TY - GEN
T1 - Work-Efficient Load Balancing
AU - Sharma, Gokarna
AU - Rai, Suresh
AU - Busch, Costas
AU - Trahan, Jerry L.
AU - Vaidyanathan, Ramachandran
PY - 2015/5/7
Y1 - 2015/5/7
N2 - Load balancing is the key to many parallel and distributed applications. We consider the following load balancing problem: given any undirected connected graph and an initial weight distribution on the nodes, determine a schedule to move weights across edges so as to have (almost) equal weights on the nodes. Weights are moved across edges in rounds, and, in a round, weights are moved between the adjacent nodes exactly once. We study this problem in both static and dynamic networks. Previously studied diffusion and dimension exchange algorithms are slow in practice in the sense that they require many rounds of weight exchanges. In this paper, we present a class of algorithms that are work-efficient, i.e., they reduce the number of rounds (i.e., iterations) of weight exchanges needed to balance the load. In our algorithms, a node exchanges load with its neighbors sequentially (one neighbor at a time) and the load at that node is updated before subsequent exchanges with other neighbors. Simulation results on six network topologies show that our algorithms balance the load quite work-efficiently compared to previous algorithms.
AB - Load balancing is the key to many parallel and distributed applications. We consider the following load balancing problem: given any undirected connected graph and an initial weight distribution on the nodes, determine a schedule to move weights across edges so as to have (almost) equal weights on the nodes. Weights are moved across edges in rounds, and, in a round, weights are moved between the adjacent nodes exactly once. We study this problem in both static and dynamic networks. Previously studied diffusion and dimension exchange algorithms are slow in practice in the sense that they require many rounds of weight exchanges. In this paper, we present a class of algorithms that are work-efficient, i.e., they reduce the number of rounds (i.e., iterations) of weight exchanges needed to balance the load. In our algorithms, a node exchanges load with its neighbors sequentially (one neighbor at a time) and the load at that node is updated before subsequent exchanges with other neighbors. Simulation results on six network topologies show that our algorithms balance the load quite work-efficiently compared to previous algorithms.
KW - Convergence
KW - Diffusion algorithms
KW - Dimension exchange algorithms
KW - Dynamic networks
KW - Load balancing
KW - Work-efficiency
UR - http://www.scopus.com/inward/record.url?scp=84946575929&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84946575929&partnerID=8YFLogxK
U2 - 10.1109/ICPPW.2014.17
DO - 10.1109/ICPPW.2014.17
M3 - Conference contribution
AN - SCOPUS:84946575929
T3 - Proceedings of the International Conference on Parallel Processing Workshops
SP - 27
EP - 36
BT - Proceedings - 43rd International Conference on Parallel Processing Workshops, ICPPW 2014
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 43rd International Conference on Parallel Processing Workshops, ICPPW 2014
Y2 - 9 September 2014 through 12 September 2014
ER -