PREVENTIVE MIGRATION VS. PREVENTIVE CHECKPOINTING FOR EXTREME SCALE SUPERCOMPUTERS
Abstract
An alternative to classical fault-tolerant approaches for large-scale clusters is failure avoidance, by which the occurrence of a fault is predicted and a preventive measure is taken. We develop analytical performance models for two types of preventive measures: preventive checkpointing and preventive migration. We instantiate these models for platform scenarios representative of current and future technology trends. We find that preventive migration is the better approach in the short term by orders of magnitude. However, in the longer term, both approaches have comparable merit with a marginal advantage for preventive checkpointing. We also develop an analytical model of the performance for fault tolerance based on periodic checkpointing and compare this approach to both failure avoidance techniques. We find that this comparison is sensitive to the nature of the stochastic distribution of the time between failures, and that failure avoidance is likely inferior to fault tolerance in the long term. Regardless, our result show that each approach is likely to achieve poor utilization for large-scale platforms (e.g., 220 nodes) unless the mean time between failures is large. We show how bounding parallel job size improves utilization, but conclude that achieving good utilization in future large-scale platforms will require a combination of techniques.
References
- Int. J. High Perform. Comput. Appl. 23(4), 309 (2009), DOI: 10.1177/1094342009347714. Crossref, ISI, Google Scholar
- Vivek Sarkar and others. Exascale software study: Software challenges in extreme scale systems, 2009. White paper. See , http://users.ece.gatech.edu/mrichard/ExascaleComputingStudyReports/ECSS%20report%20101909.pdf . Google Scholar
- Top500 Supercomputer Sites , http://www.top500.org . Google Scholar
- Int. Journal of High Performance Computing Applications 23(4), 374 (2009), DOI: 10.1177/1094342009347767. Crossref, ISI, Google Scholar
- Journal of Physics: Conference Series 78(012022), (2007), DOI: 10.1088/1742-6596/78/1/012022. Google Scholar
Nick Kolettis and N. Dudley Fulton , Software rejuvenation: Analysis, module and applications, FTCS '95: Proc. 25th Int. Symposium on Fault-Tolerant Computing (IEEE Computer Society, Washington, DC, USA, 1995) p. 381. Google Scholar- IBM J. Res. Dev. 45(2), 311 (2001), DOI: 10.1147/rd.452.0311. Crossref, ISI, Google Scholar
-
Prashasta Gujrati , A meta-learning failure predictor for blue gene/l systems , Int. Conf. Parallel Processing (ICPP) ( IEEE Computer Society , 2007 ) . Google Scholar Chao Wang , Proactive process-level live migration in hpc environments, SC '08: Proceedings 2008 ACM/IEEE Conf. Supercomputing (IEEE Press, 2008) pp. 1–12. Google ScholarSayantan Chakravorty , Celso L. Mendes and Laxmikant V. Kalé , Proactive fault tolerance in mpi applications via task migration, HiPC 2006, the IEEE High performance Computing Conference (IEEE Computer Society Press, 2006) pp. 485–496. Google ScholarA. Oliner and J. Stearley , What supercomputers say: A study of five system logs, DSN '07: Proceedings of the 37th Annual IEEE/IFIP International Conference on Dependable Systems and Networks (IEEE Computer Society, Washington, DC, USA, 2007) pp. 575–584. Google Scholar- SIGMETRICS Perform. Eval. Rev. 30(1), 217 (2002), DOI: 10.1145/511399.511362. Crossref, ISI, Google Scholar
B. Schroeder and G. A. Gibson , A large-scale study of failures in high-performance computing systems, Proc. of the International Conference on Dependable Systems and Networks (2006) pp. 249–258. Google Scholar- Journal of Parallel and Distributed Computing 63(11), 1105 (2003), DOI: 10.1016/S0743-7315(03)00108-4. Crossref, ISI, Google Scholar
M. J. Koop , Performance analysis and evaluation of pcie 2.0 and quad-data rate infiniband, HOTI '08: Proceedings of the 2008 16th IEEE Symposium on High Performance Interconnects (IEEE Computer Society, Washington, DC, USA, 2008) pp. 85–92. Google ScholarL. Wang , Modeling Coordinated Checkpointing for Large-Scale Supercomputers, Proc. Int. Conf. on Dependable Systems and Networks (2005) pp. 812–821. Google Scholar- Joshua Wingstrom. Overcoming The Difficulties Created By The Volatile Nature Of Desktop Grids Through Understanding, Prediction And Redundancy. PhD thesis, University of Hawai'i at Manoa, 2009 . Google Scholar
- Communications of the ACM 17(9), 530 (1974), DOI: 10.1145/361147.361115. Crossref, ISI, Google Scholar
- Future Generation Computer Systems 22(3), 303 (2004). Crossref, ISI, Google Scholar
- J. Comput. Appl. Math. 123(1-2), 201 (2000), DOI: 10.1016/S0377-0427(00)00409-X. Crossref, ISI, Google Scholar


