Graph Database: Migration from Neo4j

Graph Database: Migration from Neo4j

Part of: Graph Database User Guide

---

Compatibility Matrix

Feature	Neo4j	HeliosDB	Notes
Cypher Query Language	Full	Core	Basic patterns, WHERE, RETURN
Bolt Protocol			Use Neo4j drivers
Graph Algorithms			PageRank, centrality, community
ACID Transactions			Snapshot isolation
Indexes			Label and property indexes
Constraints		⚠ Partial	Unique constraints planned
Full-text Search		🔄 Planned	V6.6
Spatial Queries		🔄 Planned	V6.7
Temporal Types			Via properties
User-Defined Procedures		🔄 Planned	V6.8

Legend: Supported | ⚠ Partial | 🔄 Planned | ❌ Not supported

Export from Neo4j

Using APOC Export

// Export to JSON
CALL apoc.export.json.all("export.json", {})

// Export to CSV
CALL apoc.export.csv.all("export.csv", {})

// Export specific subgraph
MATCH (p:Person)-[r:FRIEND_OF]->(f:Person)
CALL apoc.export.json.data([p, f], [r], "friends.json", {})
YIELD file, nodes, relationships
RETURN file, nodes, relationships

Using Cypher Shell

# Export to Cypher statements
cypher-shell -u neo4j -p password \
  "MATCH (n) RETURN n LIMIT 10000" > nodes.cypher

cypher-shell -u neo4j -p password \
  "MATCH ()-[r]->() RETURN r LIMIT 10000" > edges.cypher

Import to HeliosDB

Import from JSON

use serde_json::Value;
use std::fs::File;
use std::io::BufReader;

// Read Neo4j export
let file = File::open("export.json")?;
let reader = BufReader::new(file);
let data: Vec<Value> = serde_json::from_reader(reader)?;

// Import nodes
for item in data {
    if item["type"] == "node" {
        let node = Node {
            id: 0,
            label: item["labels"][0].as_str().unwrap().to_string(),
            properties: item["properties"].as_object().unwrap().clone().into_iter()
                .map(|(k, v)| (k, v.clone()))
                .collect(),
        };
        storage.add_vertex(node).await?;
    }
}

// Import relationships
for item in data {
    if item["type"] == "relationship" {
        let edge = Edge {
            id: 0,
            source: item["start"]["id"].as_u64().unwrap(),
            target: item["end"]["id"].as_u64().unwrap(),
            label: item["label"].as_str().unwrap().to_string(),
            weight: 1.0,
            properties: item["properties"].as_object().unwrap().clone().into_iter()
                .map(|(k, v)| (k, v.clone()))
                .collect(),
        };
        storage.add_edge(edge).await?;
    }
}

Bulk Import Tool

# Use HeliosDB bulk import tool
heliosdb-import \
  --format neo4j-json \
  --input export.json \
  --graph social_network \
  --batch-size 10000

Query Translation

Neo4j Cypher → HeliosDB

Most Cypher queries work as-is. Some differences:

-- Neo4j: CREATE with MERGE
MERGE (p:Person {name: 'Alice'})

-- HeliosDB: Use add_vertex API
-- (CREATE support in roadmap)

-- Neo4j: OPTIONAL MATCH
MATCH (p:Person)
OPTIONAL MATCH (p)-[:FRIEND_OF]->(f)
RETURN p, f

-- HeliosDB: Use LEFT JOIN pattern
-- (OPTIONAL MATCH planned for V6.6)

Performance Comparison

Based on benchmarks with 1M node, 5M edge graph:

Operation	Neo4j 5.x	HeliosDB 6.5	Speedup
6-degree BFS	245ms	78ms	3.1x faster
Shortest path (Dijkstra)	32ms	15ms	2.1x faster
PageRank (20 iter)	980ms	890ms	1.1x faster
Node lookup	1.2μs	0.4μs	3x faster
Bulk insert	180K/sec	500K/sec	2.8x faster
Memory usage	15.2GB	9.6GB	37% less

Conclusion: HeliosDB is 2-3x faster for most operations with lower memory footprint.

---

---

Navigation

• Previous: Integration
• Next: API Reference
• Index: Graph Database User Guide

---

Version: 6.5 Last Updated: November 17, 2025