Elastic IP Address Management and Design - Static IP Usage and Cost Optimization
Learn about Elastic IP allocation, EC2 association, cost implications of unused EIPs, and alternative approaches to consider.
Overview of Elastic IP
An Elastic IP (EIP) is a static public IPv4 address that you associate with an EC2 instance. While a regular EC2 public IP changes when the instance is stopped and started, an EIP retains the same IP address until you explicitly release it. EIPs are used when you need to point a DNS A record directly to an IP address or when you need to register an IP in a firewall allow list.
Cost Optimization and Alternatives
(As of August 2026, US East (Northern Virginia) / Asia Pacific (Tokyo)) Since February 2024, all public IPv4 addresses have been subject to hourly charges, and the unit price is 0.005 USD/hour for 1 address in both Northern Virginia and Tokyo. Not only unused EIPs but also EIPs already associated with EC2 are charged at the same unit price, so taking inventory of unnecessary EIPs is important. When you use an ALB, you reference the ALB's DNS name with a CNAME, so no EIP is needed, and the same is true when you use CloudFront. However, charges continue for the public IPv4 addresses that the load balancer itself uses in each Availability Zone, so expect that what you can reduce is only the addresses that had been assigned to the instances. Cases that require an EIP are limited to situations such as when an external system allows access by IP address or when you configure reverse DNS for an SMTP server.
EIP Design Patterns and IPv6 Migration
EIPs are put to use in failover configurations. By reassociating the EIP with another instance using Lambda when an EC2 instance fails, you can switch traffic immediately without waiting for the DNS TTL. If you want load balancing with fixed IPs, assign EIPs reserved in advance to each Availability Zone (subnet) when you create an internet-facing NLB. You can enable or disable Availability Zones at any time even after operation starts, and you can specify EIPs for subnets you add, but replacing the EIP of an already enabled zone requires deleting the subnet and adding it again, which involves creating a new network interface. To avoid the effort of switching, it is practical to reserve the required number of EIPs before creating the NLB. Migrating to IPv6 is also a way to reduce dependency on EIPs. On a path where you enable dual-stack in the VPC and expose EC2 instances directly over IPv6, those instances no longer need to have public IPv4 addresses. On the other hand, because you cannot assign EIPs to an ALB, when a fixed IP is a requirement, the options are either an NLB or the static Anycast IPs of Global Accelerator. With Global Accelerator, you can achieve fixed IPs that include failover during a regional outage, and you can consider it as an alternative to EIPs.
Public IPv4 Pricing and Reduction Strategies
(As of August 2026, US East (Northern Virginia) / Asia Pacific (Tokyo)) The unit price of public IPv4 addresses in both Northern Virginia and Tokyo is 0.005 USD/hour for both in-use (associated with a resource) and idle (allocated but unused) addresses, and holding 1 address for 1 month (calculated as 730 hours) comes to about 3.65 USD. Regarding the free tier, the official guidance is that accounts eligible for the EC2 free tier get 750 hours of in-use public IPv4 address usage free each month, but this covers only in-use addresses and does not include idle addresses. Because how the free tier is provided differs depending on when the account was created, check whether it applies to your own account on the free tier page of the billing console. Addresses that you own and bring in with BYOIP are not subject to this hourly charge. Use Public IP Insights in Amazon VPC IP Address Manager (IPAM) to visualize the usage of all public IPv4 addresses in your account and identify unnecessary EIPs and unused public IPs. If direct access to EC2 instances is not required, disable automatic public IP assignment and connect via Systems Manager Session Manager. Consolidating access through an ALB eliminates the need to assign EIPs to individual instances. It is recommended to track Elastic IP costs in Cost Explorer and take inventory of unnecessary EIPs monthly as a regular practice.
Allocation and Association Basics
You allocate EIPs per region and use them by associating them with EC2 instances, NAT gateways, or network interfaces. An EIP associated with an instance is retained even if you repeatedly stop and start the instance, so it suits uses where you want a fixed IP. If you move the association to another instance, you can replace only the server behind it without changing the IP. There is also a method in which you associate an EIP with a network interface (ENI) and move that ENI as a whole to another instance, and with it you can switch over more flexibly.
Bringing Your Own IP Addresses with BYOIP
If you already own an IP address range, you can bring it to AWS with BYOIP (Bring Your Own IP) and use it as EIPs. This is useful when, for example, IPs you have used for many years are registered in the allowlists of customers and partners and are difficult to change. AWS advertises the addresses you bring, and you can associate and use them in the same way as EIPs. As a result, you can take advantage of the AWS network infrastructure while avoiding the impact of IP changes that come with migration. Introducing it requires a procedure to prove ownership.
Quotas and Inventory Operations
EIPs have a default limit on the number of allocations per region, and you request a quota increase as needed. Because the limit is also a mechanism that prevents disorderly reservation, the basic practice is to release and clean up EIPs you are not using. If you record the purpose and owner with tags, it becomes easier to judge which EIPs are active during inventory. Because disassociated EIPs tend to remain unintentionally, if you systematize an operation that regularly identifies unassociated EIPs and releases unnecessary ones, you can prevent wasted costs and confusion in management.
Common Misconceptions and Cautions
The understanding that only unused EIPs are charged is wrong, and all public IPv4 addresses you hold, including EIPs associated with instances, are subject to charges. Also, associating an EIP alone does not ensure security. You need to design access control separately with security groups and network ACLs. A NAT gateway used for outbound traffic consumes one EIP, but this is something needed for availability. It is sound to determine at the design stage whether a fixed IP is really necessary and, if not, to shift to references based on DNS names.
Summary
Elastic IP provides static public IPv4 addresses, but since February 2024 all public IPv4 addresses have been subject to hourly charges, and the unit price is 0.005 USD/hour in both Northern Virginia and Tokyo (as of August 2026). Reduce instance-side EIPs by consolidating onto ALB or CloudFront, and handle situations that need fixed IPs with EIP assignment per NLB subnet or with the static Anycast IPs of Global Accelerator. Introducing IPv6 dual-stack is also a useful reduction measure that narrows the range of resources holding public IPv4 addresses.
References (Official AWS Resources)
The primary sources for this page are the official AWS website and documentation. Check the official pages below for the latest specifications and pricing.
If this page and the official documentation disagree, treat the official documentation as authoritative.