JOINs unsupported; relationships are links. 28 |
| Key-value access | Not offered No key-value API; data is a property graph. 1 | Yes Key/value access via Redis commands and wire protocol. 18 |
| Columnar storage | Shipped Durable projects store graph data as Arrow and Parquet. 4 | Partial Columnar time-series storage and columnar CSR graph analytical views; not general columnar tables. 1821 |
| Vector search | Shipped Exact cosine retrieval over embeddings you supply; no ANN index. 5 | Yes JVector HNSW or Vamana/DiskANN indexes; cosine, dot product, Euclidean; INT8/binary quantisation. 24 |
| Full-text search | Shipped Tantivy-backed BM25; hybrid text and vector retrieval through find(). 5 | Yes Built-in Lucene full-text engine integrated into query languages. 18 |
| Geospatial | Not offered No spatial types or indexes are documented. | Yes WKT geometries (points, polygons) with geo.* SQL functions and GeoHash-based index. 25 |
| Execution | ||
| Transactional workloads (OLTP) | Partial Local analytical scope; durable atomic writes with queued and optimistic writer modes. 6 | Yes MVCC ACID transactions over an LSM-tree storage engine with WAL. 2218 |
| Analytical workloads (OLAP) | Shipped Analytical verbs and Cypher over a local columnar project. 7 | Yes Graph Analytical Views keep a CSR read-optimized copy alongside live OLTP data. 21 |
| Graph traversal | Shipped Cypher MATCH patterns plus BFS, DFS, Dijkstra, A*, and reachability. 8 | Yes SQL MATCH/TRAVERSE, openCypher and Gremlin traversals over physical links. 18 |
| Variable-length paths | Shipped Bounded and unbounded patterns and shortestPath(). 8 | Yes Cypher *min..max quantifiers, honored by shortestPath() and allShortestPaths(). 26 |
| Graph algorithms | Shipped 94 native algorithms behind analysis verbs, not Cypher CALL procedures. 37 | Yes Bundled algo.* Cypher procedures plus SQL path functions. 20 |
| Incremental computation | Not offered Algorithms recompute on request; no incremental mode is documented. 7 | Partial Analytical views apply per-commit delta overlays (SYNCHRONOUS/ASYNCHRONOUS); algorithms rerun on demand. 21 |
| Deployment | ||
| Embedded / in-process | Shipped In-process library; no server, daemon, or connection string. 9 | Yes Link the JAR and run in-process in any JVM application. 18 |
| Server mode | Planned Optional remote peer extension planned after 0.6.0. 11 | Yes Server over HTTP, PostgreSQL wire, Bolt, MongoDB, Redis, gRPC and MCP. 18 |
| WebAssembly | Not planned A browser WebAssembly engine was declined in ADR 0027. 12 | Not established No WebAssembly build documented. |
| Distributed cluster | Not planned Distributed clusters and HA replication are out of scope. 9 | Partial Raft leader-replica HA with read scaling; single leader handles writes, no sharding documented. 23 |
| Managed cloud | Not planned No hosted compute. The graphforge.sh Hub distributes repositories; execution stays local. 13 | Not established No managed cloud service found in official docs. |
| Investigation | ||
| Branching | Experimental Research versions, branches, slices, proposals, and forks ship in 0.6.0 release candidates. 14 | Not established No database branching documented. |
| Snapshots / time travel | Shipped Named checkpoints you can diff and revert, plus transaction-time snapshots. 3 | Not established MVCC is internal; no point-in-time query documented. |
| Immutable history | Shipped Knowledge ledger records immutable assertions and evidence lineage. 16 | Not established No append-only history documented. |
| Provenance | Shipped Provenance and knowledge ledger with evidence lineage. 16 | Not established No engine-level provenance documented. |
| Competing hypotheses | Shipped Immutable hypothesis groups with membership and selection events. 3 | Not established |
| Operations | ||
| ACID transactions | Partial Acknowledged-durable atomic publication; docs do not claim general ACID or serializable isolation. 6 | Yes ACID with MVCC; READ_COMMITTED default, REPEATABLE_READ optional, no SERIALIZABLE. 22 |
| Replication | Not planned HA replication is out of scope. 9 | Yes Apache Ratis (Raft) log replication from leader to replicas with automatic election. 23 |
Showing 26 of 26 capabilities.
How is each one built?
ArcadeDB: multimodel + graph. GraphForge: an embedded property graph library in Rust, persisted as Parquet.
