Fault Tolerant Implementation of Peer-to-Peer Distributed Iterative Algorithms - Archive ouverte HAL Access content directly
Conference Papers Year :

Fault Tolerant Implementation of Peer-to-Peer Distributed Iterative Algorithms

(1) , (1)
1
The Tung Nguyen
  • Function : Author
  • PersonId : 924032
Didier El Baz

Abstract

Fault tolerance issues related to the implementation of distributed iterative algorithms via the P2PDC peer-to-peer distributed computing environment are considered. P2PDC is a decentralized environment dedicated to task parallel applications. It has been designed more particularly for the solution of large scale numerical simulation problems via distributed iterative algorithms. The environment allows frequent and direct communications between peers i.e., machines. P2PDC is based on P2PSAP, a self-adaptive communication protocol. We present new functionalities of P2PDC aimed at making our environment more robust. An adaptive fault tolerance mechanism ensures the robustness of computation to cope with peer faults. We consider also fault tolerance from an algorithmic point of view: we concentrate in particular on distributed asynchronous iterative algorithms that can tolerate some message loss. A series of computational results is presented and analyzed for a numerical simulation problem.
Fichier principal
Vignette du fichier
4914a137.pdf (211.24 Ko) Télécharger le fichier
Origin : Publisher files allowed on an open archive
Loading...

Dates and versions

hal-01153169 , version 1 (19-05-2015)

Identifiers

Cite

The Tung Nguyen, Didier El Baz. Fault Tolerant Implementation of Peer-to-Peer Distributed Iterative Algorithms. 15th IEEE International Conference on Computational Science and Engineering, Dec 2012, Paphos, Cyprus. 9p., ⟨10.1109/ICCSE.2012.103⟩. ⟨hal-01153169⟩
202 View
105 Download

Altmetric

Share

Gmail Facebook Twitter LinkedIn More