In large-scale parallel computing systems, machines and the network suffer from non-negligible faults, often leading to system crashes. The traditional method to increase reliability is to restart the ...