Network Design Method by Finding Server Placement and Protected Links to Keep Connectivity to Servers Against Link Failures
Network Design Method by Finding Server Placement and Protected Links to Keep Connectivity to Servers Against Link Failures
复制标题
通过寻找服务器放置和受保护的链路来保持与服务器的连接以防止链路故障的网络设计方法
DOI:
10.1109/incos.2016.73
复制
发表时间:
2016
期刊:
影响因子:
--
通讯作者:
Hiroyoshi Miwa
中科院分区:
文献类型:
--
作者:
Daishi Irie;Hiroyoshi Miwa
It is important for network service providers to design and operate reliable network which is robust against network failures. Especially, a network must be sufficiently connected, even if link failures occur. There are many methods to design a robust network. One is a method based on a server location problem. As the location of mirror servers affects the performance of a content delivery service, it is necessary to find a better server placement. Another method is based on link protection. It is desirable to protect all links so that the failure probability of a link is sufficiently small by sufficient backup resource and rapid recovery system. However, link protection costs are high; therefore, it is important to find the smallest number of links to be protected so that a network resulting from failures of any non-protected links provides users connectivity. In this paper, we address a network design method by combining server placement method and link protection method. Even if a network cannot be made reliable simply by server placement method or simply by link protection method, the combination of them may make the network reliable. Thus, our purpose is to find a server placement and a set of links to be protected such that, even if any non-protected links fail, all connected components include specified servers in the resulting network. First, we formulate this reliable network design method by combining server placement method and link protection method and show that this problem is NP-hard. In addition, we propose a polynomial-time algorithm to solve the problem that the number of the protected links is restricted to zero, and a polynomial-time algorithm to solve the problem that the number of the simultaneous link failures is restricted to one. Furthermore, we evaluate the algorithm by applying it to the topology of actual networks and show that the proposed algorithms work well.