Amazon Web Services EC2 – Part 2: Elastic IP Addresses
Elastic Compute Cloud (EC2)
Elastic IPs are a very neat feature of EC2. An Elastic IP is a configurable IP address assigned to your account. Unlike the rest of the features in AWS, you pay for any elastic IPs that you have allocated but not used. Amazon doesn’t want customers hoarding, and thus wasting, this resource. If you set aside the IP and use it, it’s free. If you set it aside and just keep it hanging around unused, you’ll pay a small monthly fee.
These IP addresses are connected to your account so you can assign them to any instance you might be running. A great example would be if you were running two instances in different availability zones. If something were to happen and you lost a zone (say a cable is cut) and access to that zone is lost, you can use the API (or a GUI) to reassign the IP to the other instance. You have total control over where the address points.
Elastic IPs are limited to 5 per customer by default. You can easily request a higher number if you have a good business need for it. If you allocate an IP and do not use it, you are billed $0.01 for each hour that it is not used.
In addition to the optional Elastic IPs, every instance is assigned an internal IP and a public IP. When connecting internally, it is better to use the internal IP. This is for both latency and financial reasons. The internal IP will be within the Amazon network so will be the highest bandwidth. It also incurs no transfer costs. That means you can FTP between instances and back and forth with S3 without incurring transfer fees.