Virtual IP, also called a circuitless or loopback interface, is a powerful function that can be used in many ways.
Virtual IP provides a way to assign one or more addresses to the system without the need of binding the address to a physical interface. You can use this function when you want to run multiple occurrences of a server bound to different addresses, or if you want to run other services that need to bind to default ports.
Most environments where you might want to use virtual IP are cases where you want to provide multiple paths between the local gateway and the iSeries™ server, for example, load balancing and fault tolerance. In this context, each "path" implies an additional interface, and consequently, an additional, nonvirtual IP address on the iSeries server. The existence of these multiple interfaces should only be visible on the local network. You do not want the remote clients to have to be aware of the multiple IP addresses for the iSeries server. Ideally, you will like them to view your iSeries server as a single IP address. How the inbound packet gets routed through the gateway, over the local network, and to the iSeries server should be invisible to a remote client. The way to accomplish this is by using virtual IP. Local clients should communicate with the iSeries server by any of the physical IP addresses while remote clients see only the virtual IP interface.
The virtual IP environment is for the iSeries server that acts as the server for remotely connected clients. More importantly, the virtual IP address is on a different subnet than the physical interfaces. Moreover, the virtual IP address makes your iSeries server appear as a single host, not necessarily as one attached to a larger network or subnetwork. Therefore, the subnet mask for the virtual IP interface should usually be set to 255.255.255.255.
Because the virtual IP address is not bound to a single physical
interface, the iSeries server
never responds to an Address Resolution Protocol (ARP) request to the virtual
IP address unless you enable proxy ARP for the virtual IP address. In other
words, by enabling proxy ARP, a local interface can respond to the ARP requests
on behalf of the virtual IP address. Otherwise, remote systems must have a
route defined to reach the address. 
In the preceding example, the workstations all point to one
of the 10.3.2 interfaces, on the iSeries server, as their next hop gateway.
When a packet arrives at the iSeries server, it goes through the packet processing.
If the destination address matches any address defined on the system (including
virtual IP addresses), the system processes the packet.
The DNS servers use the addresses of the requested server. In this case, all the addresses represent the same system. The virtual IP function can be used when consolidating multiple systems into one larger system.