Always On Gold v4
This architecture favors local resiliency/redundancy first and remote locations only when and an entire location is offline or fails.
This architecture enables geo-distributed writes where no/low conflict handling is expected
The Always On Gold architecture requires intervention to move between locations but all data will be replicated and available to the application when failover is initiated with full capacity.
The Always On Gold architecture includes the following:
- Four BDR data nodes (two in location A, two in location B)
- One BDR witness node in location C (optional but recommended)
- Four HARP-Proxy nodes (two in location A, two in location B)
- Two Barman nodes (one in location A, one in location B)