运营商级网络地址翻译的不敏感排队模型

基于传统网络地址翻译的地址复用技术并不能很好地适应运营商级网络地址翻译(carrier grade network address translation,CGN)体系结构,端口资源利用率和服务质量难以得到保证。为了更好研究和评价CGN在端口复用方面的性能,该文将网络地址翻译端口资源调度状态机分解成排队网络,并建立基于Erlang公式的不敏感排队模型,使用该模型分析了传统网络地址翻译和CGN的性能。实验结果表明:尽管CGN有性能方面的局限,然而该模型相比传统网络地址翻译,能在一定范围内提高IPv4地址利用率。

Insensitive queuing model for carrier grade network address translation

Carrier grade network address translation(CGN) was recently proposed to alleviate the IPv4 address depletion problem.However,some studies have suggested that traditional network address translation(NAT) address sharing techniques cannot accommodate the CGN guarantee of the port/address utilization and quality of service.CGN port sharing is analyzed by decomposing the NAT state machine to a queuing network with an insensitive queuing model used to analyze the performance of a traditional NAT and a CGN.The results show that though there are inherent drawbacks in the CGN,it achieves higher utilization of IP addresses and lower blocking probability than the traditional NAT.