Sign In

Search <book_title>...

Veritas NetBackup for Hadoop Administrator's Guide

Last Published: 2018-05-10

Product(s): NetBackup (8.1)

NetBackup for Hadoop terminologies

The following table defines the terms you will come across when using NetBackup for protecting Hadoop cluster.

Table: NetBackup terminologies

Terminology	Definition
Compound job	A backup job for Hadoop data is a compound job. The backup job runs a discovery job for getting information of the data to be backed up. Child jobs are created for each backup host that performs the actual data transfer. After the backup is complete, the job cleans up the snapshots on the NameNode and is then marked complete.
Discovery job	When a backup job is executed, first a discovery job is created. The discovery job communicates with the NameNode and gathers information of the block that needs to be backed up and the associated DataNodes. At the end of the discovery, the job populates a workload discovery file that NetBackup then uses to distribute the workload amongst the backup hosts.
Child job	For backup, a separate child job is created for each backup host to transfer data to the storage media. A child job can transfer data blocks from multiple DataNodes.
Workload discovery file	During discovery, when the backup host communicates with the NameNode, a workload discovery file is created. The file contains information about the data blocks to be backed up and the associated DataNodes.
Workload distribution file	After the discovery is complete, NetBackup creates a workload distribution file for each backup host. These files contain information of the data that is transferred by the respective backup host.
Parallel streams	The NetBackup parallel streaming framework allows data blocks from multiple DataNodes to be backed up using multiple backup hosts simultaneously.
Backup host	The backup host acts as a proxy client. All the backup and restore operations are executed through the backup host. You can configure media servers, clients, or a master server as a backup host. The backup host is also used as destination client during restores.
BigData policy	The BigData policy is introduced to: Specify the application type. Allow backing up distributed multi-node environments. Associate backup hosts. Perform workload distribution.
Application server	Namenode is referred to as a application server in NetBackup.
Primary NameNode	In a high-availability scenario, you need to specify one NameNode with the BigData policy and with the tpconfig command. This NameNode is referred as the primary NameNode.
Fail-over NameNode	In a high-availability scenario, the NameNodes other than the primary NameNode that are updated in the `hadoop.conf` file are referred as fail-over NameNodes.

Table: Hadoop terminologies

Terminology	Definition
NameNode	NameNode is also used as a source client during restores.
DataNode	DataNode is responsible for storing the actual data in Hadoop.
Snapshot-enabled directories (snapshottable)	Snapshots can be taken on any directory once the directory is snapshot-enabled. Each snapshot-enabled directory can accommodate 65,536 simultaneous snapshots. There is no limit on the number of snapshot-enabled directories. Administrators can set any directory to be snapshot-enabled. If there are snapshots in a snapshot-enabled directory, it can cannot be deleted or renamed before all the snapshots are deleted. A directory cannot be snapshot-enabled if one of its ancestors or descendants is a snapshot-enabled directory.