Set up a cross-region internal Application Load Balancer with hybrid connectivity

This page shows how to deploy a cross-region internal Application Load Balancer to load balance traffic to network endpoints that are on-premises or in other public clouds and that are reachable by using hybrid connectivity.

If you haven't already done so, review the Hybrid connectivity NEGs overview to understand the network requirements to set up hybrid load balancing.

Setup overview

The example sets up a cross-region internal Application Load Balancer for mixed zonal and hybrid connectivity NEG backends, as shown in the following figure:

Cross-region internal Application Load Balancer example for mixed zonal and hybrid connectivity NEG backends.
Cross-region internal Application Load Balancer example for mixed zonal and hybrid connectivity NEG backends (click to enlarge).

You must configure hybrid connectivity before setting up a hybrid load balancing deployment. Depending on your choice of hybrid connectivity product, use either Cloud VPN or Cloud Interconnect (Dedicated or Partner).

Set up an SSL certificate resource

Create a Certificate Manager SSL certificate resource as described in the following:

We recommend using a Google-managed certificate.

Permissions

To set up hybrid load balancing, you must have the following permissions:

  • On Google Cloud

    • Permissions to establish hybrid connectivity between Google Cloud and your on-premises environment or other cloud environments. For the list of permissions needed, see the relevant Network Connectivity product documentation.
    • Permissions to create a hybrid connectivity NEG and the load balancer. The Compute Load Balancer Admin role (roles/compute.loadBalancerAdmin) contains the permissions required to perform the tasks described in this guide.
  • On your on-premises environment or other non-Google Cloud cloud environment

    • Permissions to configure network endpoints that allow services on your on-premises environment or other cloud environments to be reachable from Google Cloud by using an IP:Port combination. For more information, contact your environment's network administrator.
    • Permissions to create firewall rules on your on-premises environment or other cloud environments to allow Google's health check probes to reach the endpoints.

Additionally, to complete the instructions on this page, you need to create a hybrid connectivity NEG, a load balancer, and zonal NEGs (and their endpoints) to serve as Google Cloud-based backends for the load balancer.

You should be either a project Owner or Editor, or you should have the following Compute Engine IAM roles.

Task Required role
Create networks, subnets, and load balancer components Compute Network Admin (roles/compute.networkAdmin)
Add and remove firewall rules Compute Security Admin (roles/compute.securityAdmin)
Create instances Compute Instance Admin (roles/compute.instanceAdmin)

Establish hybrid connectivity

Your Google Cloud and on-premises environment or other cloud environments must be connected through hybrid connectivity by using either Cloud Interconnect VLAN attachments or Cloud VPN tunnels with Cloud Router. We recommend that you use a high availability connection.

A Cloud Router enabled with global dynamic routing learns about the specific endpoint through Border Gateway Protocol (BGP) and programs it into your Google Cloud VPC network. Regional dynamic routing is not supported. Static routes are also not supported.

The VPC network that you use to configure either Cloud Interconnect or Cloud VPN is the same network that you use to configure the hybrid load balancing deployment. Ensure that your VPC network's subnet CIDR ranges do not conflict with your remote CIDR ranges. When IP addresses overlap, subnet routes are prioritized over remote connectivity.

For instructions, see the following documentation:

Set up your environment that is outside Google Cloud

Perform the following steps to set up your on-premises environment or other cloud environment for hybrid load balancing:

  • Configure network endpoints to expose on-premises services to Google Cloud (IP:Port).
  • Configure firewall rules on your on-premises environment or other cloud environment.
  • Configure Cloud Router to advertise certain required routes to your private environment.

Set up network endpoints

After you set up hybrid connectivity, you configure one or more network endpoints within your on-premises environment or other cloud environments that are reachable through Cloud Interconnect or Cloud VPN by using an IP:port combination. This IP:port combination is configured as one or more endpoints for the hybrid connectivity NEG that is created in Google Cloud later on in this process.

If there are multiple paths to the IP endpoint, routing follows the behavior described in the Cloud Router overview.

Set up firewall rules

The following firewall rules must be created on your on-premises environment or other cloud environment:

  • Create an ingress allow firewall rule in on-premises or other cloud environments to allow traffic from the region's proxy-only subnet to reach the endpoints.
  • Allowlisting Google's health check probe ranges isn't required for hybrid NEGs. However, if you're using a combination of hybrid and zonal NEGs in a single backend service, you need to allowlist the Google health check probe ranges for the zonal NEGs.

Configure Cloud Router to advertise the following custom IP ranges to your on-premises environment or other cloud environment:

  • The range of the region's proxy-only subnet.

Set up the Google Cloud environment

For the following steps, make sure you use the same VPC network (called NETWORK in this procedure) that was used to configure hybrid connectivity between the environments.

Additionally, make sure the regions used (called REGION_A and REGION_B in this procedure) are the same as those used to create the Cloud VPN tunnel or Cloud Interconnect VLAN attachments.

Optionally, you can configure DNS routing policies of type GEO to route client traffic to the load balancer VIP in the region closest to the client during regional outages.

Configure the backend subnets

Use this subnet to create the load balancer's zonal NEG backends:

Console

  1. In the Google Cloud console, go to the VPC networks page.

    Go to VPC networks

  2. Go to the network that was used to configure hybrid connectivity between the environments.

  3. In the Subnets section:

    • Set the Subnet creation mode to Custom.
    • In the New subnet section, enter the following information:
      • Provide a Name for the subnet.
      • Select a Region: REGION_A
      • Enter an IP address range.
    • Click Done.
  4. Click Create.

