Even FIB Decomposition Model Based on Storage and Traffic Balancing

The sharp expansion of the FIB table rapidly aggravates the hardware cost of Line cards(LCs) in the high-performance distributed routers. The storage optimization of FIB becomes a research hotspot. The traffic load of each LC is still very different in the current nonfull backup storage, which has a deep impact on the overall forwarding performance of the routers. An even FIB decomposition model was proposed, namely FEST, aiming for a two-dimension balance in both storage and traffic. Based on the splitting and distribution filters, FEST starts with splitting the root prefixes and utilizes the optimal adaptation algorithm to evenly distribute the routing entries and the traffic to LCs without the modification of the hardware designs of the current LCs. Eventually FEST uses the location routing to determine the location and the forwarding of every packet. The experiment results show that different LCs get very even numbers of routing entries and relatively even traffic in FEST.