Preserving the Fault-Containment of Ring Protocols Executed on Trees

Reliable and fault-tolerant distributed systems have been attracting more and more attention (see Autonomic Computing Project by IBM, http://www-03.ibm.com/autonomic/). A self-stabilizing protocol is a fault-tolerant protocol that guarantees autonomous recovery from any number of and any type of faults that can affect the data stored locally at some process(es). If the impact of the faults can be contained to the affected process(es) and some of its immediate neighbors, then the protocol is also fault-containing. We present a new method, called causal simulation, which preserves the fault-containing property of ring protocols executed on trees.