  5. To add more subnets in different regions, click Add subnet and repeat the previous steps for REGION_B

gcloud

  1. Create subnets in the network that was used to configure hybrid connectivity between the environments.

    gcloud compute networks subnets create SUBNET_A \
        --network=NETWORK \
        --range=LB_SUBNET_RANGE1 \
        --region=REGION_A
    
    gcloud compute networks subnets create SUBNET_B \
        --network=NETWORK \
        --range=LB_SUBNET_RANGE2 \
        --region=REGION_B
    

API

Make a POST request to the subnetworks.insert method. Replace PROJECT_ID with your project ID.

POST https://compute.googleapis.com/compute/v1/projects/PROJECT_ID/regions/REGION_A/subnetworks

{
 "name": "SUBNET_A",
 "network": "projects/PROJECT_ID/global/networks/NETWORK",
 "ipCidrRange": "LB_SUBNET_RANGE1",
 "region": "projects/PROJECT_ID/regions/REGION_A",
}

Make a POST request to the subnetworks.insert method. Replace PROJECT_ID with your project ID.

POST https://compute.googleapis.com/compute/v1/projects/PROJECT_ID/regions/REGION_B/subnetworks

{
 "name": "SUBNET_B",
 "network": "projects/PROJECT_ID/global/networks/NETWORK",
 "ipCidrRange": "LB_SUBNET_RANGE2",
 "region": "projects/PROJECT_ID/regions/REGION_B",
}

Replace the following:

  • SUBNET_A and SUBNET_B: the name of the subnets
  • LB_SUBNET_RANGE1 and LB_SUBNET_RANGE2: the IP address range for the subnets
  • REGION_A and REGION_B: the regions where you have configured the load balancer

Configure the proxy-only subnet

A proxy-only subnet provides a set of IP addresses that Google uses to run Envoy proxies on your behalf. The proxies terminate connections from the client and create new connections to the backends.

This proxy-only subnet is used by all Envoy-based regional load balancers in the same region of the VPC network. There can only be one active proxy-only subnet for a given purpose, per region, per network.

Console

If you're using the Google Cloud console, you can wait and create the proxy-only subnet later on the Load balancing page.

If you want to create the proxy-only subnet now, use the following steps:

  1. In the Google Cloud console, go to the VPC networks page.

    Go to VPC networks

  2. Click the name of the VPC network.
  3. On the Subnets tab, click Add subnet.
  4. Provide a Name for the proxy-only subnet.
  5. In the Region list, select REGION_A.
  6. In the Purpose list, select Cross-region Managed Proxy.
  7. In the IP address range field, enter 10.129.0.0/23.
  8. Click Add.

Create the proxy-only subnet in REGION_B

  1. Click Add subnet.
  2. Provide a Name for the proxy-only subnet.
  3. In the Region list, select REGION_B.
  4. In the Purpose list, select Cross-region Managed Proxy.
  5. In the IP address range field, enter 10.130.0.0/23.
  6. Click Add.

gcloud

Create the proxy-only subnets with the gcloud compute networks subnets create command.

    gcloud compute networks subnets create PROXY_SN_A \
        --purpose=GLOBAL_MANAGED_PROXY \
        --role=ACTIVE \
        --region=REGION_A \
        --network=NETWORK \
        --range=PROXY_ONLY_SUBNET_RANGE1
    
    gcloud compute networks subnets create PROXY_SN_B \
        --purpose=GLOBAL_MANAGED_PROXY \
        --role=ACTIVE \
        --region=REGION_B \
        --network=NETWORK \
        --range=PROXY_ONLY_SUBNET_RANGE2
    

Replace the following:

  • PROXY_SN_A and PROXY_SN_B: the name of the proxy-only subnets
  • PROXY_ONLY_SUBNET_RANGE1 and PROXY_ONLY_SUBNET_RANGE2: the IP address range for the proxy-only subnets
  • REGION_A and REGION_B: the regions where you have configured the load balancer

API

Create the proxy-only subnets with the subnetworks.insert method, replacing PROJECT_ID with your project ID.

    POST https://compute.googleapis.com/compute/v1/projects/PROJECT_ID/regions/REGION_A/subnetworks

    {
      "name": "PROXY_SN_A",
      "ipCidrRange": "PROXY_ONLY_SUBNET_RANGE1",
      "network": "projects/PROJECT_ID/global/networks/NETWORK",
      "region": "projects/PROJECT_ID/regions/REGION_A",
      "purpose": "GLOBAL_MANAGED_PROXY",
      "role": "ACTIVE"
    }
  
    POST https://compute.googleapis.com/compute/v1/projects/PROJECT_ID/regions/REGION_B/subnetworks

    {
      "name": " PROXY_SN_B",
      "ipCidrRange": "PROXY_ONLY_SUBNET_RANGE2",
      "network": "projects/PROJECT_ID/global/networks/NETWORK",
      "region": "projects/PROJECT_ID/regions/REGION_B",
      "purpose": "GLOBAL_MANAGED_PROXY",
      "role": "ACTIVE"
    }
  

Create firewall rules

In this example, you create the following firewall rules for the zonal NEG backends on Google Cloud:

  • fw-allow-health-check: An ingress rule, applicable to the instances being load balanced, that allows traffic from Google Cloud health checking systems (130.211.0.0/22 and 35.191.0.0/16). This example uses the target tag allow-health-check to identify the zonal NEGs to which it should apply.
  • fw-allow-ssh: An ingress rule that allows incoming SSH connectivity on TCP port 22 from any address. You can choose a more restrictive source IP range for this rule; for example, you can specify just the IP ranges of the systems from which you will initiate SSH sessions. This example uses the target tag allow-ssh to identify the virtual machine (VM) instances to which it should apply.
  • fw-allow-proxy-only-subnet: An ingress rule that allows connections from the proxy-only subnet to reach the zonal NEG backends.

