Why mpls is more efficient than traditional ip? Please explain from the working principle.

Traditional ip is based on the router cpu, that is, after the data packet arrives, remove the three-layer ip header, check the destination ip address and mask, and then match with the routing table one by one, according to the longest mask matching principle. Generally speaking, the patrol path is based on software search.

Mpls uses fixed-length labels for forwarding, that is, after the data packet arrives, the 2.5-layer mpls packet header is disassembled to find the label value, and then it is replaced and forwarded directly according to the label forwarding table, which is more efficient. And because the tag length is fixed, it can be processed by asic and np, which further improves the forwarding efficiency.