Neo4j vs PostgreSQL

Neo4j is a graph-native database; PostgreSQL is a relational database that can handle graph queries through recursive CTEs. The choice depends on how central relationships are to your data model.

Architecture Differences

Neo4j uses index-free adjacency: each node physically stores pointers to its neighbors. This means traversing a relationship is a constant-time pointer hop, regardless of total data size. For deep traversals (6+ hops), this architectural difference provides orders-of-magnitude performance advantage.

PostgreSQL stores relationships as foreign keys between tables. Traversing relationships requires joining tables, and each additional hop adds a join operation. Recursive CTEs can traverse arbitrary depth, but performance degrades as depth increases.

Performance Characteristics

For a "friends of friends of friends" query across millions of users, Neo4j completes in milliseconds while PostgreSQL's recursive CTE may take seconds or even time out. This difference is the fundamental argument for graph databases in social network system design.

For queries that filter primarily on entity attributes (WHERE clauses on columns), PostgreSQL typically outperforms Neo4j. The sweet spot for Neo4j is pattern-matching queries that combine relationship traversal with property filters.

Trade-offs

Neo4j excels at its niche but is a specialized tool. You typically need another database alongside it for non-graph workloads. PostgreSQL can handle light graph workloads adequately, avoiding the operational complexity of managing two databases.

Neo4j's Graph Data Science library provides built-in algorithms for PageRank, community detection, shortest path, and centrality calculations. Implementing these in SQL would be impractical.

Graph Use Cases in System Design

In system design interviews, graph databases appear in social network design, recommendation engines, fraud detection (finding suspicious relationship patterns), and knowledge graphs. Understanding when graph traversal performance matters is key.

Real-World Usage

Neo4j powers eBay's product recommendations, NASA's knowledge management, and Walmart's supply chain optimization. PostgreSQL handles graph-like queries at companies that prefer operational simplicity over graph-native performance.

Learn about graph algorithms and data modeling patterns. See our interview questions and pricing.