GraphForge
GraphForge is a Rust library that runs inside your process: no daemon, server, or connection string. Projects persist to a local directory as Arrow and Parquet data plus JSON metadata, and every query and analysis verb returns an Apache Arrow table. Algorithms run in the same native engine as Cypher and are called as API verbs (rank, cluster, paths, analyze, similar), not as Cypher procedures.
ArcadeDB
ArcadeDB is a Java engine that runs embedded in a JVM or as a server exposing HTTP, PostgreSQL wire, Bolt, MongoDB, Redis, gRPC, and MCP. Its six query languages read and write the same records. Graph Analytical Views keep a compressed sparse row snapshot of chosen vertex and edge types for algorithms, refreshed off, synchronously, or asynchronously. Vector search uses JVector, with HNSW and DiskANN-style indexes.
ArcadeDB lets one engine speak many data models and protocols. GraphForge does fewer things: one graph model and one query language, plus analysis verbs and investigation history built around them.
What the documentation says about ArcadeDB
- ArcadeDB accepts six query languages (SQL, Cypher, Gremlin, GraphQL, MongoDB QL, Redis) that read and write the same records.18
- ArcadeDB runs embedded in a JVM process or as a server reachable over HTTP, PostgreSQL wire, Bolt, MongoDB, Redis, gRPC and MCP.18
- ArcadeDB is Apache 2.0 licensed with no Enterprise edition gate.18
- Graph algorithms are invoked as CALL algo.<name>() in openCypher, while dijkstra(), bellmanFord() and shortestPath() are also SQL functions.20
- A Graph Analytical View stores a named Compressed Sparse Row snapshot of selected vertex/edge types; its update modes are OFF, SYNCHRONOUS or ASYNCHRONOUS.21
- High availability uses Apache Ratis (Raft): one leader coordinates writes, and replicas serve reads and forward writes to the leader.23
- Vector indexes use JVector 4.0.0, supporting both HNSW and Vamana (DiskANN) graphs.24
- Geometries are stored as WKT strings and indexed by decomposing them into GeoHash tokens stored in the LSM-tree.25
- Release 26.10.2 added a JVM-wide query admission gate across all remote protocols and Customizable Contraction Hierarchies for shortest paths.17
Which one is faster?
Not measured yet. No head-to-head run has been published, so this page shows no performance chart.
Charts appear only after a reproducible run exists: pinned versions, disclosed hardware, generated data, raw results, and scripts.
No comparative results yet
The suites below apply to GraphForge and ArcadeDB. Each will be reported separately per benchmark class, never merged into one leaderboard.
Ingestion and writes
| Suite | Workloads | Metrics | Status |
|---|---|---|---|
| INGEST | CSV, JSONL, Parquet, graph edge lists | Records/s, MB/s, total elapsed | Not run |
| MUTATE | Insert, update, delete, upsert | Ops/s, p50/p95/p99 latency | Not run |
Query, analytics, and system
| Suite | Workloads | Metrics | Status |
|---|---|---|---|
| LOOKUP | ID lookup, indexed property filter | p50/p95/p99 latency | Not run |
| TRAVERSE | BFS, k-hop, variable-length path | Latency, visited edges/s | Not run |
| JOIN | Multi-relation pattern matching | Latency, rows/s | Not run |
| ALGO | PageRank, WCC, SCC, Louvain, SSSP | Time, memory, correctness | Not run |
| HYBRID | Vector + graph + metadata filters | Recall@k, latency, throughput | Not run |
| EMBED | Startup, load, reopen, binary size | ms, MiB, disk footprint | Not run |
Which graph algorithms ship?
GraphForge: 94 native algorithms. ArcadeDB: 25 or more named in its docs, delivered as bundled.
Algorithms are listed by name, as documented. Counts of extension modules or third-party libraries are not presented as native coverage.
