Testing a redundancy set up

Testing a redundancy set up

I was hoping someone could help me further understand something.
I was told that simply pulling an Ethernet cable to force a communication error, and watch the back up system switch on was the wrong way to do this. Instead to power down the primary server. I was hoping someone could shed some light into why pulling the Ethernet cable is the wrong approach.
Thanks

This is a migrated post! Originally posted on 10.05.2016 by user SteveM. Please be aware that information can be outdated.