Abstract: A network attached storage (NAS) cluster can run with a set of heterogeneous hardware nodes, where not all nodes in the cluster have access to the same target connectivities. In one aspect, network data management protocol (NDMP) sessions can be redirected from a first node of the NAS cluster to a second node of the NAS cluster to balance NDMP load across the NAS cluster and improve resource utilization across cluster. Further, the NDMP load can be actively monitored to balance it continuously to increase resource utilization efficiency.

Abstract: A network attached storage (NAS) cluster can run with a set of heterogeneous hardware nodes, where not all nodes in the cluster have access to the same target connectivities. In one aspect, network data management protocol (NDMP) sessions can be redirected from a first node of the NAS cluster to a second node of the NAS cluster to balance NDMP load across the NAS cluster and improve resource utilization across cluster. Further, the NDMP load can be actively monitored to balance it continuously to increase resource utilization efficiency.

Abstract: Performing parallel file restoration from a serialized storage device using a network data management protocol is presented herein. A system can comprise a reader, a parser component, and a set of writers. The reader can be configured to copy data from a tape drive to memory buffers as buffered data. The parser component can be configured to determine, within the buffered data, a location of a file header record representing a file that has been stored in the tape drive, and generate, based on the file header record, a header work item for facilitating restoration of the file in a file system. Further, the set of writers can be configured to create, based on the header work item using the buffered data, the file in the file system.

Abstract: Implementations are provided herein for a self-healing scalable ring communication topology that enables a multi-stream restartable backup. A standard network management protocol tree-walk process can be altered to support parallel tree-walk. Parallel tree-walk provides for splitting a backup directory among multiple sessions, where each session, in parallel, can walk the tree of portions of a backup directory and stream the results to separate backup storage devices. By allowing multiple backup sessions to work together to back up a single root directory, the backup process becomes more scalable to very large scale data storage systems. In addition, if a single stream of a multi-stream backup experiences a failure, only that stream need be restarted and other streams of the backup can continue without or with very little interruption.

Abstract: Implementations are provided herein for incremental backup using a change tree. A change tree is a database file for tracking file system changes of a specific dataset, such as a directory of a file system. A backup process can create and/or initiate a change tree when creating a backup snapshot of a dataset. After the change tree is created, all file system changes inside the dataset can be tracked. The next incremental backup can then take advantage of the change tree to backup changes without traversing the entire dataset. Thus, incremental backups can be more efficient and are better able to keep up with tape streaming speed.