Console

  1. In the Google Cloud console, go to the Firewall policies page.

    Go to Firewall policies

  2. Click Create firewall rule to create the rule to allow traffic from health check probes:

    1. Enter a Name of fw-allow-health-check.
    2. For Network, select NETWORK.
    3. For Targets, select Specified target tags.
    4. Populate the Target tags field with allow-health-check.
    5. Set Source filter to IPv4 ranges.
    6. Set Source IPv4 ranges to 130.211.0.0/22 and 35.191.0.0/16.
    7. For Protocols and ports, select Specified protocols and ports.
    8. Select TCP and then enter 80 for the port number.
    9. Click Create.
  3. Click Create firewall rule again to create the rule to allow incoming SSH connections:

    1. Name: fw-allow-ssh
    2. Network: NETWORK
    3. Priority: 1000
    4. Direction of traffic: ingress
    5. Action on match: allow
    6. Targets: Specified target tags
    7. Target tags: allow-ssh
    8. Source filter: IPv4 ranges
    9. Source IPv4 ranges: 0.0.0.0/0
    10. Protocols and ports: Choose Specified protocols and ports.
    11. Select TCP and then enter 22 for the port number.
    12. Click Create.
  4. Click Create firewall rule again to create the rule to allow incoming connections from the proxy-only subnet:

    1. Name: fw-allow-proxy-only-subnet
    2. Network: NETWORK
    3. Priority: 1000
    4. Direction of traffic: ingress
    5. Action on match: allow
    6. Targets: Specified target tags
    7. Target tags: allow-proxy-only-subnet
    8. Source filter: IPv4 ranges
    9. Source IPv4 ranges: PROXY_ONLY_SUBNET_RANGE1 and PROXY_ONLY_SUBNET_RANGE2
    10. Protocols and ports: Choose Specified protocols and ports
    11. Select TCP and then enter 80 for the port number.
    12. Click Create.

gcloud

  1. Create the fw-allow-health-check-and-proxy rule to allow the Google Cloud health checks to reach the backend instances on TCP port 80:

    gcloud compute firewall-rules create fw-allow-health-check \
        --network=NETWORK \
        --action=allow \
        --direction=ingress \
        --target-tags=allow-health-check \
        --source-ranges=130.211.0.0/22,35.191.0.0/16 \
        --rules=tcp:80
    
  2. Create the fw-allow-ssh firewall rule to allow SSH connectivity to VMs with the network tag allow-ssh. When you omit source-ranges, Google Cloud interprets the rule to mean any source.

    gcloud compute firewall-rules create fw-allow-ssh \
        --network=NETWORK \
        --action=allow \
        --direction=ingress \
        --target-tags=allow-ssh \
        --rules=tcp:22
    
  3. Create an ingress allow firewall rule for the proxy-only subnet to allow the load balancer to communicate with backend instances on TCP port 80:

    gcloud compute firewall-rules create fw-allow-proxy-only-subnet \
        --network=NETWORK \
        --action=allow \
        --direction=ingress \
        --target-tags=allow-proxy-only-subnet \
        --source-ranges=PROXY_ONLY_SUBNET_RANGE1,PROXY_ONLY_SUBNET_RANGE2 \
        --rules=tcp:80
    

Set up the zonal NEG

For Google Cloud-based backends, we recommend you configure multiple zonal NEGs in the same region where you configured hybrid connectivity.

For this example, set up a zonal NEG (with GCE_VM_IP_PORT type endpoints) in the REGION_A region. First create the VMs in the ZONE_A zone. Then create a zonal NEG in the ZONE_A zone, and then add the VMs' network endpoints to the NEG. To support high availability, set up a similar zonal NEG in the REGION_B region. If backends in one region happen to be down, traffic fails over to the other region.

Create VMs

Console

  1. In the Google Cloud console, go to the VM instances page.

    Go to VM instances

  2. Repeat steps 3 to 8 for each VM, using the following name and zone combinations.

    • Name: of vm-a1
      • Zone: ZONE_A in the region REGION_A
      • Subnet: SUBNET_A
    • Name: of vm-b1
      • Zone: ZONE_B in the region REGION_B
      • Subnet: SUBNET_B
  3. Click Create instance.

  4. Set the name as indicated in the preceding step.

  5. For the Region, choose as indicated in the earlier step.

  6. For the Zone, choose as indicated in the earlier step.

  7. In the Boot disk section, ensure that Debian GNU/Linux 12 (bookworm) is selected for the boot disk options. Click Choose to change the image if necessary.