GraphForge
94 Rust algorithms run in the engine behind five analysis verbs (rank, cluster, paths, analyze, similar) and return Arrow tables.37
A selection of the catalog:
- PageRank
- ArticleRank
- Betweenness (Brandes)
- Closeness
- Harmonic closeness
- Eigenvector
- HITS
- CELF influence maximization
- Triangle count
- K-core decomposition
- Louvain
- Leiden
- Label propagation
- Infomap
- Walktrap
- Girvan-Newman
- Weakly connected components
- Strongly connected components
- Dijkstra
- A*
- Bellman-Ford
- Yen's k-shortest paths
- Max flow / min cut
- Steiner tree
- Minimum spanning tree
- Topological sort
- Node2Vec
- GraphSAGE
- Fast random projection
- Node similarity (Jaccard)
ArcadeDB
A bundled catalog of roughly 70 algorithms is exposed as openCypher procedures (CALL algo.<name>) plus SQL path functions, optionally accelerated by Graph Analytical Views.20
Delivered as: bundled. A selection of the catalog:
- PageRank (algo.pagerank)
- Personalized PageRank
- ArticleRank
- Betweenness Centrality
- Closeness Centrality
- Harmonic Centrality
- Eigenvector Centrality
- Katz Centrality
- HITS
- Louvain
- Leiden
- Label Propagation
- Weakly Connected Components
- Strongly Connected Components
- Triangle Count
- K-Core Decomposition
- Dijkstra
- A*
- Bellman-Ford
- K-Shortest Paths
- All-Pairs Shortest Paths
- Maximum Flow
- Minimum Spanning Tree
- Node2Vec
- FastRP
Where can each one run?
GraphForge runs in-process only. ArcadeDB: embedded yes, server yes, cluster partial.
| Mode | GraphForge | ArcadeDB |
|---|---|---|
| Embedded / in-process | Shipped In-process library; no server, daemon, or connection string. 9 | Yes Link the JAR and run in-process in any JVM application. 18 |
| Server mode | Planned Optional remote peer extension planned after 0.6.0. 11 | Yes Server over HTTP, PostgreSQL wire, Bolt, MongoDB, Redis, gRPC and MCP. 18 |
| WebAssembly | Not planned A browser WebAssembly engine was declined in ADR 0027. 12 | Not established No WebAssembly build documented. |
| Distributed cluster | Not planned Distributed clusters and HA replication are out of scope. 9 | Partial Raft leader-replica HA with read scaling; single leader handles writes, no sharding documented. 23 |
| Managed cloud | Not planned No hosted compute. The graphforge.sh Hub distributes repositories; execution stays local. 13 | Not established No managed cloud service found in official docs. |
| Replication | Not planned HA replication is out of scope. 9 | Yes Apache Ratis (Raft) log replication from leader to replicas with automatic election. 23 |
What will my team write?
GraphForge: openCypher from Python, Node.js, Rust. ArcadeDB: SQL (ArcadeDB dialect), openCypher, Gremlin, GraphQL (subset), MongoDB QL, Redis commands.
ArcadeDB developers can choose their language: SQL with link traversal, openCypher with algo.* procedures, Gremlin, or MongoDB-style queries. GraphForge developers write openCypher, which passes the full 2024.3 TCK, and call algorithms as Python, Node.js, or Rust verbs.
- Query languages
- GraphForge openCypher
- ArcadeDB SQL (ArcadeDB dialect), openCypher, Gremlin, GraphQL (subset), MongoDB QL, Redis commands
- SDKs and bindings
- GraphForge Python, Node.js, Rust
- ArcadeDB Java (embedded and remote), Python, JavaScript/TypeScript, Go
- Written in
- GraphForge Rust
- ArcadeDB Java
- Surfaces
- GraphForge Python, Node.js, Rust crates, CLI, VS Code extension, Notebooks, Agent skills
What are the trade-offs?
ArcadeDB has one write leader and no SERIALIZABLE isolation. GraphForge has one data model and no server.
A fair assessment cuts both ways.
GraphForge
- One data model; no document, key-value, time-series, or geospatial storage.
- No wire-protocol compatibility layers (PostgreSQL, Bolt, MongoDB, Redis).
- No server or high-availability mode.
ArcadeDB
- ArcadeDB does not support SQL SERIALIZABLE isolation; only READ_COMMITTED (default) and REPEATABLE_READ are offered.22
- ArcadeDB SQL does not support relational JOIN syntax; relationships are traversed as links with dot notation.28
- openCypher map-valued properties are refused, and 84 openCypher TCK scenarios (2.2%) are skipped as documented limitations.26
- An HA cluster has exactly one leader coordinating all writes, so write throughput does not scale out with replicas.23
- The Neo4j importer disables the WAL during import for throughput.19
How would I migrate?
Load ArcadeDB data through Arrow, pandas, or Polars. Standard openCypher carries over; procedure calls become verb calls.
