Design, Modeling, and Evaluation of a Scalable Multi-level Checkpointing System
By A. Moody, G. Bronevetsky, K. Mohror, Bronis R. de Supinski. 2010 ACM/IEEE International Conference for High Performance Computing, Networking, Storage and Analysis, 2010
633 citations recorded by the source corpus.
Subfields: Computer Networks and Communications, Hardware and Architecture
View Paper: https://www.semanticscholar.org/paper/8a19dfd6e417bd35a7b04a2adc22549d45b7ad2c
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Computer Networks and Communications
- Kiwan Maeng
- Carole-Jean Wu
- Rolf Riesen
- Pradip Bose
- R. Brightwell
- Subhasish G. Mitra
- Brandon Lucia
- Sven Leyffer
In Hardware and Architecture
- Kiwan Maeng
- Pradip Bose
- Subhasish G. Mitra
- Brandon Lucia
- Murali Annavaram
- Christian Engelmann
- Franck Cappello
- Michael G. Rabbat
Citations in the corpus
- Addressing failures in exascale computing (2014, background): “Currently, two environments provide multilevel checkpoint/restart: SCR (scalable checkpoint/restart) [114] and FTI (fault tolerance interface) [13].”
- Evaluating the viability of process replication reliability for exascale systems (2011, background): “Multi-level checkpointing [27] is a library-based approach for controlling checkpointing to multiple storage targets, including memory-based checkpoints, local checkpoint storage, and remote checkpoints, into a single system.”
- This paper is included in the Proceedings of the 19th USENIX Symposium on Networked Systems Design and Implementation. (background): “Checkpoint optimization schemes include techniques to reduce latency [31], coordinating across multiple snapshots for efficient reconstruction [27,35], using different checkpoint resolutions for providing varying levels of recovery [8,20].”
- Toward Exascale Resilience: 2014 update (2014, methodology): “A similar approach has been used in SCR [78].”
- CPR: Understanding and Improving Failure Tolerant Training for Deep Learning Recommendation with Partial Recovery (2020, methodology): “Many optimizations orthogonal to CPR have been proposed to speed up checkpointing, including using multi-level hierarchy (Moody et al., 2010; Bautista-Gomez et al., 2011), adding asynchronous operations (Nico-lae et al., 2019), leveraging memory access patterns (Nicolae & Cappello, 2013; Carbone et…”