  8. In the Advanced options section, expand Networking, and then do the following:

    • Add the following Network tags: allow-ssh,allow-health-check, and allow-proxy-only-subnet.
    • In the Network interfaces section, click Add a network interface make the following changes, and then click Done:
      • Network: NETWORK
      • Subnetwork: as indicated in the earlier step.
      • Primary internal IP: Ephemeral (automatic)
      • External IP: Ephemeral
    • Expand Management. In the Automation field, copy and paste the following script contents. The script contents are identical for all VMs:

      #! /bin/bash
      apt-get update
      apt-get install apache2 -y
      a2ensite default-ssl
      a2enmod ssl
      vm_hostname="$(curl -H "Metadata-Flavor:Google" \
      http://metadata.google.internal/computeMetadata/v1/instance/name)"
      echo "Page served from: $vm_hostname" | \
      tee /var/www/html/index.html
      systemctl restart apache2
      
  9. Click Create.

gcloud

Create the VMs by running the following command, using these combinations for the name of the VM and its zone. The script contents are identical for both VMs.

gcloud compute instances create VM_NAME \
    --zone=GCP_NEG_ZONE \
    --image-family=debian-12 \
    --image-project=debian-cloud \
    --tags=allow-ssh,allow-health-check,allow-proxy-only-subnet \
    --subnet=LB_SUBNET_NAME \
    --metadata=startup-script='#! /bin/bash
      apt-get update
      apt-get install apache2 -y
      a2ensite default-ssl
      a2enmod ssl
      vm_hostname="$(curl -H "Metadata-Flavor:Google" \
      http://metadata.google.internal/computeMetadata/v1/instance/name)"
      echo "Page served from: $vm_hostname" | \
      tee /var/www/html/index.html
      systemctl restart apache2'
  • VM_NAME of vm-a1
    • The zone GCP_NEG_ZONE as ZONE_A in the region REGION_A
    • The subnet LB_SUBNET_NAME as SUBNET_A
  • VM_NAME of vm-b1
    • ZoneGCP_NEG_ZONE as ZONE_B in the region REGION_B
    • Subnet LB_SUBNET_NAME as SUBNET_B

Create the zonal NEG

Console

To create a zonal network endpoint group:

  1. In the Google Cloud console, go to the Network Endpoint Groups page.

    Go to Network Endpoint Groups

  2. Repeat steps 3 to 8 for each zonal NEG, using the following name and zone combinations:

    • Name: neg-1
      • Zone: ZONE_A in the region REGION_A
      • Subnet: SUBNET_A
    • Name: neg-2
      • Zone: ZONE_B in the region REGION_B
      • Subnet: SUBNET_B
  3. Click Create network endpoint group.

  4. Set the name as indicated in the preceding step.

  5. Select the Network endpoint group type: Network endpoint group (Zonal).

  6. Select the Network: NETWORK

  7. Select the Subnetwork as indicated in earlier step.

  8. Select the Zone as indicated in earlier step.

  9. Enter the Default port: 80.

  10. Click Create.

Add endpoints to the zonal NEG:

  1. In the Google Cloud console, go to the Network Endpoint Groups page.

    Go to the Network endpoint groups

  2. Click the Name of the network endpoint group created in the previous step. You see the Network endpoint group details page.

  3. In the Network endpoints in this group section, click Add network endpoint. You see the Add network endpoint page.

  4. Select a VM instance to add its internal IP addresses as network endpoints. In the Network interface section, the name, zone, and subnet of the VM is displayed.

  5. Enter the IP address of the new network endpoint.

  6. Select the Port type.

    1. If you select Default, the endpoint uses the default port 80 for all endpoints in the network endpoint group. This is sufficient for our example because the Apache server is serving requests at port 80.
    2. If you select Custom, enter the Port number for the endpoint to use.
  7. To add more endpoints, click Add network endpoint and repeat the previous steps.

  8. After you add all the endpoints, click Create.

gcloud

  1. Create zonal NEGs (with GCE_VM_IP_PORT endpoints) using the name, zone, and subnet combinations. Use the gcloud compute network-endpoint-groups create command.

    gcloud compute network-endpoint-groups create GCP_NEG_NAME \
        --network-endpoint-type=GCE_VM_IP_PORT \
        --zone=GCP_NEG_ZONE \
        --network=NETWORK \
        --subnet=LB_SUBNET_NAME
    
    • Name: neg-1
      • Zone GCP_NEG_ZONE: ZONE_A in the region REGION_A
      • Subnet LB_SUBNET_NAME: SUBNET_A
    • Name: neg-2
      • Zone GCP_NEG_ZONE: ZONE_B in the region REGION_B
      • Subnet LB_SUBNET_NAME: SUBNET_B

    You can either specify a port using the --default-port option while creating the NEG, or specify a port number for each endpoint as shown in the next step.

  2. Add endpoints to neg1 and neg2.

    gcloud compute network-endpoint-groups update neg1 \
        --zone=ZONE_A \
        --add-endpoint='instance=vm-a1,port=80'
    
    gcloud compute network-endpoint-groups update neg2 \
        --zone=ZONE_B \
        --add-endpoint='instance=vm-b1,port=80'
    

Set up the hybrid connectivity NEG

When creating the NEG, use a zone that minimizes the geographic distance between Google Cloud and your on-premises or other cloud environment. For example, if you are hosting a service in an on-premises environment in Frankfurt, Germany, you can specify the europe-west3-a Google Cloud zone when you create the NEG.

And, if you're using Cloud Interconnect, the zone used to create the NEG is in the same region where the Cloud Interconnect attachment was configured.

Hybrid NEGs support only the distributed Envoy health checks.

