Implementation and Usability Evaluation of a Cloud Platform for Scientific Computing as a Service (SCaaS)

  • Authors:
  • Prasad Saripalli;Curt Oldenburg;Ben Walters;N. Radheshyam

  • Affiliations:
  • -;-;-;-

  • Venue:
  • UCC '11 Proceedings of the 2011 Fourth IEEE International Conference on Utility and Cloud Computing
  • Year:
  • 2011

Quantified Score

Hi-index 0.00

Visualization

Abstract

Scientific computing requires simulation and visualization involving large data sets among collaborating teams. Cloud platforms offer a promising solution via ScaaS. We report on the architecture, implementation and User Experience (UX) evaluation of one such SCaaS platform implementing TOUGH2V2.0, a numerical simulator for sub-surface fluid and heat flow, offered as a service. Results from example simulations, with virtualization of workloads in a multi-tenant, Virtual Machine (VM)-based cloud platform, are presented. These include fluid production from a geothermal reservoir, diffusive and advective spreading of contaminants, radial flow from a CO2 injection well and gas diffusion of a chemical through porous media. Prepackaged VM pools deployed autonomically ensure that sessions are provisioned elastically on demand. Users can access data-intensive visualizations via a web-browser. Authentication, user state and sessions are managed securely via an Access Gateway, to autonomically redirect and manage the workflows when multiple concurrent users are accessing their own sessions. Usability in the cloud and the traditional desktop are comparatively assessed, using several UX metrics. Simulated network conditions of different quality were imposed using a WAN emulator. Usability was found to be good for all the simulations under even moderately degraded network quality, as long as latency was not well above 100 ms. Hosting of a complex scientific computing application on an actual, global Enterprise cloud platform (as opposed to earlier remoting platforms) and its usability assessment, both presented for the first time, are the essential contributions of this work.