VRRP (Virtual Router Redundancy Protocol) / HSRP (Hot Standby Router Protocol) / GLBP (Gateway Load Balancing Protocol)
What it is for
A host has one default gateway. If that router dies the whole subnet is cut off, and the host will not switch by itself.
★★★ VRRP's answer: two routers share a virtual IP that hosts use as their gateway. The master owns it normally and a backup takes over when it fails -- entirely invisibly to the hosts.
★ The virtual MAC is the key
★★★ A virtual IP alone is not enough. A host's ARP cache holds "gateway IP -> some MAC"; if the MAC changed on failover the host would keep sending to the old one until the entry ages out, minutes later.
★★★ So VRRP also defines a virtual MAC (00:00:5e:00:01:VRID). Master and backup use the same MAC, so a host's ARP cache remains correct after failover.
★★ The new master also sends a gratuitous ARP so switches immediately point that MAC at it.
★ Common misconceptions
★★★ "VRRP lets both gateways share the load" -- it does not. Only the master forwards; the backup merely waits. Load sharing needs GLBP (Cisco-proprietary) or alternating master roles per VLAN.
★★★ "Hosts must re-ARP on failover" -- they need not, which is the point of the virtual MAC: it did not change, so their caches stay correct.
★★ "VRRP and HSRP are the same" -- functionally almost, but VRRP is the standard (RFC 5798) while HSRP and GLBP are Cisco-proprietary. Mixed vendors must use VRRP.