Proceedings of the IFIP Int'l Conference on Networking (Networking), Valencia, Spain, May 2011.
The "X-day" estimation is March 3rd 2011, which means that IPv4 addresses are running out soon. An IP router has to match each incoming packet's IP destination address against all stored prefixes in its forwarding table at wire speed. The challenges are very great here; not only the routers have to keep up with with the ultra-high Giga bits per second (Gbps) link speeds, they have to keep up with switch to the 128-bit IPv6 address space while the number of prefixes quickly grows. Commercially, many routers employ Ternary Content Addressable Memory (TCAM) to facilitate fast IP lookup. However, TCAMs are power-eager, expensive, and not scalable. Recent studies suggest that keeping the forwarding table in a more efficient SRAM and utilizing pipelined access and trie-based data structures make a better solution. But with IPv6 prefixes, trie depth and the number of nodes can form a bottleneck. We propose a new scalable IPv6 forwarding engine based on a multibit trie architecture that has the potential to achieve a high throughput of 4.9 Giga packets per second.