We have to read a lot of historic data (2+ TB) and it is taking too long because of CPU limitations (can't get connector to use all 4 cores), so we are thinking about having multiple instances of the connector to import different portions of the historic data but only one instance to do the change streams but we can't find a clean way to configure this since it seems all instances would automatically switch to the change stream once they are done with their corresponding Copy Existing portion. Thanks for any advice or help you can give us with this.