Ask Your Question

Revision history [back]

While Floating IPs are indeed public IP addresses, that's not the entire point of them.

The IP address that each instance receives is known as a Fixed IP address. In the majority of OpenStack installations, the Fixed IP range is composed of a private IP range. But it doesn't have to be. In fact, your Fixed IP range _could_ be a public IP subnet if you so wished.

The problem is that most people don't have a large enough public IP range to make it worthwhile to use as their Fixed IP range. Let's say, for example, you have access to a /24 subnet (254 usable IPs) and you make that your Fixed IP range. You've just limited yourself to 254 instances in your cloud.

If you made your Fixed IP range something like 10.0.0.0/16, you can accommodate ~65,500 instances. Now your Fixed IP range is no longer your bottleneck.

But public IP address limitations still aren't the point of Floating IPs, even though Floating IPs inadvertently resolve that issue.

Floating IPs are to Fixed IPs as volume storage is to instance storage: one can be moved from instance to instance while the other is stuck on a single instance. A Floating IP can be thought of as the antithesis of a Fixed IP.

Modern design of infrastructure inside the cloud says that you should make everything as disposable as possible. If you lose an instance, just use your automation tools to fire a new one up. It's a great goal to work towards, but reality doesn't always work that way. Just as attaching and detaching volumes serve their purpose, so does attaching and detaching IPs. That's really the point of a Floating IP.