Distributed network address translation-protocol translation over active network

Abstract

IPv6 is proposed as a new version of the internetworking protocol designed to address the scalability and service shortcomings of the current standard. During the coexisting period of IPv4 and IPv6, IPv6 would not be compatible with IPv4. Programs and systems based on one protocol cannot communicate with those based on the other. Consequently, it is necessary to develop a smooth transition mechanism that enables applications to continue working while the network is being upgraded. Among these translation mechanisms suggested so far, the Network Address Translation - Protocol Translation is applicable in general situation. However, NAT-PT has several limitations. Within these limitations, as the load of translation requests increase, edge router may become a performance bottleneck, due to the topology limitation in NAT-PT. We can solve this problem by placing several active NAT-PT routers. In this thesis, we propose and simulate three methods that translate packets using fixed probability, dynamic probability and utilizing resource to the maximum. According to the simulation result, the fixed probability method can be used when the number of translation requests is small amounts, but performance is decreased as translation request increase. In the both cases of dynamic probability method and maximum utilization method, if processing load is placing limit on its own capability, the packet classifier distributes its own load to upper node by decreasing probability of translation. These methods show good performance in general. So, these methods fit in network that has high translation load. Consequently, by applying suggested mechanism to the IPv6 network, we could distribute NAT-PT translation load efficiently and flexibly as well as remove critical single point of failure when using one edge NAT-PT router.