PowerGraph: distributed graph-parallel computation on natural graphs

TLDR

Large‑scale graph‑structured computation underpins tasks such as targeted advertising and natural language processing, and has spurred graph‑parallel abstractions like Pregel and GraphLab, but natural power‑law graphs challenge these abstractions’ assumptions, limiting performance and scalability. This work characterizes the challenges of computing on natural power‑law graphs within existing graph‑parallel frameworks and proposes the PowerGraph abstraction to exploit graph program structure and overcome these challenges. PowerGraph introduces a distributed graph placement and representation scheme that leverages power‑law structure, and the authors analyze and experimentally compare it to two popular graph‑parallel systems. Empirical evaluations on large‑scale real‑world problems demonstrate that PowerGraph achieves order‑of‑magnitude performance gains over existing systems.

Abstract

Large-scale graph-structured computation is central to tasks ranging from targeted advertising to natural language processing and has led to the development of several graph-parallel abstractions including Pregel and GraphLab. However, the natural graphs commonly found in the real-world have highly skewed power-law degree distributions, which challenge the assumptions made by these abstractions, limiting performance and scalability.In this paper, we characterize the challenges of computation on natural graphs in the context of existing graph-parallel abstractions. We then introduce the PowerGraph abstraction which exploits the internal structure of graph programs to address these challenges. Leveraging the PowerGraph abstraction we introduce a new approach to distributed graph placement and representation that exploits the structure of power-law graphs. We provide a detailed analysis and experimental evaluation comparing PowerGraph to two popular graph-parallel systems. Finally, we describe three different implementation strategies for PowerGraph and discuss their relative merits with empirical evaluations on large-scale real-world problems demonstrating order of magnitude gains.