Research And Implementation Of Forecasting Method For High Volume Forwarding Information Base Of IPv6

It's an inevitable trend that IPv6 will be substitute for IPv4 in Internet. However, the current Forwarding Information Base in router will still expand rapidly despite of IPv6 protocol's adoption. Therefore it's needed to solve the problem of high-speed search to high volume IPv6 Forwarding Information Base in the design of IPv6 core router. For the deployment scale of IPv6 is limited, a real high volume IPv6 Forwarding Information Base doesn't exist currently. Forecast method must be adopted to generate large-capacity IPv6 Forwarding Information Base in the performance evaluation of a variety of related algorithms.For the problem of forecast to high volume IPv6 Forwarding Information Base, CISCO Corporation introduces local random expansion method based on IPv4 and ARAM forecast model IPv6 proposes a parameterized simulation method based on the structure of IPv6 routing table. But the forwarding information in the current IPv6 router hasn't been used to forecast in the above methods. So it's difficult to reflect the current characteristics of address allocation and prefix aggregation in time. This paper propses a new type of forecast method - FPR (Fixed-Point Random) algorithm. Through simulating the attributes of the future high volume IPv6 Forwarding Information Base such as table size, structure and dynamics of changes, This algorithm generated in line with the forecast the future development of the high volume Forwarding Information Base of real IPv6.The main work and contributions include as follows.(1) By the analysis of present IPv6 address architecture, several forecast techniques are introduced for high volume IPv6 Forwarding Information Base based on the extention of IPv4. By the analysis of present IPv6 address architecture, several forecast techniques are introduced for high-volume IPv6 Forwarding Information Base, including IPv4-extension-based and IPv6-router-structure-based forecast methods. Deficiencies of these ways have been listed.(2) It's difficult for current forecast method to reflect the disadvantage of present IPv6 network's address allocation and prefix aggregation, so FPR forecast method is put forward in this paper. The algorithm is based on the analysis of real IPv6 routing table's sample and generates IPv6 prefix in fixed-point random ways. This thesis researches the process of algorithm implementation and formation methods of prefix.(3) Based on real sample of IPv6 routing table in CERNET2, we use FPR methods to forecast and generate high volume IPv6 forwarding table of different scales. And for the architecture features of MPFS(Massive Parallel Forwarding and Switching), performance evaluation methods of IPv6 packet forwarding under this structure is proposed. The above work is of great significance to the evaluation on MPFS architecture. The work in this paper has referential value to some applications, such as the design of large-capacity IPv6 routers, especially performance evaluation to the lookup algorithm of IPv6 Forwarding Information Base in high-performance network processor.