by Matthias Tichy, Holger Giese, Daniela Schilling, Wladimir Pauls
Abstract:
The dependability of a software system can be improved by online redeployment of failed software components using appropriate system self-repair actions. The effect of different self-repair actions can vary to a great extent w.r.t. the resulting temporary service unavailability and reduced redundancy of services. We therefore developed an approach to efficiently compute self-repair actions which realize requested repair steps in a nearly optimal manner. We show that our approach achieves a suitable compromise between the usually infeasible optimal deployment modification w.r.t. damage minimization and repair time minimization by presenting a number of simulation results.
Reference:
Computing Optimal Self-Repair Actions: Damage Minimization versus Repair Time (Matthias Tichy, Holger Giese, Daniela Schilling, Wladimir Pauls), In Proc. of the ICSE 2005 Workshop on Architecting Dependable Systems, St. Louis, Missouri, USA (Rogério de Lemos, Alexander Romanovsky, eds.), ACM Press, 2005.
Bibtex Entry:
@InProceedings{TGSP05_ag,
AUTHOR = {Tichy, Matthias and Giese, Holger and Schilling, Daniela and Pauls, Wladimir},
TITLE = {{Computing Optimal Self-Repair Actions: Damage Minimization versus Repair Time}},
YEAR = {2005},
BOOKTITLE = {Proc. of the ICSE 2005 Workshop on Architecting Dependable Systems, St. Louis, Missouri, USA},
PAGES = {1--6},
EDITOR = {de Lemos, Rogério and Romanovsky, Alexander},
PUBLISHER = {ACM Press},
URL = {http://www.upb.de/cs/ag-schaefer/Veroeffentlichungen/Quellen/Papers/2005/wads2005.pdf},
PDF = {uploads/pdf/TGSP05_ag_wads2005.pdf},
ABSTRACT = {The dependability of a software system can be improved by online redeployment of failed software components using appropriate system self-repair actions. The effect of different self-repair actions can vary to a great extent w.r.t. the resulting temporary service unavailability and reduced redundancy of services. We therefore developed an approach to efficiently compute self-repair actions which realize requested repair steps in a nearly optimal manner. We show that our approach achieves a suitable compromise between the usually infeasible optimal deployment modification w.r.t. damage minimization and repair time minimization by presenting a number of simulation results.}
}