Iterative graph processing

Gelly enhances Flink's iterative processing capabilities to support large scale graph processing. Currently it supports implementation of the following models:

  • Vertex-Centric
  • Scatter-Gather
  • Gather-Sum-Apply

Let's start by understanding these models in the context of Gelly.

Vertex-Centric iterations

As the name suggest, these iterations are built thinking the vertex is in the center. Here each Vertex processes the same user-defined function in parallel. Each step of execution is called a superset. A vertex can send a message to another vertex as long as it knows its unique ID. This message would be used as input to the next superset.

To use Vertex-Centric iterations, the user needs to provide a ComputeFunction. We can also define ...

Get Learning Apache Flink now with O’Reilly online learning.

O’Reilly members experience live online training, plus books, videos, and digital content from 200+ publishers.