Console

To create a hybrid connectivity network endpoint group:

  1. In the Google Cloud console, go to the Network Endpoint Groups page.

    Go to Network endpoint groups

  2. Click Create network endpoint group.

  3. Repeat steps 4 to 9 for each hybrid NEG, using the following name and zone combinations.

    • Name ON_PREM_NEG_NAME: hybrid-1
      • Zone: ON_PREM_NEG_ZONE1
      • Subnet: SUBNET_A
    • Name ON_PREM_NEG_NAME: hybrid-2
      • Zone: ON_PREM_NEG_ZONE2
      • Subnet: SUBNET_B
  4. Set the name as indicated in the previous step.

  5. Select the Network endpoint group type: Hybrid connectivity network endpoint group (Zonal).

  6. Select the Network: NETWORK

  7. For the Subnet, choose as indicated in the previous step.

  8. For the Zone, choose as indicated in the previous step.

  9. Enter the Default port.

  10. Click Create

Add endpoints to the hybrid connectivity NEG:

  1. In the Google Cloud console, go to the Network Endpoint Groups page.

    Go to Network endpoint groups

  2. Click the Name of the network endpoint group created in the previous step. You see the Network endpoint group detail page.

  3. In the Network endpoints in this group section, click Add network endpoint. You see the Add network endpoint page.

  4. Enter the IP address of the new network endpoint.

  5. Select the Port type.

    1. If you select Default, the endpoint uses the default port for all endpoints in the network endpoint group.
    2. If you select Custom, you can enter a different Port number for the endpoint to use.
  6. To add more endpoints, click Add network endpoint and repeat the previous steps.

  7. After you add all the non-Google Cloud endpoints, click Create.

gcloud

  1. Create a hybrid connectivity NEG that uses the following name combinations. Use the gcloud compute network-endpoint-groups create command.

    gcloud compute network-endpoint-groups create ON_PREM_NEG_NAME \
        --network-endpoint-type=NON_GCP_PRIVATE_IP_PORT \
        --zone=ON_PREM_NEG_ZONE \
        --network=NETWORK
    
    • Name ON_PREM_NEG_NAME: hybrid-1
      • Zone ON_PREM_NEG_ZONE: ON_PREM_NEG_ZONE1
    • Name ON_PREM_NEG_NAME: hybrid-2
      • Zone GCP_NEG_ZONE: ON_PREM_NEG_ZONE2
  2. Add the on-premises backend VM endpoint to ON_PREM_NEG_NAME:

    gcloud compute network-endpoint-groups update ON_PREM_NEG_NAME \
        --zone=ON_PREM_NEG_ZONE \
        --add-endpoint="ip=ON_PREM_IP_ADDRESS_1,port=PORT_1" \
        --add-endpoint="ip=ON_PREM_IP_ADDRESS_2,port=PORT_2"
    

You can use this command to add the network endpoints you previously configured on premises or in your cloud environment. Repeat --add-endpoint as many times as needed.

Configure the load balancer

Console

gcloud

  1. Define the HTTP health check with the gcloud compute health-checks create http command.

    gcloud compute health-checks create http gil7-basic-check \
       --use-serving-port \
       --global
    
  2. Create the backend service and enable logging with the gcloud compute backend-services create command.

    gcloud compute backend-services create BACKEND_SERVICE \
      --load-balancing-scheme=INTERNAL_MANAGED \
      --protocol=HTTP \
      --enable-logging \
      --logging-sample-rate=1.0 \
      --health-checks=gil7-basic-check \
      --global-health-checks \
      --global
    
  3. Add backends to the backend service with the gcloud compute backend-services add-backend command.

    gcloud compute backend-services add-backend BACKEND_SERVICE \
      --global \
      --balancing-mode=RATE \
      --max-rate-per-endpoint=MAX_REQUEST_RATE_PER_ENDPOINT \
      --network-endpoint-group=neg1 \
      --network-endpoint-group-zone=ZONE_A \
      --network-endpoint-group=neg2 \
      --network-endpoint-group-zone=ZONE_B
    

    For details about configuring the balancing mode, see the gcloud CLI documentation for the --max-rate-per-endpoint flag.

  4. Add the hybrid NEGs as a backend to the backend service.

    gcloud compute backend-services add-backend BACKEND_SERVICE \
      --global \
      --balancing-mode=RATE \
      --max-rate-per-endpoint=MAX_REQUEST_RATE_PER_ENDPOINT \
      --network-endpoint-group=hybrid1 \
      --network-endpoint-group-zone=ON_PREM_NEG_ZONE1 \
      --network-endpoint-group=hybrid2 \
      --network-endpoint-group-zone=ON_PREM_NEG_ZONE2 \
    

    For details about configuring the balancing mode, see the gcloud CLI documentation for the --max-rate-per-endpoint parameter.

  5. Create the URL map with the gcloud compute url-maps create command.

    gcloud compute url-maps create gil7-map \
      --default-service=BACKEND_SERVICE \
      --global
    
  6. Create the target proxy.

    For HTTP:

    Create the target proxy with the gcloud compute target-http-proxies create command.

    gcloud compute target-http-proxies create gil7-http-proxy \
      --url-map=gil7-map \
      --global
    

    For HTTPS:

    To create a Google-managed certificate, see the following documentation:

    After you create the Google-managed certificate, attach the certificate to the target proxy. Certificate maps are not supported by cross-region internal Application Load Balancers.

