1 \documentclass[smallextended]{svjour3}
2 \usepackage[utf8]{inputenc}
3 \usepackage[T1]{fontenc}
10 \title{Best effort strategy and virtual load for asynchronous iterative load balancing}
12 \author{Raphaël Couturier \and
17 \institute{F. Author \at
19 Tel.: +123-45-678910\\
21 \email{fauthor@example.com} % \\
22 % \emph{Present address:} of F. Author % if needed
33 Most of the time, asynchronous load balancing algorithms have extensively been
34 studied in a theoretical point of view. The Bertsekas and Tsitsiklis' algorithm
35 is certainly the most well known algorithm for which the convergence proof is
36 given. From a practical point of view, when a node wants to balance a part of
37 its load to some of its neighbors, the strategy is not described. In this
38 paper, we propose a strategy called \texttt{best effort} which tries to balance
39 the load of a node to all its less loaded neighbors while ensuring that all the
40 nodes concerned by the load balancing phase have the same amount of load.
41 Moreover, asynchronous iterative algorithms in which an asynchronous load
42 balancing algorithm is implemented most of the time can dissociate messages
43 concerning load transfers and message concerning load information. In order to
44 increase the converge of a load balancing algorithm, we propose a simple
45 heuristic called \texttt{virtual load} which allows a node that receives an load
46 information message to integrate the load that it will receive later in its
47 load (virtually) and consequently sends a (real) part of its load to some of its
48 neighbors. In order to validate our approaches, we have defined a simulator
49 based on SimGrid which allowed us to conduct many experiments.
56 Load balancing algorithms are extensively used in parallel and distributed
57 applications in order to reduce the execution times. They can be applied in
58 different scientific fields from high performance computation to micro sensor
59 networks. They are iterative by nature. In literature many kinds of load
60 balancing algorithms have been studied. They can be classified according
61 different criteria: centralized or decentralized, in static or dynamic
62 environment, with homogeneous or heterogeneous load, using synchronous or
63 asynchronous iterations, with a static topology or a dynamic one which evolves
64 during time. In this work, we focus on asynchronous load balancing algorithms
65 where computer nodes are considered homogeneous and with homogeneous load with
66 no external load. In this context, Bertsekas and Tsitsiklis have proposed an
67 algorithm which is definitively a reference for many works. In their work, they
68 proved that under classical hypotheses of asynchronous iterative algorithms and
69 a special constraint avoiding \texttt{ping-pong} effect, an asynchronous
70 iterative algorithm converge to the uniform load distribution. This work has
71 been extended by many authors. For example, DASUD propose a version working with
75 \bibliographystyle{spmpsci}
83 %%% ispell-local-dictionary: "american"
86 % LocalWords: Raphaël Couturier Arnaud Giersch Abderrahmane Sider
87 % LocalWords: Bertsekas Tsitsiklis SimGrid DASUD