Multi-node backup using Data Protection for SAP HANA

Answer

When the SAP HANA database is in a multi-node environment, then the Data Protection for SAP HANA can be set up for the backup to run on all the nodes.

The files involved in this configuration are the initSID.utl, dsm.opt and dsm.sys.

initSID.utl

There is normally only one initSID.utl file, which is located on the shared file system and accessed by all the SAP HANA nodes. This would contain the settings, for example:

# Sample profile for Data Protection for SAP (R) HANA
#-------------------------------------------------------------------
BACKUPIDPREFIX  PRD___
MAX_SESSIONS     1    # Tivoli Storage Manager client sessions
#REDOLOG_COPIES  1        #default is 1
CONFIG_FILE  /usr/sap/PRD/SYS/global/hdb/opt/hdbconfig/initPRD.bki
#
SERVER         SAP_BACKUP     # Servername, as defined in dsm.sys
   ASNODE         SAPNODE
   SESSIONS           1       # Maximum number of sessions to the server
   #  PASSWORDREQUIRED    YES       # Use a password
   #  ADSMNODE            XXXX      # Tivoli Storage Manager Nodename
   BRBACKUPMGTCLASS    MDB       # Mgmt-Classes for database backup
   BRARCHIVEMGTCLASS   MLOG      # Mgmt-Classes for redo log backup
   # USE_AT   0 1 2 3 4 5 6      # Days when server_a is used for backup
#End of profile  ********************************************************

Note that the values for SESSIONS and MAX_SESSIONS are set to a value of 1 in this example. Although it is possible to set this to a larger value, this will not increase the number of sessions used for the backup. During the SAP HANA backup there will be 1 hdbbackint process for each SAP HANA service. Thus, you would potentially have 4 sessions for the master node and 1 session for each slave node.

dsm.opt
The dsm.opt file would exist on each machine, but would have the same SERVERNAME entry for the files on each SAP HANA node. For example, the dsm.opt file would contain the entry:

SErvername SAP_BACKUP

dsm.sys
The dsm.sys file would be in the /opt/tivoli/tsm/client/api/bin64 directory on each SAP HANA node. This would contain a unique nodename and at a minimum the following parameters:

SERVERNAME SAP_BACKUP
TCPSERVERADDRESS xxx.xxx.xxx.xxx
NODENAME NODE1
PASSWORDACCESS GENERATE

where each machine would have a NODENAME value that is unique to that SAP HANA node.
It would also be possible to utilize the stanza that exists for the Backup/Archive filesystem backups, since the nodename that is specified in the dsm.sys will only be used to connect to the Tivoli Storage Manager Server. All the SAP HANA data that is stored on the Tivoli Storage Manager Server would exist under the ASNODENAME that is specified in the .utl file.

Tivoli Storage Manager Server
On the Tivoli Storage Manager Server it would be necessary to give proxy authority to the nodes, for example:

GRANT PROXYNODE TARGET=SAPNODE AGENT=NODE1, NODE2, NODE3...

specifying the agent nodenames for each unique nodename that would be involved in this SAP HANA environment.

The Maximum number of mount points will need to be set to accommodate the number of sessions/mounts that are needed for the backup. This value would be set within the node definition on the Tivoli Storage Manager Server for the ASNODENAME. For example:
UPDATE NODE SAPNODE MAXNUMMP=8