HANA Replication

[SAPSupport]

Hi Team,

We are planning to upgrade the memory of our existing PERSON_NAME appliance and aim to minimize downtime throughout the process.

Currently, the HANA Database System Replication is configured from the Primary Node to the Secondary Node using Log Replay, and from the Secondary Node to the Tertiary Node. The upgrade strategy involves failing over from the Primary to the Secondary Node and upgrading the Primary Node first. During the upgrade, replication will continue with the Tertiary Node. After the upgrade is completed, we will initiate replication from the Secondary Node back to the Primary Node, establishing multi-node replication. Once replication is active, we will fail back to the Primary Node and reconfigure replication from the Primary to the Secondary and from the Primary to the Tertiary Node for multi-node replication. After replication is established, we will perform the Secondary Node memory upgrade, aiming to ensure continuous business operations with the Tertiary Node as a safeguard. Due to limited bandwidth, we do not want to maintain a full replica between the Primary Node and the Tertiary Node.

Please help us understand how to avoid a full replica rebuild when replication switches from the primary node to the tertiary node. Additionally, let us know how we can ensure that, during replication initialization, it will start with either a full replica or a delta update.

------------------------------------------------------------------------------------------------------------------------------------------------
Learn more about the SAP Support user and program here.

[SAPSupport · Accepted solution]

Hi,

Similar to the steps involved in updating the HANA version, you can use system replication to minimise downtime 

https://help.sap.com/docs/SAP_HANA_PLATFORM/4e9b18c116aa42fc84c7dbfd02111aba/ee3fd9a0c2e74733a74e4ad...

Regarding the full sync, you are starting off with multi-tier system replication and then changing to multi-target when you register Dr to primary instead of secondary.
That is fine but will involve a full sync of DR as it takes the data from primary to dr 
The persistence of DR will be incompatible with primary and it wont be able to just sync the delta as it was registered to secondary previously and now is being registered to the primary.

[RajivVMM · Accepted solution]

Thanks for the Quick Support,

[RajivVMM · Accepted solution]

Team, we have sufficient space available in the /hana/log file system, and data load will be minimal during the hardware upgrade activity. While performing the hardware upgrade on Node A, we will not unregister it from Node B. The replication will active from SITEB to SITEA and from SITEB to SITEC. The hardware upgrade is expected to take approximately 15 hours. After the upgrade, when starting replication, will it begin with a full replica or a delta? How can we determine whether Node A's replication will start as a delta or a full replica? Are there any parameter settings that can help us avoid a full replica?

Awaiting for your response, still need clarify on the above query