This tip applies to Jet 3.5 and Jet 4.0 (it even applies to Jet 3.0 but only for indirect synchs). There is no code... consider this a "concept" tip if you are using replication.
The basic principle is to do, on your server that hosts indirect or internet synchronization (so it has a Jet Synchronizer, etc.), you do not just keep a single copy of a replica, you keep multiple replicas from the same set, all managed by that same Synchronizer. Set the 'internal' schedule for them synchronize regularly so that you know they are all up to date.
Remember that anytime you do an indirect or internet synch, you are NOT synchronizing to a replica, you are going to the synchronizer and saying "find me a partner." It looks for a valid member of a replica set and uses it. Thus, if you have more than one there and one is marked invalid for some reason (one of those awful timeouts, for example) then it will pick another.
Many people have noted that indirect/internet synchs optimize a bit *too* much.... if they think a single replica is no good then that one is never tried again; yet if you do a direct synch then Jet figures out it is still good and restores it to its former glory. People complain about having to do a direct synch over a WAN.... yet if you have a replica farm then direct synchs happen all the time, and replicas with such problems will constantly be "fixed" by these direct synchs..... and thus as you add replicas the info is freely propagated.
So, the advantages are:
a) synchs will happen more reliably since there will always be a replica there to do the work
b) When bad things happen to a replica, they will often be "fixed' by the farm.
The only disadvantage is the extra space..... which is no big deal at all.these days. So make a replica farm today!!!!