9.7.2 Hot-failover of Transaction Manager using shared file system
This is the simplest of all Transaction Manager HA setups. It requires all cluster members to have access to a shared file system on NAS or SAN, where the transaction logs are stored, see Figure 9-13.
Normally, every cluster member runs its own Transaction Manager. When a failover occurs, another cluster member will be nominated to perform recovery processing for the failed peer according to the Clustered TM Policy of the core group. The recovery process completes in-doubt transactions, releases any locks in the back-end database and then releases the transaction logs. No new work will be performed beyond recovery processing. The Transaction Manager will fail back when the failed server is restarted.
Figure 9-13 ND V6 transaction service high availability using NAS