Publication Type

Conference Proceeding Article

Version

acceptedVersion

Publication Date

12-2017

Abstract

Recent studies showed that single-machine graph processing systems can be as highly competitive as clusterbased approaches on large-scale problems. While several outof-core graph processing systems and computation models have been proposed, the high disk I/O overhead could significantly reduce performance in many practical cases. In this paper, we propose GraphMP to tackle big graph analytics on a single machine. GraphMP achieves low disk I/O overhead with three techniques. First, we design a vertex-centric sliding window (VSW) computation model to avoid reading and writing vertices on disk. Second, we propose a selective scheduling method to skip loading and processing unnecessary edge shards on disk. Third, we use a compressed edge cache mechanism to fully utilize the available memory of a machine to reduce the amount of disk accesses for edges. Extensive evaluations have shown that GraphMP could outperform state-of-the-art systems such as GraphChi, X-Stream and GridGraph by 31.6x, 54.5x and 23.1x respectively, when running popular graph applications on a billion-vertex graph.

Keywords

Graph Processing, Big Data, Parallel Computing

Discipline

Computer and Systems Architecture | Software Engineering

Research Areas

Software and Cyber-Physical Systems

Publication

Proceedings of the 23rd International Conference on Parallel and Distributed Systems (ICPADS): 2017 IEEE, Shenzhen, China, December 15-17

First Page

1

Last Page

10

ISBN

1521-9097

Identifier

10.1109/ICPADS.2017.00045

Publisher

IEEE

City or Country

Shenzhen, China

Additional URL

https://doi.org/10.1109/ICPADS.2017.00045

Share

COinS