    To create a self-managed certificate, see the following documentation:

    Assign your file paths to variable names.

    export LB_CERT=PATH_TO_PEM_FORMATTED_FILE
    
    export LB_PRIVATE_KEY=PATH_TO_PEM_LB_PRIVATE_FILE
    

    Create an all region SSL certificate using the gcloud certificate-manager certificates create command.

    gcloud certificate-manager certificates create gilb-certificate \
      --private-key-file=$LB_CERT \
      --certificate-file=$LB_PRIVATE_KEY \
      --scope=all-regions
    

    Use the SSL certificate to create a target proxy with the gcloud compute target-https-proxies create command

    gcloud compute target-https-proxies create gil7-https-proxy \
      --url-map=gil7-map \
      --certificate-manager-certificates=gilb-certificate \
      --global
    
  7. Create two forwarding rules, one with a VIP IP_ADDRESS1 in the REGION_A region and another one with a VIP IP_ADDRESS2 in the REGION_B region. For the forwarding rule's IP address, use the LB_SUBNET_RANGE1 or LB_SUBNET_RANGE2 IP address range. If you try to use the proxy-only subnet, forwarding rule creation fails.

    For custom networks, you must reference the subnet in the forwarding rule. Note that this is the VM subnet, not the proxy subnet.

    For HTTP:

    Use the gcloud compute forwarding-rules create command with the correct flags.

    gcloud compute forwarding-rules create FWRULE_A \
      --load-balancing-scheme=INTERNAL_MANAGED \
      --network=NETWORK \
      --subnet=SUBNET_A \
      --subnet-region=REGION_A \
      --address=IP_ADDRESS1 \
      --ports=80 \
      --target-http-proxy=gil7-http-proxy \
      --global
    
    gcloud compute forwarding-rules create FWRULE_B \
      --load-balancing-scheme=INTERNAL_MANAGED \
      --network=NETWORK \
      --subnet=SUBNET_B \
      --subnet-region=REGION_B \
      --address=IP_ADDRESS2 \
      --ports=80 \
      --target-http-proxy=gil7-http-proxy \
      --global
    

    For HTTPS:

    Create the forwarding rule with the gcloud compute forwarding-rules create command with the correct flags.

    gcloud compute forwarding-rules create FWRULE_A \
      --load-balancing-scheme=INTERNAL_MANAGED \
      --network=NETWORK \
      --subnet=SUBNET_A \
      --subnet-region=REGION_A \
      --address=IP_ADDRESS1 \
      --ports=443 \
      --target-https-proxy=gil7-https-proxy \
      --global
    
    gcloud compute forwarding-rules create FWRULE_B \
      --load-balancing-scheme=INTERNAL_MANAGED \
      --network=NETWORK \
      --subnet=SUBNET_B \
      --subnet-region=REGION_B \
      --address=IP_ADDRESS2 \
      --ports=443 \
      --target-https-proxy=gil7-https-proxy \
      --global
    

Connect your domain to your load balancer

After you create the load balancer, note the IP address that is associated with the load balancer—for example, IP_ADDRESS1 and IP_ADDRESS2. To point your domain to your load balancer, create an A record by using Cloud DNS or domain registration service. If you added multiple domains to your SSL certificate, you must add an A record for each one, all pointing to the load balancer's IP address.

Test the load balancer

Create a VM instance to test connectivity

  1. Create a client VM:

    gcloud compute instances create l7-ilb-client-a \
        --image-family=debian-12 \
        --image-project=debian-cloud \
        --network=NETWORK \
        --subnet=SUBNET_A \
        --zone=ZONE_A \
        --tags=allow-ssh
    
    gcloud compute instances create l7-ilb-client-b \
        --image-family=debian-12 \
        --image-project=debian-cloud \
        --network=NETWORK \
        --subnet=SUBNET_B \
        --zone=ZONE_B \
        --tags=allow-ssh
    
  2. Use SSH to connect to each client instance.

    gcloud compute ssh l7-ilb-client-a \
       --zone=ZONE_A
    
    gcloud compute ssh l7-ilb-client-b \
       --zone=ZONE_B
    
  3. Verify that the IP address is serving its hostname.

    • Verify that the client VM can reach both IP addresses. The command should succeed and return the name of the backend VM which served the request:

      curl IP_ADDRESS1
      
      curl IP_ADDRESS2
      

      For HTTPS testing, replace curl with:

      curl -k -s 'https://test.example.com:443' --connect-to test.example.com:443:IP_ADDRESS1:443
      
      curl -k -s 'https://test.example.com:443' --connect-to test.example.com:443:IP_ADDRESS2:443
      

      The -k flag causes curl to skip certificate validation.

    • Optional: Use the configured DNS record to resolve the IP address closest to the client VM. For example, DNS_ENTRY can be service.example.com.

      curl DNS_ENTRY
      

Run 100 requests

Run 100 curl requests and confirm from the responses that they are load balanced.

For HTTP:

  • Verify that the client VM can reach both IP addresses. The command should succeed and return the name of the backend VM which served the request:

    {
      RESULTS=
      for i in {1..100}
      do
        RESULTS="$RESULTS:$(curl --silent IP_ADDRESS1)"
      done
      echo "***"
      echo "*** Results of load-balancing to IP_ADDRESS1: "
      echo "***"
      echo "$RESULTS" | tr ':' '\n' | grep -Ev "^$" | sort | uniq -c
      echo
    }
    
    {
      RESULTS=
      for i in {1..100}
      do
        RESULTS="$RESULTS:$(curl --silent IP_ADDRESS2)"
      done
      echo "***"
      echo "*** Results of load-balancing to IP_ADDRESS2: "
      echo "***"
      echo "$RESULTS" | tr ':' '\n' | grep -Ev "^$" | sort | uniq -c
      echo
    }
    

