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.

What it is

DISC decomposes subgraph counting into a sequence of relational queries. This lets the system reuse relational optimization and distributed execution techniques instead of relying only on graph-native operators.

Key ideas

  • Relational formulation of local subgraph counting.
  • Distributed execution over Spark for large-scale graph workloads.
  • General support for pattern graphs and orbits beyond narrow special cases.