System and method for implementing a block-based backup restart

摘要

A system and method for block-based restarts are described. A data storage system interfaces with one or more nodes of a network file system on which a volume is provided in order to read data stored on the volume on a block-by-block basis. Backup data sets capable of recreating the data on the volume are generated from the data blocks read from the volume. The system can interface with a backup memory resource and write the backup data sets to the backup memory resource in a sequential order. As the backup data sets are generated and written to the backup memory resource, restart checkpoints for the data set are also regularly generated and stored for use in restarting the backup process in the event of a recoverable failure in the transfer.

主权项

1. A method comprising:
reading, by a storage server implementing a backup session, data at a physical block level from a volume associated with one or more nodes of a network; generating backup data sets from read data read from the volume; writing the backup data sets to a backup memory resource in an ordered sequence; generating a plurality of checkpoints for the backup data sets, wherein a checkpoint, comprising an identifier of a latest block number of the read data read from the volume for writing to the backup memory resource, is generated after a programmed interval during the backup session; storing the plurality of checkpoints within a checkpoint location; and performing a backup restart in response to detecting a failure in the backup session, wherein the performing a backup restart comprises:
retrieving the plurality of checkpoints from the checkpoint location;selecting a restart checkpoint from the plurality of checkpoints; andrestarting the backup session to generate additional backup data sets from the read data beginning at a block identified by the restart checkpoint.