Replication continues to backlog with hundreds of new backups coming in every day. One example of this is a SQL server with multiple DBs performing transaction logs every 15 minutes.
Customer has hundreds of jobs performed per day. Replication just doesn't seem able to keep up with the volume of jobs.
In the [Replication] section of the master.ini on the source add the option to disable threaded send for VCD.
Alternatively you can lower the number of send threads and still increase the MaxConcurrent count without overloading the system.
By turning off this option we are able to increase the number of workers much more than we would other wise. The threaded component with a large number or workers could overwhelm an appliance.