Dynamic management of highly replicated data

An efficient replication control protocol, called the dynamic group protocol, for managing replicated data objects that have more than five replicas is presented. Like the grid protocol, the dynamic group protocol requires only O( square root n) messages per access to enforce mutual consistency among n replicas. Unlike other protocols aimed at providing fast access, this protocol adapts itself to changes in site availability and network connectivity, which allows it to tolerate n-2 successive replica failures. The availability of a replicated object consisting of n replicas managed by the dynamic group protocol when the n replicas are on the same LAN segment, is evaluated under standard Markovian assumptions and found to equal that of an object with the same number of replicas, managed by the dynamic-linear voting protocol.<<ETX>>