Planning Parallel backup paths and backup servers
Data Protection for SAP can use several communication links between IBM Storage Protect™ clients to control alternative backup paths and alternative backup servers. This feature can increase throughput by transferring data over multiple paths simultaneously or to and from several servers in parallel. It can improve the availability of the IBM Storage Protect client to server communication and enable disaster recovery backup to a remote IBM Storage Protect server.
Each path in the initSID.utl profile is defined by a server statement. These correspond to definitions in the IBM Storage Protect client system option file dsm.sys.
The server 1..n statement denotes IBM Storage Protect servers that are defined in the Data Protection for SAP profile. This definition corresponds to the statement SERVERNAME server 1..n in the IBM Storage Protect client option file or files. These servers are identified by their TCPSERVERADDRESS and can be on one system (multiple paths) or several systems (multiple servers).
SESSIONS denotes the number of parallel sessions that Data Protection for SAP schedules for the path.
If only one path is used, SESSIONS must be equal to MAX_SESSIONS, which specifies the total number of parallel sessions to be used (equivalent to number of tape drives or management classes). Data Protection for SAP attempts to communicate with the IBM Storage Protect server by using the first path in the profile. If this attempt is successful, Data Protection for SAP starts the number of parallel sessions as specified for this path.
If the attempt is unsuccessful, this path is skipped and Data Protection for SAP continues to the next path. This process continues until as many sessions are active as were specified in the total session number (MAX_SESSIONS). If this number is never reached (for example, because several paths where inactive), Data Protection for SAP ends the backup job.
Backup hdbbackint processes
When SAP HANA runs a database backup operation through Data Protection for SAP HANA, a dedicated hdbbackint process for each backup object is started. It is possible to use multiple sessions in a single hdbbackint process, and each hdbbackint process transfers a single object.
You do not need to configure multiple sessions for database backup operations. Multiple hdbbackint processes run in parallel, and all objects are backed up in parallel. The number of objects depends on the number of SAP HANA nodes in the environment. All nodes host an index server that holds the payload of the database. The backup objects from the index servers are typically the largest objects. The controller node hosts a name server, a statistic server, and other services. Backup objects from these additional services are typically smaller than the index server. If the backups must go directly to tape, then the number of mount points must include the additional services.
The Data Protection for SAP HANA parameters SESSIONS and MAX_SESSIONS define the upper limits of parallel streams that can be run. The SAP HANA configuration parameter 'parallel_data_backup_backint_channels' defines the number of parallel sessions or channels that are opened by SAP HANA. When you configure Data Protection for SAP HANA by using 'setup.sh', the script prompts for the number of sessions to use. This value is used for the SESSIONS (for the single server that is configured by 'setup.sh'), MAX_SESSIONS, and 'parallel_data_backup_backint_channels' parameters. In case one of these parameters is changed after the initial configuration, the other parameters might depend on it and must be updated as well.
The 'parallel_data_backup_backint_channels' parameter has no effect when the size of the HANA service that is subject to backups is below the value of the SAP HANA parameter 'parallel_data_backup_backint_size_threshold'. When running 'setup.sh', the 'parallel_data_backup_backint_channels' parameter is set automatically to a value of 2 GB. If necessary, this value can be adjusted in the header of the /opt/tivoli/tsm(tpd_hana/setup.sh file). In future invocations of the 'setup.sh script, the new value takes effect.
Redo log files
Configure multiple sessions for a single hdbbackint process for backing up redo log files. When you are using the redo log copy feature of IBM Storage Protect for Enterprise Resource Planning, each redo log file is saved simultaneously in multiple storage pools on the IBM Storage Protect servers. Ensure that you have the sane number of sessions opened as redo log copies that are created to be able to duplicate and transfer data.