NetBackup for Greenplum technical overview
NetBackup for Greenplum uses the gpbackup and gprestore commands to backup and restore Greenplum databases. These commands are not part of the Tanzu Greenplum database software installation. The Tanzu Greenplum Backup and Restore software package contains the gpbackup and gprestore commands so it must be installed with the database software. The NetBackup for Greenplum agent is a command-line plug-in for gpbackup and gprestore. (The NetBackup for Greenplum plug-in command-line is nbgpbackup.)
Greenplum is a distributed database that runs on multiple hosts. The database uses one coordinator host and multiple segment hosts. Each segment host can contain multiple segments. Each host must have the NetBackup client software installed. The Greenplum agent uses the Datastore policy type.
For more information on the Greenplum Database, see your Greenplum documentation.
The gpbackup command calls nbgpbackup multiple times during a backup. On the segment hosts nbpgbackup is called many times to protect a database. For the best performance, the agent attempts to put multiple files from a segment into an image. To achieve this action the agent uses shared memory to pass the data to a nbgpbackup process that communicates with the NetBackup servers. Each NetBackup stream is a separate process that services its shared memory. After the segment host backups are complete gpbackup also calls nbgpbackup multiple times on the coordinator host. But unlike with the segment hosts, each of those calls results in separate backup images.
Each segment on each segment host is backed up into a separate image. To achieve this action the number_of_streams option in the config file must be at least equal to the number of segments per host. If more than one stream per segment is configured, then there can be multiple images for each segment.