Using A Double Weighted Clustering Technique for Fragment Allocation in Wide Area Networks
George Semeczko · 1997
: Many papers have been written about resource allocation problems in distributed systems. In solving these optimization problems there has always been a trade-off between the optimality of the solution, the expense in deriving a solution and the complexity of the problem being solved. For problems in Wide Area Networks, the complexity of the problem is immediately more complicated than those for Local Area Networks as the variable communication costs become another factor in the optimization function. In this paper, we show that in practice Wide Area Networks have more structure than usually assumed. Given this, a structured model is presented that can be used to approximate Wide Area Networks and allow "better" solutions to optimization problems due to its properties. A measure is also presented that can be used to determine how close the allocation is to an optimal solution. 1. Introduction The problem of fragment allocation in distributed database systems is very dependant upon t...