We are setting up GoldenGate to keep 3 of our Oracle databases in sync. Regarding configuration:
We plan to have an instance of GoldenGate running on all 3 nodes. Each node will run an extract on (local) Transaction/Redo Logs to create local Trail files, then run Replicat (locally) to the other 2 databases.
These connections are using TCP/IP so there should not be any lost packets? If Replicat running over the network sends and fails, will GG reattempt until it does or just move on? (Resulting in out of sync?)
DataPump/ Checkpoint process seems to be industry standard for replication over the network remedying the above process.
What are the advantages to using the DataPump/Checkpoints in this process vs what I described above?
Does the Datapump/Checkpoint process add additional syncing for redundancies ?