The important thing to understand about loopback interfaces is that they are virtual interfaces. They have the characteristics of interfaces in that you can assign addresses to them, can route packets to them, can source packets from them. But they are virtual and have no physical implementation, so there is no cord that can get unplugged and take them down. So one big advantage of loopback interfaces is that they remove phycial dependencies that might impact real interfaces.
So to answer your question about why you would want to create a loopback interface, they are usually more stable than a physical interface. If you have a router that has two serial interfaces and you can get to the router via either of the interfaces, if you telnet to serial 0 it only works if that interface is up. But if you telnet to loopback 0 then it should work as long as either of the serial interfaces is up. So loopback interfaces are frequently used as the address used to manage the router, frequently used as the source address of management packets, or logging packets, and things like that.
There are really not any praticular ramifications other than the ramifications of adding any interface on a router. If you want the address to be reachable you must make sure that your routing protocol advertises it. There is not anything about loopback that says whether to use private addresses (172.16.0.0 etc) or public addresses. Either kind of address works well on loopbacks. Choosing whether to use public or private would depend on what you were attempting to accomplish. If you want it to be known at the ISP then it should be public. If you want it known only inside your network then private may be what you want.
HTH
Rick
HTH
Rick
