Performance Techniques for COTS Systems

  • Authors:
  • Erik Putrycz;Murray Woodside;Xiuping Wu

  • Affiliations:
  • National Research Council of Canada;Carleton University;Carleton University

  • Venue:
  • IEEE Software
  • Year:
  • 2005

Quantified Score

Hi-index 0.00

Visualization

Abstract

The increased adoption of COTS-based systems creates new challenges for managing performance through an application's life cycle. Because the source code of the components involved isn't available, many existing techniques are inadequate. However, the availability of the components before the system is built provides early data on their performance properties. The authors' approach combines modeling based on this data with new tracing techniques. Both address internal concurrency in the COTS components and its effect on performance. The models help predict performance, plan the architecture and deployment, and plan measurements, and trace analysis helps diagnose problems in detail.