SAP HANA provides three different mechanisms to increase the availability.
- Host Auto-Failover – At least one standby node added to a SAP HANA system. These nodes are configured to work in standby mode (SAP HANA scale-out)
- Storage Replication – The storage used on the SAP HANA node replicates all data to another SAP HANA node. This replication works without a control process from the SAP HANA system. The Storage Replication is provided by hardware partners.
- System Replication – SAP HANA replicates all data by using their own feature to a secondary SAP HANA node. Data is constantly pre-loaded on the secondary SAP HANA node. (SAP HANA scale-up)
With this SAP HANA Recovery kit SAP HANA systems can be controlled through an activated system replication in SIOS LifeKeeper.
The Recovery Kit is able to start the SAP HANA system on all nodes and perform the Takeover process of system replication. To ensure the functionality of the SAP HANA system, the following processes and states are continuously tested:
- SAP Host Agent on all nodes
- SAP Start Service of instance (sapstartsrv) on all nodes
- State of SAP HANA database on all nodes
- State of system Replication mode (primary on active node, sync|syncmem|async on secondary nodes)
If the required processes or the databases are not active, it will try to start them again. In case of an unsuccessful restart of the processes, the status of the SAP HANA resource is marked as faulty (OSF) for the corresponding server. Manual intervention is necessary to correct this condition and to set the state of SAP HANA resource to “in Service” and monitored (ISP).
In case of an invalid state of the System Replication, the SAP HANA resource is also placed in the state faulty (OSF). It has to be decided with the help of a database administrator if a takeover is to
be performed or how the System Replication mode should be corrected.
In the case of an invalid state of the System Replication, the SAP HANA resource is also placed in the state faulty (OSF). It has to be decided with the help of the database administrator if a takeover is to be performed or how the System Replication mode is to be corrected.
In carrying out the “Out of Service” action only the database on the primary node is stopped. The database on the secondary node remains active as does the System Replication mode. If the entire SAP HANA system is to be shut down, the control flag “!volatile!noHANAremove” must first be deleted by using:
/opt/LifeKeeper/bin/flg_remove –f “!volatile!noHANAremove”