TY - GEN

T1 - Robust network supercomputing without centralized control

AU - Davtyan, Seda

AU - Konwar, Kishori M.

AU - Shvartsman, Alexander A.

PY - 2011

Y1 - 2011

N2 - Traditional approaches to network supercomputing employ a master process and a large number of potentially undependable worker processes that must perform a collection of tasks on behalf of the master. In such a centralized scheme, the master process is a performance bottleneck and a single point of failure. This work develops an original approach that eliminates the master and instead uses a decentralized algorithm, where each worker is able to determine locally that all tasks have been performed, and to collect locally the results of all tasks. The failure model assumes that the average probability of a worker returning a wrong result is inferior to 1/2. A randomized synchronous algorithm for n processes and n tasks is presented. The algorithm terminates in θ(log n) rounds, and it is proved that upon termination the workers know the results of all tasks with high probability, and that these results are correct with high probability. The message complexity of the algorithm is θ(n log n), and the bit complexity is O(n2 log3 n).

AB - Traditional approaches to network supercomputing employ a master process and a large number of potentially undependable worker processes that must perform a collection of tasks on behalf of the master. In such a centralized scheme, the master process is a performance bottleneck and a single point of failure. This work develops an original approach that eliminates the master and instead uses a decentralized algorithm, where each worker is able to determine locally that all tasks have been performed, and to collect locally the results of all tasks. The failure model assumes that the average probability of a worker returning a wrong result is inferior to 1/2. A randomized synchronous algorithm for n processes and n tasks is presented. The algorithm terminates in θ(log n) rounds, and it is proved that upon termination the workers know the results of all tasks with high probability, and that these results are correct with high probability. The message complexity of the algorithm is θ(n log n), and the bit complexity is O(n2 log3 n).

KW - distributed algorithms

KW - fault-tolerance

KW - internet supercomputing

KW - randomized algorithms

UR - http://www.scopus.com/inward/record.url?scp=79959897254&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=79959897254&partnerID=8YFLogxK

U2 - 10.1145/1993806.1993860

DO - 10.1145/1993806.1993860

M3 - Conference contribution

AN - SCOPUS:79959897254

SN - 9781450307192

T3 - Proceedings of the Annual ACM Symposium on Principles of Distributed Computing

SP - 293

EP - 294

BT - PODC'11 - Proceedings of the 2011 ACM Symposium Principles of Distributed Computing

T2 - 30th Annual ACM SIGACT-SIGOPS Symposium on Principles of Distributed Computing, PODC'11, Held as Part of the 5th Federated Computing Research Conference, FCRC

Y2 - 6 June 2011 through 8 June 2011

ER -