An Experimental Study to Determine Task Size for Rollback Recovery Systems

  • Authors:
  • Shambhu J. Upadhyaya;Kewal K. Saluja

  • Affiliations:
  • State Univ. of New York, Buffalo, NY;Univ. of Wisconsin, Madison

  • Venue:
  • IEEE Transactions on Computers
  • Year:
  • 1988

Quantified Score

Hi-index 14.98

Visualization

Abstract

The effects of using a recovery cache to save the variables of a program are studied. A novel optimization model for rollback is formulated to include the effects of a recovery cache in rollback systems. The parameters of the model proposed are the maximum recovery time, the cache size, and the save and load time associated with the task size. The results are also discussed of an experimental study conducted to estimate the parameters of the programs that are critical for arriving at a suitable task size or cache size to minimize the cost of recovery.