Into GraphForge
- GraphForge 0.5.x loads data through Arrow, pandas, and Polars bulk calls or Cypher scripts; it has no native CSV loader.15
- Algorithms are API verbs (rank, cluster, paths, analyze, similar), not Cypher CALL procedures, so procedure calls become verb calls.7
- GraphForge passes the full openCypher 2024.3 TCK, so standard openCypher queries are the portable part of a migration.10
Moving from ArcadeDB
- The built-in importer reads OrientDB and Neo4j exports, GraphML, GraphSON, XML, JSON, JSONL, CSV and RDF, plain or ZIP/GZip compressed.27
- Neo4j data migrates via an APOC JSONL export loaded by the Neo4j importer.19
- Following Cypher 25/GQL, ArcadeDB rejects legacy syntax such as USING PERIODIC COMMIT and {param} parameters, so older Neo4j queries may need rewriting.26
- Neo4j Bolt drivers and PostgreSQL clients can connect directly, reducing driver changes when porting applications.18
What else do people ask about ArcadeDB?
Is ArcadeDB open source? ArcadeDB is Apache-2.0 with about 70 bundled algorithms; plain Cypher ports, procedure calls do not.
Is ArcadeDB open source?
Yes. ArcadeDB is Apache-2.0 with no enterprise edition gate. Paid support plans are optional. GraphForge is Apache-2.0 too.
How many graph algorithms does ArcadeDB have?
Its documentation lists roughly 70 algorithms exposed as CALL algo.<name>() procedures, plus SQL path functions such as dijkstra() and shortestPath(). GraphForge lists 94 native algorithms called as API verbs.
Can ArcadeDB and GraphForge share Cypher queries?
Plain openCypher patterns usually carry over. Procedure calls do not, since GraphForge has no CALL procedures, and ArcadeDB documents some skipped TCK scenarios and refuses map-valued properties.
How was this researched?
From 28 primary sources, accessed 2026-10-19, each cited where it is used.
Capabilities were checked against primary documentation, repositories, and release notes, accessed 2026-10-19. Each cell records two things: the capability’s state and the evidence behind it. Every GraphForge cell carries a release status. Shipped means stable 0.5.2; experimental means 0.6.0 release candidates only. GraphForge is pre-1.0 and makes no backward-compatibility guarantee yet.
- Yes documented natively
- Partial limited or caveated
- Extension official extension or module
- Paid edition paid or enterprise edition only
- Removed removed, with the version cited
- Not established not confirmed by our research; not a “no”
- Shipped in stable 0.5.2
- Partial shipped with documented limits
- Experimental in 0.6.0 release candidates only
- Planned on the roadmap
- Not offered outside the product's model
- Not planned explicitly out of scope
Benchmark classes
- Equivalent workloads. The same logical operation with the same correctness requirement on every engine.
- Native optimized workloads. Each engine's best supported implementation, including projections, indexes, extensions, and native algorithms.
- End-to-end application workloads. Time from raw data to a usable result: import, indexing, projection, computation, and output retrieval.
Sources · accessed 2026-10-19
- GraphForge repository README — GitHub
- GraphForge releases — GitHub
- GraphForge 0.5.0 changelog — GitHub
- Storage architecture — GraphForge Book
- Analytics integration: search and retrieval — GraphForge Guide
- ADR 0018: acknowledged durability and isolation — GitHub
- Algorithm verbs — GraphForge Book
- Cypher guide — GraphForge Guide
- openCypher compatibility — GraphForge Reference
- openCypher TCK compliance — GraphForge Reference
- Remote authority peer extension (#215) — GitHub
- ADR 0027: native runtime boundary — GitHub
- Release scope and roadmap — GraphForge Docs
- Keep research history — GraphForge Guide
- Graph construction — GraphForge Guide
- Knowledge public API v1 — GraphForge Book
- Release 26.10.2 — ArcadeData/arcadedb GitHub
- What is ArcadeDB — ArcadeDB Docs
- Neo4j Importer — ArcadeDB Docs
- Graph Algorithms — ArcadeDB Docs
- Graph OLAP Engine — ArcadeDB Docs
- Transactions — ArcadeDB Docs
- High Availability — ArcadeDB Docs
- Vector Search — ArcadeDB Docs
- Geospatial Index — ArcadeDB Docs
- OpenCypher Compatibility — ArcadeDB Docs
- Importer — ArcadeDB Docs
- SQL Introduction — ArcadeDB Docs
Try GraphForge on your own graph.
Install the package, open a project, and run the same notebook we measure with.