In networks of clusters, a collection of computing nodes works innconcert to solve large application problems. The nodes exchange messagesnand coordinate their efforts by sending and receiving messages through anrouting network. Consequently the achieved performance of such machinesndepends critically on that of their routing networks. Commonly there isna routing table at each node of the network. When a message is to bensent, the routing table is consulted for the direction. Multi-node labelninterval routing scheme (MNLIRS) is a routing framework which provides ancompressed routing table for efficient routing with fault-tolerance. Innthe scheme, fault-tolerant tables can be introduced to improve thencapability of end-to-end fault-tolerant delivery. In this paper, thenalgorithms to generate the fault-tolerant table are presented andnevaluated
展开▼