Relational-Core Graph Analytics Querying graphs at SQL scale, and why the node/edge model is a performance tax, not a truer picture of connected data
The paper argues graph analytics can run faster by translating graph queries onto existing relational tables instead of rebuilding data as nodes and edges.
Gene Zhang presents ClickGraph and DeltaGraph, which convert Cypher into SQL over native schemas and run on ClickHouse, Databricks, or lakehouse files. The claim is that property-graph modeling adds overhead because relationships already exist in tables, columns, and foreign keys. The abstract says benchmarks show a columnar engine beating Neo4j by two to four orders of magnitude, with additional measurements across LDBC’s Social Network Benchmark suite. Source: ArXiv · AI/CL/LG's note.
Gene Zhang presents ClickGraph and DeltaGraph, which convert Cypher into SQL over native schemas and run on ClickHouse, Databricks, or lakehouse files. The claim is that property-graph modeling adds overhead because relationships already exist in tables, columns, and foreign keys. The abstract says benchmarks show a columnar engine beating Neo4j by two to four orders of magnitude, with additional measurements across LDBC’s Social Network Benchmark suite. Source: ArXiv · AI/CL/LG's note.
score 5