What is a floating IP?

The internet – plainly put – consists of many computers connected by cables, fibre optic cables, and wireless receivers. They exchange data based on a common ‘language'. This common standard is known as the Internet Protocol (IP). Data is arranged in such a way that computers, which un­der­stand the common protocol, can interpret it. An IP address, also referred to as an 'IP', makes digital devices de­tect­able in a network. It is a crucial pre­requis­ite so that elec­tron­ic data packets can be delivered reliably. The devices com­mu­nic­ate with one another, for example, over the internet. The IP address ensures that data from the sender reaches the correct recipient – for example, from a web browser to a web server or vice versa. An IP address can be assigned to both single and multiple devices at the same time. Likewise, a single device can have multiple IP addresses at the same time. However, in order to be able to un­der­stand exactly what a floating IP is, you first need to know the dif­fer­ence between dynamic and static IP addresses.

When a computer connects to the internet, in most cases the Internet Service Provider (ISP) assigns a dynamic IP address to it. Dynamic IP addresses are the most cost-effective standard for users and providers. They are char­ac­ter­ised by the fact that they are only assigned tem­por­ar­ily and change after a certain time, which is either fixed (e.g. for 24 hours), or is irregular. The user then receives a new dynamic IP address for their computer from the re­spect­ive internet service provider and the previous address will then be signed to a different user.

A static IP, on the other hand, is a fixed address and is per­man­ently assigned to a device. Static IP addresses are found mainly in the web server or e-mail server area, or wherever offers or website content must be ac­cess­ible via a fixed URL , so that users or processes can (re)find them without any problems. Computers in a network or peri­pher­al devices (such as printers) have fixed IPs, so that the in­di­vidu­al devices within the network can easily com­mu­nic­ate with one another.

So that users don’t have to remember complex numbers, it’s possible to assign a domain name to a static IP address e.g. www.example.org. The numerical IP, the 'con­nec­tion number' of a device in the network, is therefore trans­lated into a name that can easily be re­membered. This is generally only reserved for static IPs. It doesn’t make much sense for dynamic IPs since the user changes so fre­quently.

A floating IP is usually a public, routable IP address that is not auto­mat­ic­ally assigned to an entity. Instead, a project owner assigns them to one or more entities tem­por­ar­ily. The re­spect­ive entity has an auto­mat­ic­ally assigned, static IP for com­mu­nic­a­tion between instances in a private, non-routable network area, as well as via a manually assigned floating IP. This makes the entity’s services outside a cloud or network re­cog­nis­able and therefore achiev­able.

In ap­pro­pri­ately con­figured failover scenarios, an IP 'floats' to another active unit in the network so that it can take on the function of a dormant entity without a time delay, and can then answer incoming requests.

Users obtain floating IPs for their projects from different pools that the system ad­min­is­trat­or con­fig­ures and provides as server resources. As soon as a user receives a floating IP, they become the 'owner'. They can assign it to an entity, remove it, and then assign it to another at any time. Even if an entity is ter­min­ated, the user does not 'lose' the as­so­ci­ated floating IP. It remains as a resource and can still be assigned to another entity when needed.

A major reason for using several parallel floating IP pools is that each pool can be operated by another internet service provider or can also be assigned by other external networks. This ensures that the con­nectiv­ity or avail­ab­il­ity is main­tain­able even if an internet service provider should fail due to a mal­func­tion.

Maximum avail­ab­il­ity is one of the key factors in every pro­duc­tion en­vir­on­ment. In the com­mu­nic­a­tion network, however, a single error can cause ap­plic­a­tions to fail. De­velopers do sleep better knowing that their ap­plic­a­tions are designed to withstand any con­ceiv­able error scenarios. The goal is to provide a highly available piece of in­fra­struc­ture with minimal downtime.

A floating IP can serve as a flexible load balancing address, helping to balance peak loads by dis­trib­ut­ing incoming network traffic to different network nodes. Network nodes are devices which connect two (or more) trans­mis­sion paths of a tele­com­mu­nic­a­tion network. As with a computer that dis­trib­utes workflows across multiple pro­cessors, load balancing also handles large amounts of sim­ul­tan­eous requests or more complex cal­cu­la­tions by splitting the load across multiple parallel systems.

If a primary load balancer or a central ap­plic­a­tion server in a cluster fails on one side, a floating IP can be im­me­di­ately assigned a redundant ap­plic­a­tion server or a secondary load balancer in a cor­res­pond­ingly con­figured system. The IP 'floats' to the active unit, which im­me­di­ately carries out the desired processes. An unplanned change between network services is referred to as 'fail­over'. This kind of pro­tec­tion is es­pe­cially re­com­men­ded for critical ap­plic­a­tions.

A planned change from a primary to a secondary system is referred to as a 'switchover'. The targeted trans­mis­sion of services is not triggered by errors, but is usually con­trolled by a system ad­min­is­trat­or. A classic reason for a switchover is, for example, routine main­ten­ance of the primary or secondary systems where a parallel instance tem­por­ar­ily takes over its function.

One of the main ad­vant­ages of floating IPs is their flex­ib­il­ity – the free and needs-oriented as­signab­il­ity. Floating IPs are therefore suitable for use in both failover and switchover en­vir­on­ments – for example, for per­form­ing upgrades of ap­plic­a­tions or entire sites with minimal downtime. While an upgrade is applied to one entity, another one takes on the traffic. Once the upgrade has been suc­cess­fully completed, the traffic is re­dir­ec­ted to the updated unit.

Another advantage: even if several or even many different entities are concealed behind a service being offered, the floating IP appears on the surface to users (who make use of the service) rather than the server’s IP that offers the re­spect­ive service.