Abstract
The performance of striping-based video servers really depends on the placement of video segments on the disks. However, it seems to be virtually impossible to find the optimal one out of them because there are a great number of placement patterns. In this paper, we propose a bland-new method that finds the optimal placement pattern of video segments solving the constraint optimization problem. First, finding the optimal segment placement is converted into the form of a constraint optimization problem, then it is solved using the GA (Genetic Algorithm). Solving this problem leads to find the optimal placement pattern based on the access probabilities of the segments. In the formulation, the objective functions investigate their effect on the performance of the server. As a result, the placement pattern that provides the required performance can be easily found by our method. In basic simulations, our proposed method was able to find a placement that is superior to other conventional placements.
