IPv6 termination for external Application Load Balancers and external proxy Network Load Balancers

Google Cloud supports IPv6 clients with proxy load balancers. The load balancer accepts IPv6 connections from your users, and then proxies those connections to your backends.

You can configure IPv4 and IPv6 (dual-stack) IP addresses for the following proxy load balancers:

Feature Application Load Balancer Proxy Network Load Balancer
External External
Dual-stack frontends global
Dual-stack backends global1

1 Support IPv4 and IPv6 (dual-stack) instance groups and zonal NEG backends. Zonal NEGs support dual-stack on only GCE_VM_IP_PORT type endpoints.

The following example shows how load balancing works for global IPv6 clients:

Global IPv6 load balancing.
Global IPv6 load balancing (click to enlarge).

IPv6 termination lets you handle IPv6 requests from your users and proxy them over IPv4 or IPv6 to your backends. By using IPv6, you can do the following:

  • Use a single anycast IPv6 address for multi-region deployment. You only need one load balancer IPv6 address for application instances running across multiple regions. This means that your DNS server has a single AAAA record and that you don't need to load balance among multiple IPv6 addresses. Caching of AAAA records by clients is not an issue because there's only one address to cache. User requests to the IPv6 address are automatically load balanced to the closest healthy backend with available capacity.

  • Load balance HTTP, HTTPS, HTTP/2, TCP, and SSL/TLS IPv6 client traffic.

  • Overflow across regions with a single IPv6 load balancer address. If backends in one region are out of resources or unhealthy, the global load balancer automatically directs requests from users to the next closest region with available resources. When the closest region has available resources, global load balancing reverts back to serving by this region. Global load balancing requires that you use the Premium Tier of Network Service Tiers.

  • Run dual stack deployments. To serve both IPv6 and IPv4 clients, create two load balancer IP resources—one for IPv6 and the other for IPv4—and associate both with the same application instances. IPv4 clients connect to the IPv4 address while IPv6 clients connect to the IPv6 address. These clients are then automatically load balanced to the closest healthy IPv4 or IPv6 dual-stack backends with available capacity. To see which load balancers support dual-stack backends, see Table: Backend services and supported backend types. We provide IPv6 forwarding rules without charge, so you pay for only the IPv4 ones.

    Directing IPv4 and IPv6 traffic to the same backends.
    Directing IPv4 and IPv6 traffic to the same backends (click to enlarge).

IPv6 termination and proxy

Configuring IPv6 termination for your load balancers lets your backends appear as IPv6 applications to your IPv6 clients, as shown in the following figure:

IPv6 termination for load balancing.
IPv6 termination for load balancing (click to enlarge).

When a user connects to the load balancer through IPv6, the following happens:

  1. Your load balancer, with its IPv6 address and forwarding rule, waits for user connections.
  2. An IPv6 client connects to the load balancer using IPv6.
  3. The load balancer acts as a reverse proxy and terminates the IPv6 client connection. Based on the backend service IP address selection policy it places the request into an IPv4 or IPv6 connection to a backend.
  4. On the reverse path, the load balancer receives the response from the backend, and then places it into the IPv6 connection back to the original client.

IPv6 address allocation for load balancer forwarding rules

When you configure an external load balancer, you provide it with one or more global forwarding rules, each with an external, publicly routed IPv4 or IPv6 IP address (or both). You can use this IP address in the DNS records for your site.

When you create a forwarding rule, you can either use a static IP address reserved for your project or you can have the forwarding rule automatically acquire an ephemeral IP address when you create the rule. A static IP address is reserved to your project, and you can keep it until you deliberately release it. An ephemeral address belongs to the forwarding rule as long as the forwarding rule exists. If you delete the forwarding rule, the ephemeral address is released back into the Google Cloud pool.

If you need both an IPv4 and IPv6 address for your load balancer, you can create two forwarding rules, associating an IPv4 address with one and an IPv6 address with the other. You can then associate both rules with the same load balancer.

IPv6 address format

Google Cloud allocates a /64 IPv6 address range to IPv6 forwarding rules. The Google Cloud CLI lists IPv6 addresses with the least significant 64 bits set to 0, but the load balancer accepts traffic on the full range. Therefore, you might see other load balancer IPv6 addresses in the allocated range in X-Forwarded-For headers depending on which IPv6 server IP address the client connected to.

When formatting an IPv6 address, Google Cloud follows the recommendations in RFC 5952,section 4.

Client IP header with IPv6 termination for external Application Load Balancers

When the load balancer proxies the IPv6 connection from the client to an IPv4 connection to your backend, the original source IP address is replaced with the load balancer's IP address. However, backends often need to know the original source IP address for logging, for decision making, or for other purposes. Google Cloud provides an HTTP header that is propagated to the backends that includes the original IPv6 client IP address.

HTTP headers for IPv6 are similar to those for IPv4. The format for requests is as follows:


The last element shows the load balancer IP address. The second to last element shows the client IP address as seen by the load balancer. There might be other elements in the X-Forwarded-For header when the client or intervening proxies add other X-Forwarded-For headers before sending the request to the load balancer.

An example X-Forwarded-For header may look like this:

X-Forwarded-For: 2001:db8:abcd:1::1234, 2607:f8b0:4005:801::200e

2001:db8:abcd:1::1234 is the client's IPv6 address. 2607:f8b0:4005:801::200e is the IPv6 address of the external Application Load Balancer.


Forwarding rules for IPv6 termination are provided at no additional cost. You are not charged for ephemeral IPv6 addresses. Reserved IPv6 addresses are charged at existing rates regardless of whether they are in use. Otherwise, pricing for IPv6 load balancing is the same as pricing for IPv4 load balancing. For load balancing pricing details, see Network pricing.


  • IPv6 traffic is not supported with regional external Application Load Balancers, cross-region internal Application Load Balancers, regional internal Application Load Balancers, regional internal proxy Network Load Balancers, cross-region internal proxy Network Load Balancer, or regional external proxy Network Load Balancers.
  • Classic proxy Network Load Balancers and classic Application Load Balancers don't support dual-stack backends; the requests are proxied to IPv4 backends.

What's next

* IP address selection protocol * Choose a migration workflow