SIFT: a scalable iterative-unfolding technique for filtering execution traces

  • Authors:

  • A. V. Miranskyy;N. H. Madhavji;M. S. Gittens;M. Davison;M. Wilding;D. Godwin;C. A. Taylor

  • Affiliations:

  • University of Western Ontario, London, ON, Canada;University of Western Ontario, London, ON, Canada;University of the West Indies, Cave Hill, Barbados;University of Western Ontario, London, ON, Canada;IBM Canada Ltd., Markham, ON, Canada;IBM Canada Ltd., Markham, ON, Canada;IBM Canada Ltd., Markham, ON, Canada

  • Venue:
  • CASCON '08 Proceedings of the 2008 conference of the center for advanced studies on collaborative research: meeting of minds
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

Comparing program execution traces can be useful for numerous purposes, such as software testing, system security analysis, program comprehension, software evolution and other areas of software development. Unfortunately, trace comparison techniques that operate on execution traces containing full execution details are too slow for use in large-scale production system environments. In order to speed up the comparisons, we propose a technique (called SIFT) for "filtering-out" irrelevant traces from a given set so that only the relevant few, residual, traces are then used for comparison. Our solution involves multiple levels of trace compression, each with a different degree of abstraction. These traces are compared iteratively while filtering out dissimilar traces. This paper describes the compression and comparison algorithms. Prototype results from a significant case study show that the SIFT approach is efficient and scalable for use in an industrial software development environment.