DISC
Distributed graph analytics system that decomposes subgraph counting into relational queries.
DISC is a specialized graph system on Spark for subgraph query processing, especially subgraph counting, expressed in a relational way.
Unlike graph systems that operate directly on graph-native operators, DISC decomposes subgraph counting into a sequence of relational queries, which lets the system reuse relational optimization and distributed execution techniques.
The project showed that relational execution can be a practical route for large-scale graph workloads when the decomposition strategy is designed carefully.
Links: Code ยท VLDB 2020 paper