what happens during failover?

[Former Member]

What happens during a failover scenario while the standby server is reading the data and log volumes? Do the users that were connected to the node that failed kick out? what happen to the transactions that were not commited or saved to Data volume or Logs? Since HANA doesn't share Disk is this more prone to data lost than a standard RDBMS?

[Former Member]

Hi All,

Fundamentally , When an active host fails ( identified by master name server ), a standby automatically takes a place since the standby host may take over operations from any of the primary hosts it needs shared access to all the DB volumes. This can be accomplished by a shared , networked shared server , by using a distributed file system.

Through  synchronous replication and  continueslog shipping  we can achieve near zero  RTO.

Regards,

Sankar

[Former Member]

Eduardo,

What happens during node failover to standby node

To be able to take over the database partition from a failed node, the standby node has to load the savepoints and database logs of the failed node to recover the database partition and resume operation in place of the failed node.  The master node  recognizes the node failure and directs the standby node to take over from the failed node. To recreate database partition  in memory to be able to take over the role of within the cluster, standby node reads the savepoints and database logs of failed node from the shared file system, reconstructs the savepoint data in memory, and re-applies the logs so that the partition data in memory is exactly like it was before node failed. After this is complete, standby node is in operation as an active Index Server, and the database cluster has recovered. Data access in non-failed nodes can continue.

    The users connected to the failed node will experience outage and will be re-connected once the standby node becomes slave. Only the committed transactions are written to disk during the save point process.

[Former Member]

Note that committed transactions are on a shared log disk in A HA scenario. If a DB node has power outage then uncommitted transactions will receive a SQL error at the app server. Like any other DB.

Then the master name server, when it recognizes a node is down, instructs the standby to mount the disk of the failed node and start the index sever. It will load the last save point and replay logs. The RPO is near-zero.

Also note that if you require a low RTO, use system replication. With synchronous replication the transactions are committed on both sides at the same time.