Network Loopbacks Why do they work and do not work at the same time?

How many times have you needed to troubleshoot a circuit and required a means of telling if layer one connectivity was present? Enter the good old loopback, which can be hard-wired or virtual. A hard-wired loopback is where redirection of the physical medium takes place so that the output and the input are together. For example, in an Ethernet loopback, this is done by connecting pin 1 to pin 3, pin 2 to pin 6, pin 4 to pin 7, and pin 5 to pin 8. The connection mentioned should theoretically give you a loopback that will provide you with 10\100\1000 Mbps loopback functionality, but in all practicality, the 1000Mbps link will not work due to cross-talk. The other hard-wired loopback is the fiber loopback. The loopback occurs when a fiber patch cable between the input and output of a circuit exists. The loop will redirect all traffic from the transmit (TX) to the receive (RX); this will work for OC3 to OC192 circuits, to go any higher, may work but with the threat of non-linearities occurring within the looped media.

The problem with loopbacks is that they may cause as many problems as they solve. Take, for instance, if your switch is not a non-blocking switch. As soon as you put that loopback in, your switch will go into a packet storm death spiral. Why? I have tried to explain this to a class of Master-level students, and I always get the statement, "I never knew that." The lack of understanding why the death spiral occurs, to me, denotes holes in that student's learning process.

Let me explain what occurs. When a switch receives a packet, it will flood the other ports of the controller, but since it can not see its outgoing packets, the port sending the traffic does not respond to the traffic. However, since you have a loopback in place, the switch port receives its traffic and floods the packets out of the ports, including itself; this keeps occurring until the functionality of the switch is diminished by the loopback.

Hence the loopback works until it kills the switch. That is the reason that you need to do quick troubleshooting on blocking controllers (switches).

I have used loopbacks many times, but as I tell people on managed switches, I make sure that loopback detection is on and loopback prevention is also on because I have seen networks die because of leaving loopbacks up. So the information above is just some helpful advice on using loopbacks.

Thank you,