A productive system’s hardware configuration can be changed by using HANA System Replication. With minimized or zero downtime the hardware can be exchanged and the HANA version can be upgraded.
In a primary system to one or numerous secondary systems all modifications to persistent data are replicated by HANA System Replication synchronously or asynchronously. With different hardware equipment, configuration, or with hardware distributed by the hardware partner the secondary system can function. Different virtualization configurations can be found in the primary and secondary systems. With a higher HANA version the secondary than the primary system too can run.
From the primary system to the secondary system the runtime of the takeover will be the downtime for changing the hardware. Typically in the range of very few minutes this occurs.
- The same numbers of worker hosts are present in the primary and secondary systems. With no standby host on single side the number of standby hosts can differ.
- Via virtual IP address (es) or virtual hostname(s) the productive system should be accessible, from the viewpoint of the HANA clients, this can be swapped atomically.
- The same byte swap ordering (little/big endian) are present in the primary and secondary system.
To exchange the hardware with minimized downtime follow the below given steps. For registration activities and System Replication activation either use HANA shell or HANA Studio commands.
Initial Configuration (during primary online operation):
- As in the primary system with the same number of worker hosts the secondary system should be setup.
- For System Replication the primary system should be enabled.
- At the primary system the secondary system should be registered. In case of ~ less than 1 GBit/sec connection or slow network connection the asynchronous replication should be used.
- The secondary system should be started
- To the secondary all persistent data will be automatically transferred by the primary and via log shipping the replication of the changes should be started immediately.
- As shown with the system view wait till the transfer of all data has been completed PUBLIC.M_SERVICE_REPLICATION
- With the view PUBLIC.M_SERVICE_REPLICATION if the replication is still active should be checked
- For the client the primary system should be made inaccessible e.g. hostnames or the IP addresses are switched
- The secondary should be in sync with the primary and this should be checked in case of asynchronous replication. The column LAST_LOG_POSITION_TIME of PUBLIC.M_SERVICE_REPLICATION should be viewed
- The secondary system should be taken over.
- The old primary should be stopped and the System Replication configuration should be unregistered
- In the new primary System Replication should be disabled
On the primary system and secondary system ensure that clients cannot work in parallel.
With different configuration, hardware equipment or hardware the operation of HANA System Replication taking primary and secondary systems is distributed by multiple hardware partners. To hardware exchange or version upgrades the above described conditions are not restricted. This operation can be too long term based operation.