To break the bottleneck of slow routing lookup in packet forwarding on a multi-core processor platform
this paper presents a parallel routing lookup algorithm based on the divided multi-branch trie.This algorithm divides a multi-branched trie into several subtrees according to the number of processor cores
and each of them becomes a separate multi-branched trie.In the subtrees
it abolishes prefix lookup and composes a large intermediate node and adopts fixed-step query between the intermediate nodes
within which it uses the binary tries for expression.The experimental results show that this algorithm is characteristic of small number of memory access
fast looking-up speed
small memory space occupation and small update overhead and furthermore
it is applicable to both IPv4 and IPv6addresses.
关键词
并行路由查找算法Trie树多核处理器前缀匹配
Keywords
parallel routing lookup algorithmtriemulti-core processorprefix match