Modeling the Impact of Checkpoints on Next-Generation Systems

Oldfield,; Arunagiri,; Teller,; Seelam,; Varela,; Riesen,; Röth,

doi:10.1109/msst.2007.4367962

Cited by 86 publications

(88 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Based on a study of twenty-two (anonymous) HPC systems, Gibson and Schroeder observe that failure rate grows in proportion to the number of sockets, and give an "optimistic" estimate of 0.1 failures per socket per year [18]. Dual-socket nodes are common in that study, corresponding to a mean time to node failure of 5 years (also used in [31]). ASCI White's MTBI has been reported to be 5 hours in 2001 and 40 hours in 2003 (after the platform had stabilized).…”

Section: Systems Surveymentioning

confidence: 99%

“…If node mean time to failure (1/λ) is 5-10 years [18,31], it will be more cost-effective to run redundant rather than non-redundant on a system size of 15-20 thousand nodes. If 1/λ is 20 years, crossover occurs at a system size of approximately 28k nodes.…”

Section: Conditions Where Dual-redundant Computation Is Cost Effectivementioning

confidence: 99%

“…Whereas some studies assume a fixed time to write a checkpoint [10,41,34], the simulations underlying Figure 10(a) assume that checkpoint time δ increases linearly with job size n (as does [31]). This is practical -the more nodes, the more memory to checkpoint.…”

Section: Conditions Where Dual-redundant Computation Is Cost Effectivementioning

confidence: 99%

“…This is practical -the more nodes, the more memory to checkpoint. Although network capacity also increases with n, the I/O subsystem quickly becomes the limiting factor [31]. We also assume a fast filesystem per Section 2, and apply the same 80% of theoretical bandwidth limitation, such that δ = n * 16GB/(.8 * 240GB/s).…”

Section: Conditions Where Dual-redundant Computation Is Cost Effectivementioning

confidence: 99%

See 3 more Smart Citations

A Model-Based Case for Redundant Computation

Stearley

Robinson

Ferreira

et al. 2011

Self Cite

View full text Add to dashboard Cite

Despite its seemingly nonsensical cost, we show through modeling and simulation that redundant computation merits full consideration as a resilience strategy for next-generation systems. Without revolutionary breakthroughs in failure rates, part counts, or stable-storage bandwidths, it has been shown that the utility of Exascale systems will be crushed by the overheads of traditional checkpoint/restart mechanisms. Alternate resilience strategies must be considered, and redundancy is a proven unrivaled approach in many domains. We develop a distribution-independent model for job interrupts on systems of arbitrary redundancy, adapt Daly's model for total application runtime, and find that his estimate for optimal checkpoint interval remains valid for redundant systems. We then identify conditions where redundancy is more cost effective than non-redundancy. These are done in the context of the number one supercomputers of the last decade, showing that thorough consideration of redundant computation is timely -if not overdue.

show abstract

Section: Systems Surveymentioning

confidence: 99%

Section: Conditions Where Dual-redundant Computation Is Cost Effectivementioning

confidence: 99%

Section: Conditions Where Dual-redundant Computation Is Cost Effectivementioning

confidence: 99%

Section: Conditions Where Dual-redundant Computation Is Cost Effectivementioning

confidence: 99%

See 2 more Smart Citations

A Model-Based Case for Redundant Computation

Stearley

Robinson

Ferreira

et al. 2011

Self Cite

View full text Add to dashboard Cite

show abstract

“…This poses challenges for system software in the areas of scalability, resiliency and programmability. The shear scale of future systems necessitate much improved resiliency, or else the majority of an application's runtime will be spent in overhead due to checkpoints and restarts [32]. Programming models are another area requiring attention.…”

Section: Introductionmentioning

confidence: 99%

Summary of multi-core hardware and programming model investigations

Kelly¹,

Pedretti²,

Levenhagen³

2008

View full text Add to dashboard Cite

This report summarizes our investigations into multi-core processors and programming models for parallel scientific applications. The motivation for this study was to better understand the landscape of multi-core hardware, future trends, and the implications on system software for capability supercomputers. The results of this study are being used as input into the design of a new open-source light-weight kernel operating system being targeted at future capability supercomputers made up of multi-core processors. A goal of this effort is to create an agile system that is able to adapt to and efficiently support whatever multi-core hardware and programming models gain acceptance by the community.3

show abstract

Unified model for assessing checkpointing protocols at extreme‐scale

Bosilca

Bouteiller

Brunet

et al. 2013

Concurrency and Computation

View full text Add to dashboard Cite

In this article, we present a unified model for several well-known checkpoint/restart protocols. The proposed model is generic enough to encompass both extremes of the checkpoint/restart space, from coordinated approaches to a variety of uncoordinated checkpoint strategies (with message logging). We identify a set of crucial parameters, instantiate them and compare the expected efficiency of the fault tolerant protocols, for a given application/platform pair. We then propose a detailed analysis of several scenarios, including some of the most powerful currently available HPC platforms, as well as anticipated Exascale designs. The results of this analytical comparison are corroborated by a comprehensive set of simulations. Altogether, they outline comparative behaviors of checkpoint strategies at very large scale, thereby providing insight that is hardly accessible to direct experimentation. Un modèle unifié pour l'évaluation des protocoles de checkpointà très largeéchelle Résumé :Nous présentons ici un modèle unifié de plusieurs protocoles de sauvegarde de points de reprise (checkpoints) et de redémarrage. Le modèle proposé est suffisamment générique pour contenir les deux extrêmes des techniques de checkpoint/restart, d'une approche coordonnéeà toute une famille de stratégies non-coordonnées (avec enregistrement de messages).Nous identifions un ensemble de paramètres cruciaux, les instancions et comparons l'espérance de l'efficacité des protocoles de tolérance aux pannes, pour un couple donné application/plate-forme. Nous proposons une analyse détaillée de plusieurs scénarios, incluant certaines des plates-formes de calcul existantes les plus puissantes, ainsi que des anticipations sur les futures plates-formes exascale. Les résultats de cette analyse sont corroborés par un ensemble de simulations. Ensemble, ces résultats illustrent le comportement relatif des différentes stratégiesà largeéchelle, fournissant des enseignements qu'il serait très difficile, voire impossible, d'obtenir par l'expérimentation directe.

show abstract

Modeling the Impact of Checkpoints on Next-Generation Systems

Cited by 86 publications

References 25 publications

A Model-Based Case for Redundant Computation

A Model-Based Case for Redundant Computation

Summary of multi-core hardware and programming model investigations

Unified model for assessing checkpointing protocols at extreme‐scale

Contact Info

Product

Resources

About