Depending on how many tables you have I would also look into having the triggers simply put the primary key for each table into a table that is used to drive a separate process that will iteratively move new records over to the off-site machine. This can be much easier and less risk/trouble than replication if you don't already have experience with that subsystem, which it sounds like you don't. I have used this mechanism many times in the past to do what you are looking to do, and also to do other things such as incremental data warehouse populations.