WGB: Towards a universal graph benchmark

Abstract

Graph data are of growing importance in many recent applications. There are many systems proposed in the last decade for graph processing and analysis. Unfortunately, with the exception of RDF stores, every system uses different datasets and queries to assess its scalability and efficiency. This makes it challenging (and sometimes impossible) to conduct a meaningful comparison. Our aim is to close this gap by introducing Waterloo Graph Benchmark (WGB), a benchmark for graph processing systems that offers an efficient generator that creates dynamic graphs with properties similar to real-life ones. WGB includes the basic graph queries which are used for building graph applications.