For HTTPS:

  • Verify that the client VM can reach both IP addresses. The command should succeed and return the name of the backend VM which served the request:

    {
      RESULTS=
      for i in {1..100}
      do
        RESULTS="$RESULTS:$(curl -k -s 'https://test.example.com:443' --connect-to test.example.com:443:IP_ADDRESS1:443)"
      done
      echo "***"
      echo "*** Results of load-balancing to IP_ADDRESS1: "
      echo "***"
      echo "$RESULTS" | tr ':' '\n' | grep -Ev "^$" | sort | uniq -c
      echo
    }
    
    {
      RESULTS=
      for i in {1..100}
      do
        RESULTS="$RESULTS:$(curl -k -s 'https://test.example.com:443' --connect-to test.example.com:443:IP_ADDRESS2:443)"
      done
      echo "***"
      echo "*** Results of load-balancing to IP_ADDRESS2: "
      echo "***"
      echo "$RESULTS" | tr ':' '\n' | grep -Ev "^$" | sort | uniq -c
      echo
    }
    

Test failover

  1. Verify failover to backends in the REGION_A region when backends in the REGION_B are unhealthy or unreachable. We simulate this by removing all the backends from REGION_B:

    gcloud compute backend-services remove-backend BACKEND_SERVICE \
       --balancing-mode=RATE \
       --network-endpoint-group=neg2 \
       --network-endpoint-group-zone=ZONE_B
    
  2. Use SSH to connect to the client VM in REGION_B.

    gcloud compute ssh l7-ilb-client-b \
       --zone=ZONE_B
    
  3. Send requests to the load balanced IP address in the REGION_B region. The output is similar to the following:

    {
    RESULTS=
    for i in {1..100}
    do
      RESULTS="$RESULTS:$(curl -k -s 'https://test.example.com:443' --connect-to test.example.com:443:IP_ADDRESS2:443)"
    done
    echo "***"
    echo "*** Results of load-balancing to IP_ADDRESS2: "
    echo "***"
    echo "$RESULTS" | tr ':' '\n' | grep -Ev "^$" | sort | uniq -c
    echo
    }
    

Additional configuration options

This section expands on the configuration example to provide alternative and additional configuration options. All of the tasks are optional. You can perform them in any order.

Configure DNS routing policies

If your clients are in multiple regions, you might want to make your cross-region internal Application Load Balancer accessible by using VIPs in these regions. This multi-region setup minimizes latency and network transit costs. In addition, it lets you set up a DNS-based, global, load balancing solution that provides resilience against regional outages. For more information, see Manage DNS routing policies and health checks.

gcloud

To create a DNS entry with a 30 second TTL, use the gcloud dns record-sets create command.

gcloud dns record-sets create DNS_ENTRY --ttl="30" \
  --type="A" --zone="service-zone" \
  --routing-policy-type="GEO" \
  --routing-policy-data="REGION_A=gil7-forwarding-rule-a@global;REGION_B=gil7-forwarding-rule-b@global" \
  --enable-health-checking

Replace the following:

  • DNS_ENTRY: DNS or domain name of the record-set

    For example, service.example.com

  • REGION_A and REGION_B: the regions where you have configured the load balancer

API

Create the DNS record by making a POST request to the ResourceRecordSets.create method. Replace PROJECT_ID with your project ID.

POST https://www.googleapis.com/dns/v1/projects/PROJECT_ID/managedZones/SERVICE_ZONE/rrsets
{
  "name": "DNS_ENTRY",
  "type": "A",
  "ttl": 30,
  "routingPolicy": {
    "geo": {
      "items": [
        {
          "location": "REGION_A",
          "healthCheckedTargets": {
            "internalLoadBalancers": [
              {
                "loadBalancerType": "globalL7ilb",
                "ipAddress": "IP_ADDRESS",
                "port": "80",
                "ipProtocol": "tcp",
                "networkUrl": "https://www.googleapis.com/compute/v1/projects/PROJECT_ID/global/networks/lb-network",
                "project": "PROJECT_ID"
              }
            ]
          }
        },
        {
          "location": "REGION_B",
          "healthCheckedTargets": {
            "internalLoadBalancers": [
              {
                "loadBalancerType": "globalL7ilb",
                "ipAddress": "IP_ADDRESS_B",
                "port": "80",
                "ipProtocol": "tcp",
                "networkUrl": "https://www.googleapis.com/compute/v1/projects/PROJECT_ID/global/networks/lb-network",
                "project": "PROJECT_ID"
              }
            ]
          }
        }
      ]
    }
  }
}

Update client HTTP keepalive timeout

The load balancer created in the previous steps has been configured with a default value for the client HTTP keepalive timeout.

To update the client HTTP keepalive timeout, use the following instructions.

Console

  1. In the Google Cloud console, go to the Load balancing page.

    Go to Load balancing.

