MODELING AND OPTIMIZATION OF DATA ASSIGNMENT IN A DISTRIBUTED INFORMATION-SYSTEM

Abstract

The assignment of data among several processing nodes for distributed processing is discussed. The performance metric of interest is the mean system response time including both the node delay and communication delay. A queueing optimization model is formulated as a non-linear programming problem with linear constraints. We employ the Newton method and propose an iterative solution procedure which gives the optimal policy. In addition, we illustrate a distributed information system with a numerical example to demonstrate the suitability of our methodology.