CDAP: A Cultural Algorithm for Data Placement in Big Data Workflows

The performance of executing big data workflows in the Cloud highly depends on the placement of the workflow original datasets. An optimal data placement reduces the data movement among virtual machines, and as a result, it significantly reduces the workflow makespan. Data placement is an NP-hard problem; therefore in this paper, we propose CDAP (Cultural algorithm Data Placement), a novel metaheuristic data placement strategy based on Cultural Algorithms (CA) to improve the performance of a workflow by minimizing the data movement among the virtual machines during workflow executions. The effectiveness of CDAP is demonstrated through extensive experiments where we evaluated our proposed method against a set of well-known data placement strategies.