  2. Click the name of the load balancer that you want to modify.
  3. Click Edit.
  4. Click Frontend configuration.
  5. Expand Advanced features. For HTTP keepalive timeout, enter a timeout value.
  6. Click Update.
  7. To review your changes, click Review and finalize, and then click Update.

gcloud

For an HTTP load balancer, update the target HTTP proxy by using the gcloud compute target-http-proxies update command:

    gcloud compute target-http-proxies update TARGET_HTTP_PROXY_NAME \
        --http-keep-alive-timeout-sec=HTTP_KEEP_ALIVE_TIMEOUT_SEC \
        --global
    

For an HTTPS load balancer, update the target HTTPS proxy by using the gcloud compute target-https-proxies update command:

    gcloud compute target-https-proxies update TARGET_HTTPS_PROXY_NAME \
        --http-keep-alive-timeout-sec=HTTP_KEEP_ALIVE_TIMEOUT_SEC \
        --global
    

Replace the following:

  • TARGET_HTTP_PROXY_NAME: the name of the target HTTP proxy.
  • TARGET_HTTPS_PROXY_NAME: the name of the target HTTPS proxy.
  • HTTP_KEEP_ALIVE_TIMEOUT_SEC: the HTTP keepalive timeout value from 5 to 600 seconds.

Enable outlier detection

You can enable outlier detection on global backend services to identify unhealthy serverless NEGs and reduce the number the requests sent to the unhealthy serverless NEGs.

Outlier detection is enabled on the backend service by using one of the following methods:

  • The consecutiveErrors method (outlierDetection.consecutiveErrors), in which a 5xx series HTTP status code qualifies as an error.
  • The consecutiveGatewayFailure method (outlierDetection.consecutiveGatewayFailure), in which only the 502, 503, and 504 HTTP status codes qualify as an error.

Use the following steps to enable outlier detection for an existing backend service. Note that even after enabling outlier detection, some requests can be sent to the unhealthy service and return a 5xx status code to the clients. To further reduce the error rate, you can configure more aggressive values for the outlier detection parameters. For more information, see the outlierDetection field.

Console

  1. In the Google Cloud console, go to the Load balancing page.

    Go to Load balancing

  2. Click the name of the load balancer whose backend service you want to edit.

  3. On the Load balancer details page, click Edit.

  4. On the Edit cross-region internal Application Load Balancer page, click Backend configuration.

  5. On the Backend configuration page, click Edit for the backend service that you want to modify.

  6. Scroll down and expand the Advanced configurations section.

  7. In the Outlier detection section, select the Enable checkbox.

  8. Click Edit to configure outlier detection.

    Verify that the following options are configured with these values:

    Property Value
    Consecutive errors 5
    Interval 1000
    Base ejection time 30000
    Max ejection percent 50
    Enforcing consecutive errors 100

    In this example, the outlier detection analysis runs every one second. If the number of consecutive HTTP 5xx status codes received by an Envoy proxy is five or more, the backend endpoint is ejected from the load-balancing pool of that Envoy proxy for 30 seconds. When the enforcing percentage is set to 100%, the backend service enforces the ejection of unhealthy endpoints from the load-balancing pools of those specific Envoy proxies every time the outlier detection analysis runs. If the ejection conditions are met, up to 50% of the backend endpoints from the load-balancing pool can be ejected.

  9. Click Save.

  10. To update the backend service, click Update.

  11. To update the load balancer, on the Edit cross-region internal Application Load Balancer page, click Update.

gcloud

  1. Export the backend service into a YAML file.

    gcloud compute backend-services export BACKEND_SERVICE_NAME \
      --destination=BACKEND_SERVICE_NAME.yaml --global
    

    Replace BACKEND_SERVICE_NAME with the name of the backend service.

  2. Edit the YAML configuration of the backend service to add the fields for outlier detection as highlighted in the following YAML configuration, in the outlierDetection section:

    In this example, the outlier detection analysis runs every one second. If the number of consecutive HTTP 5xx status codes received by an Envoy proxy is five or more, the backend endpoint is ejected from the load-balancing pool of that Envoy proxy for 30 seconds. When the enforcing percentage is set to 100%, the backend service enforces the ejection of unhealthy endpoints from the load-balancing pools of those specific Envoy proxies every time the outlier detection analysis runs. If the ejection conditions are met, up to 50% of the backend endpoints from the load-balancing pool can be ejected.

    name: BACKEND_SERVICE_NAME
    backends:
    - balancingMode: UTILIZATION
      capacityScaler: 1.0
      group: https://www.googleapis.com/compute/v1/projects/PROJECT_ID/regions/REGION_A/networkEndpointGroups/SERVERLESS_NEG_NAME
    - balancingMode: UTILIZATION
      capacityScaler: 1.0
      group: https://www.googleapis.com/compute/v1/projects/PROJECT_ID/regions/REGION_B/networkEndpointGroups/SERVERLESS_NEG_NAME_2
    outlierDetection:
      baseEjectionTime:
        nanos: 0
        seconds: 30
      consecutiveErrors: 5
      enforcingConsecutiveErrors: 100
      interval:
        nanos: 0
        seconds: 1
      maxEjectionPercent: 50
    port: 80
    selfLink: https://www.googleapis.com/compute/v1/projects/PROJECT_ID/global/backendServices/BACKEND_SERVICE_NAME
    sessionAffinity: NONE
    timeoutSec: 30
    ...
    

    Replace the following:

    • BACKEND_SERVICE_NAME: the name of the backend service
    • PROJECT_ID: the ID of your project
    • REGION_A and REGION_B: the regions where the load balancer has been configured.
    • SERVERLESS_NEG_NAME: the name of the first serverless NEG
    • SERVERLESS_NEG_NAME_2: the name of the second serverless NEG
  3. Update the backend service by importing the latest configuration.

    gcloud compute backend-services import BACKEND_SERVICE_NAME \
      --source=BACKEND_SERVICE_NAME.yaml --global
    

    Outlier detection is now enabled on the backend service.

What's next