16th Euromicro Conference on Parallel, Distributed and Network-Based Processing (PDP 2008) (2009)
Feb. 18, 2009 to Feb. 20, 2009
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/PDP.2009.19
Distributed computing platforms contribute for a large part to some of the most powerful computers.Such architectures raise new challenges, typically in terms of scheduling, adaptability and security. This paper addresses the issue of result-checking in distributed environments, where tasks or their results could have been corrupted due to benign or malicious acts.Using a macro-data flow representation of the program execution, this article presents a novel approach based on work-stealing scheduling to dynamically adapt the execution to sabotage while keeping a reasonable slowdown rate.Unlike static adaptation or adaptation at the source code level, a dynamic adaptation at the middleware level is proposed, enforcing separation of concepts and programming transparency.This article contains both conceptual and experimental results that show the interest, feasibility and limits of the concept.
dynamic adaptation, work stealing, sabotage tolerance
S. Varrette, J. Pazat, S. Guelton and T. Gautier, "Dynamic Adaptation Applied to Sabotage Tolerance," 16th Euromicro Conference on Parallel, Distributed and Network-Based Processing (PDP 2008)(PDP), Weimar, Germany, 2009, pp. 237-244.