Traffic Forwarding via Dual Matching in Large Networks

The patent segments the routing decision process into two distinct matching operations: first matching on destination IP address to determine a next-hop group, then matching on service classification field to potentially redirect to a differing next-hop group. This segmentation allows each matching operation to be simpler and more scalable, avoiding the need for a single complex policy-based routing table that would require exponential space.

The patent introduces an intermediary structure called 'next-hop group' that acts as a mediator between the destination IP address and the final next-hop selection. The FIB table maps destination IP addresses to next-hop groups, and then service classification matching determines the actual next-hop within that group. This intermediary layer enables scalable implementation by decoupling the routing decisions.

The patent divides the routing matching process into two separate operations: first matching destination IP address against FIB table entries, then matching service classification field against policy rules. This segmentation enables the system to handle large numbers of routes because each matching operation works with a subset of the total routing information, fitting within available transistor counts and memory sizes while still supporting large-scale networks.

The patent adds a dimensional layer to the routing process by introducing service classification as a separate matching dimension after destination IP address matching. Instead of trying to match both criteria in a single two-dimensional lookup table (which would require exponential space), the system performs sequential matching along different dimensions, achieving scalability by operating in multiple sequential dimensions rather than a single high-dimensional space.