A Replication Set is a two-target Replication configuration where the two targets are aware of each other, and are configured to continue replication in the event of a server failure.
The awareness between the targets allow for one of the targets to transition to the source, and synchronize with the remaining target. You can configure a Replication Set to prioritize which target is responsible for transitioning to the source. For example, a Replication Set can be configured so that the secondary target transitions to a source when both the original source and the primary target are lost. When configuring a Replication Set, all databases must have after-imaging enabled.
Communication between agents is required for Replication Sets. This is known as inter-agent communication. The initial connection between agents occurs during the startup of the agents. After the server has connected to all of the agents, it enters a state for inter-agent connection initialization. When the server is in this state, it sends a message to each agent with the connection information needed to connect to the other agent. When each connected agent receives the message, the agent calls a message dispatch function to handle connecting to the other agent. The listening agents process inter-agent connection requests in the same manner that server-agent connections are processed. After the connected agent has completed the inter-agent network with the other agent, it sends a confirmation message to the server to let the server know that the inter-agent network has been established. Once the server has received confirmation that the inter-agent network has been established from all agents, the server begins startup synchronization. With the new inter-agent communication network, agents can ping each other to determine if the agent is still alive and connected.
The figure below shows the source and target databases, and inter-agent communication in an OpenEdge Replication Set model.
