Workstation Cluster’s Hadoop Distributed File System Simulation and Modeling

Abstract

On desktop PCs, the capacity of the hard disks is increasing. Therefore, this study simulated and modeled Hadoop distributed file systems on a workstation cluster. The main aim was to determine how effective and feasible it would be to exploit idle computational storage. Architecturally, the proposed model was that which relied on HDFS (Hadoop Distributed File System). Also, CPN tools were used during model implementation. Hence, the tools constituted CPN ML programming language and Colored Petri Dish Nets. To ensure that the availability of the workstations was characterized within the model, the data collection process occurred in a computer lab for about 40 days. Indeed, findings demonstrated that the HDFS’s source code could be modified and the number of replicas specified to ensure that the process of reading and writing files is reliable, even in the wake of failures that could arise when computers in the cluster of workstations are turned on and off.