A mechanism and method for maintaining a consistent state in a non-volatile random access memory system without constraining normal computer operation is provided, thereby enabling a computer system to recover from faults, power loss, or other computer system failure without a loss of data or processing continuity. In a typical computer system, checkpointing data is either very slow, very inefficient or would not survive a power failure. In embodiments of the present invention, a non-volatile random access memory system is used to capture checkpointed data, and can later be used to rollback the computer system to a previous checkpoint. This structure and protocol can efficiently and quickly enable a computer system to recover from faults, power loss, or other computer system failure.