ABC: Aggregation before Communication, a Communication Reduction Framework for Distributed Graph Neural Network Training and Effective Partition

TL;DR

This paper introduces ABC, a communication reduction framework for distributed GNN training that leverages aggregation before communication, improving efficiency especially for dynamic graphs.

Contribution

The paper proposes a novel lossless communication reduction method, ABC, exploiting permutation-invariance to improve distributed GNN training efficiency and partitioning strategies.

Findings

ABC reduces communication overhead in distributed GNN training.

Vertex-cut partitioning outperforms edge-cut in communication efficiency.

The method is especially effective for dynamic graphs with stochastic changes.

Abstract

Graph Neural Networks(GNNs) are a family of neural models tailored for graph-structure data and have shown superior performance in learning representations for graph-structured data. However, training GNNs on large graphs remains challenging and a promising direction is distributed GNN training, which is to partition the input graph and distribute the workload across multiple machines. The key bottleneck of the existing distributed GNNs training framework is the across-machine communication induced by the dependency on the graph data and aggregation operator of GNNs. In this paper, we study the communication complexity during distributed GNNs training and propose a simple lossless communication reduction method, termed the Aggregation before Communication (ABC) method. ABC method exploits the permutation-invariant property of the GNNs layer and leads to a paradigm where vertex-cut is proved to admit a superior communication performance than the currently popular paradigm (edge-cut). In addition, we show that the new partition paradigm is particularly ideal in the case of dynamic graphs where it is infeasible to control the edge placement due to the unknown stochastic of the graph-changing process.