Improving Computation Effectiveness by the means of Elasticity Property

By Honghai Yang


The dynamical cloud computing platform (DC2) established by the NSF Spatiotemporal Innovation Center, George Mason University has provided the opportunities of handling data intensive applications with scale computing and storage capabilities, easy accessibility by common users and scientists, on demand availability, and easy maintenance.

In order to test the elasticity of this cloud platform, we designed an experiment using Bowyer Watson algorithm, which can randomly create any given number of points and generate the Delaunay Triangulation. The Delaunay triangulation and its dual (the Voronoi Diagram) have been used widely in terrain modelling, 3D scene visualization, computer graphics, pattern recognition, finite element analysis and cartographic generalization.

The experiment is conducted on an AWS windows virtual machine, and the configuration is as follows: CPU E5-2676 2.4 GHz, RAM 4.0 GB and Windows Server 2012 R2. As shown in these figures (figure 1 and 2), with a fixed number of CPUs, the execution time improves gradually as the number of points increases. Furthermore, there is a high CPU utilization and still time consuming when a large number of point involved in computation.

To leverage the elasticity of STHCP, we are using auto scaling configuration of AWS, importing parallel framework mechanism to adjust the application, and fixing the size of the point set and setting calculated the relative speed-up and efficiency with different number of same environmental virtual machines.

Apply for Cloud Resources

Requesting access to spatiotemporal Hybrid Cloud Platform resources is fairly easy. Simply fill out the online application and submit one page proposal (use your project name as file name) to describe the project objectives. Apply now