After you are replicating data, you can measure the effectiveness of your configuration and tune the speed of replication processes.
The default value for each option, which is 4096, is the best practice setting. If you want to improve the performance of a node replication process, try tuning the REPLBATCHSIZE and the REPLSIZETHRESH server options. Change the default values only after you monitor the node replication performance over several sessions. When the default values of the options are increased, the server requires more space in the active log. You might need to allocate space for the active log that is two or more times larger than an active log size that uses the default size of 4096. In addition, the server might require a longer initialization time when you start.
Use a trial and error method to increase the server options. You can increase the server options in any order. Start by increasing one of the options incrementally by 10%. If replication performance does not improve, revert the setting to its original value. Increase the other option incrementally by 10%. Ensure that you monitor active log usage during the first few replication operations to ensure sufficient active log space is available. Larger transactions run longer and use more active log space that causes other server processes to run slowly. If server processes run slowly, reduce the options until the replication and other server processes can complete.