An SQL-Based Query Language and Engine for Graph Pattern Matching

The interest for graph databases has increased in the recent years. Several variants of graph query languages exist - from low-level programming interfaces to high-level, declarative languages. In this paper, we describe a novel SQL-based language for modeling high-level graph queries. Our approach is based on graph pattern matching concepts, specifically nested graph conditions with distance constraints, as well as graph algorithms for calculating nested projections, shortest paths and connected components. Extending SQL with graph concepts enables the reuse of syntax elements for arithmetic expressions, aggregates, sorting and limits, and the combination of graph and relational queries. We evaluate the language concepts and our experimental SAP HANA Graph Scale-Out Extension (GSE) prototype (This paper is not official SAP communication material. It discusses a research-only prototype, not an existing or future SAP product. Any business decisions made concerning SAP products should be based on official SAP communication material.) using the LDBC Social Network Benchmark. In this work we consider only complex read-only queries, but the presented language paves the way for a SQL-based graph manipulation language formally based on graph transformations.