Distributed recovery block based fault-tolerant multicasting
Gul N. Khan, Gurdeep Singh Hura, Gu Wei · 2004
This paper presents a fault-tolerant and adaptive multicasting method that employs a modified distributed recovery block (DRB) approach. The section of a distributed system spanning between the source and destination nodes is partitioned into series of overlapping DRB groups on the multicasting paths. Each DRB group consists of three nodes: a current node and two successor nodes. The source node becomes a current node in the first DRB group and it partitions the destination node-list to form the message header for delivering the message to its successor nodes. Each successor node receives the message, executes a recovery block try and performs the acceptance test. The successful successors become the current nodes of next DRB groups and update their destination lists. A prototype version of the method is simulated for 2D mesh and hypercube topologies to evaluate its performance.