A higher order estimate of the optimum checkpoint interval for restart dumps

By J. Daly. Future generations computer systems, 2006

675 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Evaluating the viability of process replication reliability for exascale systems (2011, methodology): “Daly’s model [10] estimates that such a system should checkpoint once every 27 minutes and would achieve only 44% system utilization.”
Predicting the number of fatal soft errors in Los Alamos national laboratory's ASC Q supercomputer (2005, background): “simulations in [25], nonoptimal choice of the checkpoint restart interval may add as many as 15 or more hours to the length of time required to complete a 500-h calculation on a machine”
PLFS: a checkpoint filesystem for parallel applications (2009, background): “Even using optimal checkpoint frequencies [9], checkpointing has become the driving workload for parallel file systems and the challenge it imposes grows with each successively larger supercomputer [32, 42].”
Addressing failures in exascale computing (2014, methodology): “) The performance of global checkpoint/restart schemes has been analyzed by multiple authors (Young, 1974; Daly, 2006).”
Fault Tolerance in Iterative-Convergent Machine Learning (2018, methodology): “The total running time T of the system can be modeled as (Daly, 2006): T = Tsolve + Tdump + Trework + Trestart where Tsolve is the normal runtime of the program without any failures, Tdump is the time taken saving a checkpoint, Trework is the time spent repeating lost work due to restoring to a…”