FalkorDB

FalkorDB is a property graph database with a linear algebra-based execution engine. It leverages GraphBLAS to store graph structures as sparse adjacency matrices, executing OpenCypher queries through matrix operations instead of traditional traversals, enhanced by AVX SIMD instructions and OpenMP parallelism. The system runs as a Redis Module (.so shared library) requiring Redis 7.4+, with native multi-tenant graph-level isolation, full-text search, vector similarity search, and range indexing, accessible via both RESP and Bolt protocols.

Core Architecture#

┌─────────────────────────────────────────────┐
│  Protocol Layer: RESP (Redis) / Bolt          │
├─────────────────────────────────────────────┤
│  Query Engine: OpenCypher Parser              │
│     ↓                                         │
│  Execution Backend: GraphBLAS Linear Algebra  │
│     ↓                                         │
│  Storage Engine: Sparse Adjacency Matrix      │
├─────────────────────────────────────────────┤
│  Index Layer: Full-Text / Vector / Range      │
├─────────────────────────────────────────────┤
│  Multi-Tenant: Graph-level Namespace Isolation│
├─────────────────────────────────────────────┤
│  Persistence: Redis RDB/AOF                   │
└─────────────────────────────────────────────┘

Graph Storage & Query Engine#

• Sparse Matrix Representation: First queryable property graph database using sparse adjacency matrices for optimized storage and query performance
• Linear Algebra Execution: GraphBLAS-based, replacing traditional traversals with matrix operations
• AVX SIMD Acceleration: Performance optimization via AVX instruction set
• OpenCypher Compatible: OpenCypher query language support with proprietary extensions
• Property Graph Model: Full support for attributed nodes and relationships

Multi-Tenancy & Deployment#

• Native Multi-Tenancy: Graph-level isolation with independent namespaces, zero overhead
• Horizontal Scaling: Linear scalability
• High Availability: Cluster deployment, multi-region support
• Persistence & Backup: Redis RDB/AOF + cloud service 12-hour auto-backup
• Security: TLS, VPC, graph-level access control
• Dual Protocol: RESP (Redis) + Bolt

GenAI / GraphRAG#

• GraphRAG SDK: Built-in toolkit for advanced graph reasoning and generative AI tasks
• Auto Ontology Detection: Automatic ontology generation from unstructured data
• Built-in Agent Orchestration: Native agent orchestration capabilities
• Vector Search Integration: Native vector-graph hybrid retrieval

Quick Start#

Docker Deployment (Recommended)#

docker run -p 6379:6379 -p 3000:3000 -it --rm -v ./data:/var/lib/falkordb/data falkordb/falkordb

• Port 6379: FalkorDB Server (RESP protocol)
• Port 3000: FalkorDB Browser (Web UI)

Source Compilation#

git clone --recurse-submodules -j8 https://github.com/FalkorDB/FalkorDB.git
apt-get install build-essential cmake m4 automake peg libtool autoconf python3 python3-pip libgomp1
make
# Output: bin/<arch>/src/falkordb.so

Load as Redis Module#

redis-server --loadmodule /path/to/module/src/falkordb.so

Prerequisite: Redis 7.4+

Usage Example#

Python Client#

from falkordb import FalkorDB

db = FalkorDB(host='localhost', port=6379)
g = db.select_graph('MotoGP')
g.query("""CREATE
           (:Rider {name:'Valentino Rossi'})-[:rides]->(:Team {name:'Yamaha'}),
           (:Rider {name:'Dani Pedrosa'})-[:rides]->(:Team {name:'Honda'})""")

res = g.query("""MATCH (r:Rider)-[:rides]->(t:Team)
                 WHERE t.name = 'Yamaha' RETURN r.name""")
for row in res.result_set:
    print(row[0])

Use Cases#

• GraphRAG / AI Retrieval Augmentation: Combining LLMs with domain knowledge graphs to reduce hallucinations
• Agentic AI: Memory and reasoning support for AI Agents
• Fraud Detection: Graph traversal for complex relationship chain analysis
• Security Graphs: Threat intelligence queries (Case: Securin 7-hop query in 350ms)
• Snowflake Integration: Native graph database in Snowflake AI Data Cloud

Unconfirmed Items#

• GraphBLAS implementation version details
• Benchmark methodology and datasets unverified
• Securin case deployment scale specifics
• Snowflake integration availability and implementation
• SSPLv1 compliance implications
• OpenCypher compatibility coverage (CTP pass rate)
• GraphRAG SDK supported LLM providers and embedding models
• Bolt protocol version number