Authors: Stephen Dowsland, Mark Turner, Nick Holliman
The cloud offers opportunities to visualize and explore big data problems in ways that would not otherwise be possible within the constraints of a desktop or mobile device’s memory and compute power. This paper describes the architecture of, and presents performance results from, a new scalable platform for rendering visualizations in the cloud. The platform is known as the Urban Insight Cloud Engine (UICE). By adopting existing technologies, we have created a cloud-provider agnostic architecture that can utilize both cloud CPU and GPU resources to speed up the creation of complex, high quality visualizations. The platform will enable future research into these challenges when creating visualizations in the cloud for consumption by thin clients; allowing us to establish the benefits and limits of scalability. We will outline the current capabilities of the platform, including scalability performance results and explore the intended direction in terms of improvement and optimization of the rendering pipeline.
Keywords: remote rendering, visualization, scalability, cloud computing, urban data, big data