文摘
Current generation routers are being replaced by scalable next generation routers in order to satisfy the growing demand of Internet traffic in core networks. The next generation routers, on the hardware side, are provided with a very large switching capacity, high speed interfaces and additional computing and memory resources. There is a need to re-design the router software in a distributed way in order to efficiently exploit advanced hardware features of the next generation routers. One of the key functions of a router is to manage routes obtained from different routing protocols. The current centralized architecture of the routing table manager will likely face difficulties to handle the growth of routing tables and path computation requests, particularly for high traffic-density networks. This paper proposes two distributed IGP route management schemes for routers. The first scheme enhances the scalability at the control card level. The second one uses the parallel processing of the switch fabric in order to increase the handled number of routes by adding more line cards. Scalability evaluation is done for the two distributed schemes and a centralized architecture in terms of CPU cycles used and required memory.