An Optimality Proof for Asynchronous Recovery Algorithms in Distributed 
Systems 

Information Processing Letters 55, pp. 117 - 121, 1995 

We prove the optimality of asynchronous recovery algorithms for distributed 
systems in the sense that irrespective of the order of roll backs by sites, 
all sites incrementally converge to a unique consistent cut which is the 
"latest" of all past consistent cuts formed by the local recovery points of 
the sites.