A floating IP address is a concept within OpenStack. These are public IP addresses that can be dynamically mapped to instances in a cloud. This is realized by SNATaaS in a one-to-one mapping of a public IP address from an IP address pool to a private IP address associated with an instance. Internally, the one-to-one mapping is implemented by NAT, see Network address translation
A network interface can have both a floating and a private IP assigned to it at the same time. The private IP address is used to access the instance within the private network, and the floating IP address from public networks. These addresses are resources defined within a region.
Floating IP address can be used to create architectures without any single points of failure. In a highly available architecture, it is necessary to implement a mechanism with automatic failure detection of the active server and reassignment of the floating IP address to a standby server.
In Neutron, an IP address can be associated to any number of virtual machines in a shared manner, known as a virtual IP (VIP) or shared IP. A virtual machine is allocated a default Neutron port to which its default IP address is assigned.
A VIP can be paired with the same port, so that the virtual machine has both its default IP and shared IP on a single interface. The VIP is associated with a a floating IP on the external network. This technique is used in redundant node configurations, such as load balancing.
How to manage floating IPs are described in Configure Network