A Non-Blocking Coordinated Checkpointing Algorithm for Message-Passing Systems
Houssem Mansouri, Nadjib Badache, Makhlouf Aliouat, Al‐Sakib Khan Pathan · 2015
This paper proposes an efficient non-blocking coordinated checkpointing algorithm for distributed message passing system which uses transitive dependency information. The processes synchronize their checkpointing activities so that a globally consistent set of checkpoints is always maintained in the system. These algorithms do not require channels to be FIFO (First-In, First-Out) and ensure that each checkpoint taken is part of a consistent global checkpoint. Our scheme also records a minimum number of checkpoints by making sure that only few processes are required to take checkpoints in any execution - it uses very less control-message cost when compared to other related works.