Publications of Torsten Hoefler

Jens Domke, Torsten Hoefler, Satoshi Matsuoka:

Fail-in-Place Network Design: Interaction between Topology, Routing Algorithm and Failures

(presented in New Orleans, LA, USA, Nov. 2014, Proceedings of the International Conference on High Performance Computing, Networking, Storage and Analysis (SC14) )

Abstract

Hardware failures in networks are omnipresent and often inevitable. The growing system size of high performance computers results in a steady decrease of the mean time between failures. Exchanging network components often requires whole system downtime which increases the cost of failures. In this work, we study a fail-in-place strategy where broken network elements remain untouched for longer periods of time. Each hardware failure of a network component will decrease the overall bandwidth of the network and the rerouting delay will cause disconnects for a subset of the paths. We show, that a fail-in-place strategy is feasible for today’s networks and the degradation is manageable, and provide guidelines for the design. Our network failure simulation toolchain allows system designers to extrapolate the performance degradation based on expected failure rates, and it can be used to evaluate the current state of a system. In a case study of real-world HPC systems, we will analyze the performance degradation throughout the system’s lifetime under the assumption that faulty network components are not repaired, which results in a recommendation to change the used routing algorithm to improve the network performance as well as the fail-in-place characteristic. We firmly believe that fail-in-place is a practical strategy to move towards the next-generation large-scale networks.

Documents

download article:

download slides:

BibTeX

@inproceedings{fail-in-place,
  author={Jens Domke and Torsten Hoefler and Satoshi Matsuoka},
  title={{Fail-in-Place Network Design: Interaction between Topology, Routing Algorithm and Failures}},
  year={2014},
  month={Nov.},
  location={New Orleans, LA, USA},
  note={Proceedings of the International Conference on High Performance Computing, Networking, Storage and Analysis (SC14)},
  source={http://www.unixer.de/~htor/publications/},
}


serving: 52.14.125.137:28285	© Torsten Hoefler