転送性能予測を用いたミラーサーバ選択方式

概要

論文の詳細を見る
The Server mirroring technique that dispersing contents in more than one network organization is being used widely. This is to avoid access concentration to the server from the increase of the Internet user. In such a case, route to each server is different to every user. And network condition changes by the time. So the technique that the server which is always the most suitable is chosen becomes necessary in consideration of the network traffic. In this paper server selection method is proposed. The technique that server selection uses information when contents are transferred. And the transfer performance is predict by using the back propagation method of the neural network and it selects it. Then the evaluation that the information measured over the actual network is applied to the proposal technique is done. We show its validity.
近畿大学の論文
2006-12-20