VW Electronics and Information Technology, Volume: 1, Issue: 1, 04-11 (2019)
Received: Sept. 01, 2019
Accepted: Sept. 21, 2019
Published online: Sept. 23, 2019
Ibad ur Rahman1
Yaseen Muahmmad2
Shahzad Ahmed3
Muhammad Hanif 4*
1Militry College of Signals, NUST, Pakistan. ibaduet@gmail.com
2Univeristy of Engineering and Technology, Peshawar, Pakistan. yaseen@live.com
3Department of Electronics and Computer Engineering, Hanyang University, Seoul, South Korea shahzad1@hanyang.ac.kr
4Division of Computer Science and Engineering, Hanyang University, Seoul, South Korea
All authors contributed equally to this work
*Corresponding author email: honeykhan@hanyang.ac.kr
Abstract: In recent times, the Bulk Synchronous parallel (BSP) model, which splits the underlying running programs into multiple super-steps, has gain popularity in distributed graph processing systems. However, high number of message and data exchange in each super-step will create a long epoch of time, often called as network overhead or communication delays. In this paper, we add an additional step in the synchronization phase through decoupling the synchronization in to local and global synchronization. This approach move some of the data and messages exchange over the network to locally optimized pointers for the next super-step. We prototyped the proposed system and the preliminary evaluation shows decoupled mechanism outperforms baseline system on the average. The system help eliminate the unnecessary communication between the instances and localize some of the DAG nodes for the next super-step.
