Efficient broadcast in heterogeneous networks of workstations using two sub-networks
International Journal of Parallel Programming
A Message Scheduling Scheme for All-to-All Personalized Communication on Ethernet Switched Clusters
IEEE Transactions on Parallel and Distributed Systems
Scheduling for atomic broadcast operation in heterogeneous networks with one port model
The Journal of Supercomputing
Scheduling for atomic broadcast operation in heterogeneous networks with one port model
GPC'08 Proceedings of the 3rd international conference on Advances in grid and pervasive computing
| Hi-index | 0.00 |
This paper presents three efficient contention-free algorithmsfor broadcasting on heterogeneous networks of workstations(HNOW). In an HNOW, many multiple speed types of workstationsthat their send and receive overhead are distinct. Previousresearch has shown that finding an optimal routing scheme inan HNOW is not easy [4, 1], because properly arranging thescheduling tree in routing is difficult. Therefore, this investigationfocuses mainly on enhancing the performance of an HNOWby properly arranging fastest nodes to the internal nodes ofupper levels in the scheduling tree. The fastest node first isfundamental in designing an efficient algorithm. This paperpresents three schemes named EBS, VBBS, and VBBSWF. Allof these three schemes can be implemented in O(nlog(n)) time,where n is the number of workstations. They are all contention-freewhen broadcasting in an HNOW. The post-order recursivedoubling is used as as a platform to design our algorithms.Based on the simulation result, the performance outperformsthe broadcast with minimal steps [5] and the scheduling tree[9] generated by dynamic programing in an HNOW.