Network Flow Heuristic algorithm for a distributed web service selection problem

In this paper a new model for a distributed Web service system is presented. The proposed system is composed of multiple Web service components having multiple alternative versions distributed among multiple servers. For a given set of requests an allocation is to be found that maximizes total client satisfaction subject to the resource constraints of the servers. To solve this multidimensional multi knapsack problem, which is NP hard, we propose a heuristic using a variant of the network flow maximization algorithm. Not only the heuristic is polynomial but also it inherently rules out the number of requests from contributing in time complexity of the algorithm.