Visualising the train garbage collector

  • Authors:
  • Tony Printezis;Alex Garthwaite

  • Affiliations:
  • University of Glasgow, Glasgow, Scotland;Sun Microsystems Laboratories, Burlington, MA

  • Venue:
  • Proceedings of the 3rd international symposium on Memory management
  • Year:
  • 2002

Quantified Score

Hi-index 0.00

Visualization

Abstract

This paper presents a novel method for visualising an incremental garbage collector, based on the well-known Train algorithm, that generates concise snapshots of its state and informative graphs of its operation over time. To obtain these visualisations we used GC-spy, a generic heap visualisation framework. We show how this easy-to-use tool provided a visualisation model that was effective in both confirming our pre-existing beliefs about the collector's operation and, more interestingly, highlighting unexpected patterns in its behaviour. Based on this successful experience, we advocate the use of similar visualisation approaches to better understand and ultimately tune, profile, and improve other equally complex garbage collectors.