Parallel Implementation of Single-Source Shortest Path Algorithm Based on Haloop

Abstract

As a typical problem of graph theory, Single-Source Shortest Path(SSSP) has a wide range of applications and research. The traditional MapReduce framework such as Hadoop has been applied to SSSP problem. However, in this way, it may need writing and reading the disk frequently, transmitting data largely, beacuse of SSSP’s iterative. Haloop is a parallel programming framework which makes an improvement based on Hadoop framework, and adapts itself to iterative programming. Hence, this article represents the SSSP problem with an iterative way firstly, and then we put forward the implementation mechanism of SSSP algorithm based on Haloop. Though testing and analyzing, the implementation based on Haloop obviously improves the efficiency of program execution, compared with the traditional realization mechanism.