Neo4j¶

Neo4j is the world's leading graph database platform with a mature ecosystem and enterprise features.

Overview¶

Property	Value
Type	Client-Server Graph Database
Language	Java (core), Cypher (query)
License	AGPL v3 (Community) / Commercial
Architecture	Network service
Query Language	Cypher
Website	neo4j.com
GitHub	github.com/neo4j/neo4j

Performance Summary¶

### Overall Score: **87.5/100** 🥈 | Metric | Score | Rank | |--------|-------|------| | Ingestion | 🥈 2nd | 95K nodes/s | | Traversal | 🥈 2nd | 1.2ms avg | | PageRank | 🥈 2nd | 2.1s (10 iter) | | Memory | 5th | 380 MB |

Key Characteristics¶

Strengths¶

Mature Ecosystem - 15+ years of development
GDS Library - Graph Data Science algorithms
Visualization Tools - Neo4j Desktop, Bloom
Enterprise Features - Backup, clustering, security
Cypher Standard - Original creator of Cypher

Weaknesses¶

High Memory Usage - 88% more than CongraphDB
Slow Startup - ~2 seconds to initialize
Network Overhead - Client-server latency
Complex Setup - Requires separate server process
Heavyweight - Java runtime + page cache

Performance Analysis¶

Ingestion Performance¶

Scale	Nodes/s	vs CongraphDB
Small (10K)	95,000	-24%
Medium (100K)	92,000	-22%
Large (1M)	88,000	-20%

Why slower: - Network protocol overhead - Transaction log writes - Index maintenance during ingestion - Java object allocation

Traversal Performance¶

Hops	Small	Medium	Large
1-hop	0.20ms	0.22ms	0.25ms
2-hop	0.50ms	0.55ms	0.65ms
3-hop	1.20ms	1.35ms	1.55ms
4-hop	2.80ms	3.20ms	3.80ms
5-hop	6.50ms	7.80ms	9.20ms

Why slower: - Network round-trip per query - Query planning overhead - Page cache misses - Cypher execution engine

Memory Usage¶

Scale	Peak Memory	vs CongraphDB
Small (10K)	380 MB	+744%
Medium (100K)	1,850 MB	+381%
Large (1M)	8,200 MB	+152%

Why high memory: - Java runtime (~250MB base) - Page cache (configurable, ~1.5GB default) - Object overhead in Java heap - Index structures in memory

When Neo4j Makes Sense¶

Choose Neo4j For¶

✅ Complex Analytics - GDS library with 60+ algorithms
✅ Multi-Language - Drivers for all major languages
✅ Enterprise Features - Backup, HA, security
✅ Team Collaboration - Multiple users accessing same DB
✅ Visualization - Neo4j Bloom for visual queries
✅ Existing Investment - Team already knows Neo4j

Avoid Neo4j For¶

❌ Embedded Use - Need database in your app
❌ Low Memory - Serverless or edge computing
❌ Fast Startup - CI/CD or testing scenarios
❌ Single User - Overkill for single-developer projects

GDS Library¶

The Graph Data Science (GDS) library is Neo4j's killer feature:

CALL gds.pageRank.stream('my-graph')
  YIELD nodeId, score
RETURN gds.util.asNode(nodeId).name AS name, score
ORDER BY score DESC
LIMIT 10

Available algorithms: - Centrality - PageRank, Betweenness, Eigenvector - Community Detection - Louvain, Label Propagation - Pathfinding - Dijkstra, A, Yen - Similarity - Node similarity, Jaccard - ML* - Node classification, link prediction

Note: GDS requires additional memory and licensing for production.

Setup Comparison¶

CongraphDB¶

npm install congraphdb
# Ready in < 1 second

Neo4j¶

# Download Neo4j
wget https://dist.neo4j.org/neo4j-community-5.0.0-unix.tar.gz
tar -xzf neo4j-community-5.0.0-unix.tar.gz

# Configure
echo "dbms.default_listen_address=0.0.0.0" >> neo4j.conf

# Start server
./bin/neo4j start

# Wait for startup (~2-30 seconds)
# Set password at http://localhost:7474

# Install Node.js driver
npm install neo4j-driver

Query Comparison¶

Same query in both databases:

CongraphDB¶

const result = conn.execute(`
  MATCH (p:Paper)-[:CITES]->(cited:Paper)
  WHERE p.year >= 2020
  RETURN cited.id, count(*) AS citations
  ORDER BY citations DESC
  LIMIT 10
`);

Neo4j¶

const session = driver.session();
const result = await session.run(`
  MATCH (p:Paper)-[:CITES]->(cited:Paper)
  WHERE p.year >= 2020
  RETURN cited.id AS id, count(*) AS citations
  ORDER BY citations DESC
  LIMIT 10
`);

Difference: Neo4j requires network round-trip; CongraphDB is direct function call.

Resources¶

Official Documentation
GDS Library
Neo4j Aura - Managed service
Neo4j Desktop - GUI management

License¶

Community Edition: AGPL v3 (open source, copyleft)
Enterprise Edition: Commercial license

Summary¶

Neo4j is the industry standard for good reason - mature ecosystem, enterprise features, and powerful analytics. However, for embedded Node.js applications requiring performance and low memory, CongraphDB offers compelling advantages.

Recommendation: Use Neo4j if you need its advanced features; use CongraphDB for embedded, high-performance scenarios.