Why it matters

Graph processing has specific patterns. GraphX enables at Spark scale. Understanding shapes graph analytics.

Advertisement

The architecture

Graph: VertexRDD + EdgeRDD.

Pregel: iterative message passing.

GraphX modelVertexRDDid + attributeEdgeRDDsrc + dst + attrPregel iterationsmessage passingAlternative: GraphFrames (DataFrame-based graph analytics)
GraphX components.
Advertisement

How it works end to end

Pregel: superstep-based BSP. Vertices process incoming messages, send new ones.

Built-in algorithms: PageRank, triangleCount, connectedComponents.

Alternative: GraphFrames (DataFrame-based).