Automatic Failure Recovery Using Snapshots and Replicas

摘要

In one embodiment, a method of coordinating data recovery in a storage stack with a hierarchy of layers includes, upon an input/output (I/O) request from a layer of the stack, issuing a help response to recover the data from a higher layer in hierarchy order. The method further includes processing the help response, at the higher layer, by issuing a return code of success or further help response to an even higher layer.

主权项

1. A method of recovering data in a storage system, the method comprising:
upon failure to fulfill an I/O request to the primary volume, determining whether the I/O request references data addresses that may exist in a snapshot or replica; consulting a change set to determine whether the referenced data addresses in the snapshot or replica are current; and fulfilling the I/O request by accessing the snapshot or replica, if the referenced data addresses are current, or issuing an error or failure status, if the referenced data addresses are not current.