A High-Performance Computing Forecast: Partly Cloudy

  • Authors:
  • Thomas Sterling;Dylan Stark

  • Affiliations:
  • Louisiana State University;Louisiana State University

  • Venue:
  • Computing in Science and Engineering
  • Year:
  • 2009

Quantified Score

Hi-index 0.00

Visualization

Abstract

Cloud computing is emerging as an important computational resource allocation trend in commercial, academic, and industrial sectors. However, for high-performance computing (HPC), extreme scalability, efficiency, reliability, and security requirements extend beyond cloud computing capabilities, at least in its current form. Clouds could serve the general data processing workload within the HPC community, some large and potentially distributed datasets, and decoupled throughput-computing tasks. Yet, the cloud concept doesn't address and can't satisfy the needs of other workflow classes requiring extreme-scale, tightly coupled capability computing, large sensitive datasets, and optimized algorithms. This article explores the relationship of clouds to this spectrum of HPC needs and predicts a partly cloudy forecast.