DBRS: Directed Acyclic Graph based Reliable Scheduling Approach in Large Scale Computing

Abstract

In large scale environments, scheduling presents a significant challenge because it is an NP-hard problem. There are basically two types of task in execution- dependent task and independent task. The execution of dependent task must follow a strict order because output of one activity is typically the input of another. In this paper, a reliable fault tolerant approach is proposed for scheduling of dependent task in large scale computing environments. The workflow of dependent task is represented with the help of a DAG (directed acyclic graph). The proposed methodology is evaluated over various parameters by applying it in a large scale computing environment- ‘grid computing’. Grid computing is a high performance computing for solving complex, large and data intensive problems in various fields. The result analysis shows that the proposed DAG based reliable scheduling (DBRS) approach increases the performance of system by decreasing the makespan, number of failures and increasing performance improvement ratio (PIR).