Investigating interactive visualisation in a Cloud computing environment

Abstract

Scientists working with large datasets without a desktop with advanced capacity may not be able to visualise the simulation output efficiently. This is because visualisation of large datasets is computationally intensive in terms of filtering, mapping, and rendering the datasets. The time taken to visualise an image from a large dataset from an underpowered desktop computer may be prolonged, which would not be an interactive experience for the scientists. The desktop can manage a small dataset efficiently compared to client/server mode; however, larger datasets require more memory and number of processors to visualise. This project investigates if interactive visualisation is feasible in a Cloud computing environment. A virtual machine (VM) was created which was then deployed in a Cloud environment at The University of Auckland to visualise large datasets. Results showed that ParaView server VM could be deployed in a Cloud environment which offers more memory and processors for the VM to be utilised. Thus, the interactive visualisation of large datasets is feasible in a Cloud computing environment. Results from the performance tests showed larger datasets require more memory and numbers of processes to perform the visualisation. However, the increases in number of processes and memory size would not necessarily improve the performance, which depend on the type and size of datasets and the ParaView operations such as filtering, mapping, and rendering. Future work on even larger datasets is warranted.