OPTIMAL NUMBER OF WEB SERVER RESOURCES FOR JOB APPLICATIONS

The main purpose of this paper is to propose optimization problems in which how many number N of web servers should be provided for net jobs with random process times. We consider the first case when a single job with random time S is processed and take up the second case when number n of jobs with successive times are processed. The number n may not be a constant value that could be predefined from the practical point, so that we modify the model in the second case by supposing n to be a random variable. Next, we introduce shortage and excess costs into models to consider both costs suffered before and after failures of server system. We obtain the total expected costs for each model and optimize them analytically. When physical server failure time and job process time are exponentially distributed, optimal numbers that minimize the expected costs are computed numerically.