Middleboxes provide a wide range of functions and realize the network. If network functions fail and cannot be recovered, network services which utilize them cannot continue. In order to improve survival probabilities of functions, this paper presents six strategies which obtain an assignment of backup servers to network functions. Backup servers have capabilities of protecting functions and recovering some of them. In the previous work, the range of server capabilities, where an assignment can be obtained, is limited. By adopting the idea of temporarily dividing each server into a set of small servers, we expand the range of server capabilities in the first two strategies. We also consider the applicability of one of the first two strategies through deriving three theorems. Furthermore, in the last four strategies, we enable to obtain an assignment without any limitation of server capabilities by utilizing the knowledge of the first two strategies. To compare our strategies, we also provide a baseline strategy, which is a greedy algorithm and try to balance burden of each server. The numerical results show that in all the settings, more than one of our strategies provides the higher survival probability than the baseline strategy within an allowable computation time.

中间盒提供了广泛的功能并实现了网络。如果网络功能失败并且无法恢复，则使用网络功能的网络服务将无法继续。为了提高功能的生存概率，本文提出了六种策略，这些策略获得了将备份服务器分配给网络功能的方法。备份服务器具有保护功能并恢复其中一些功能。在以前的工作中，可以获得分配的服务器功能范围受到限制。通过采用将每个服务器临时划分为一组小型服务器的想法，我们在前两种策略中扩展了服务器功能的范围。我们还通过推导三个定理来考虑前两种策略之一的适用性。此外，在最后四个策略中，我们可以利用前两种策略的知识来获得分配，而不受服务器功能的任何限制。为了比较我们的策略，我们还提供了一个基线策略，这是一种贪婪算法，并试图平衡每个服务器的负担。数值结果表明，在所有设置中，在允许的计算时间内，不止一种策略比基线策略提供更高的生存概率。