Visualising Distributed File System Architecture using Mixed-Reality

Distributed File Systems (DFSs) have become a key component for cloud computing. However, a large-scale cloud computing cluster environment has to face many challenges due to file or data transfer and network overheads, which lead to the development of new distributed file systems, such as Hadoop Distributed File System (HDFS). One of the key advantages of HDFS is the support for a fault-tolerant, scalable and reliable data storage. Due to this, HDFS is one of the most suitable DFSs to process MapReduce jobs. However, one of the major challenge with HDFS is that most of the techniques to analyse and identify the performance of HDFS are experimental, and hence, cannot be relied on. Therefore, this research paper purposes a mixed-reality application, HoloHDFS, to understand and visualise the architecture of HDFS and its data processing with the help of mixed-reality and holograms. An application to analyse the resources of Hadoop cluster and provide a way to highlight the inefficient workload/data distribution in a distributed file system environment. This paper describes the design and implementation of the application and discusses the encouraging results from the initial evaluation of the system, and further, discusses how the application can be extended and improvised in several ways