A recovery method for communication control processors
H. Yamashita · 2003
An efficient recovery method using checkpointing and journaling in communication control processors is presented. In the method, each process journals all received messages on stable storage together with indications of sending messages, and can recover its state independently by using them. Two ways of further reducing the recovery time are proposed: (1) establishment of checkpoints when the total amount of collected journals reaches a calculated value based on the recovery time requirements; and (2) execution in the recovery mode of the copy codes of a failed process in which message-input and output instructions have been modified.>