Network load balancing on Amazon EKS - Amazon EKS
Services or capabilities described in Amazon Web Services documentation might vary by Region. To see the differences applicable to the China Regions, see Getting Started with Amazon Web Services in China.

Network load balancing on Amazon EKS

Network traffic is load balanced at L4 of the OSI model. To load balance application traffic at L7, you deploy a Kubernetes ingress, which provisions an Amazon Application Load Balancer. For more information, see Application load balancing on Amazon EKS. To learn more about the differences between the two types of load balancing, see Elastic Load Balancing features on the Amazon website.

When you create a Kubernetes Service of type LoadBalancer, the Amazon cloud provider load balancer controller creates Amazon Classic Load Balancers by default, but can also create Amazon Network Load Balancers. This controller is only receiving critical bug fixes in the future. For more information about using the Amazon cloud provider load balancer , see Amazon cloud provider load balancer controller in the Kubernetes documentation. Its use is not covered in this topic.

For new services deployed to 1.19 or later clusters, we recommend that you use version 2.4.1 or later of the Installing the Amazon Load Balancer Controller add-on instead of the Amazon cloud provider load balancer controller. If your cluster is earlier than 1.19, then we recommend that you use version 2.3.1 of the controller. The Amazon Load Balancer Controller creates Amazon Network Load Balancers, but doesn't create Amazon Classic Load Balancers. The remainder of this topic is about using the Amazon Load Balancer Controller.

An Amazon Network Load Balancer can load balance network traffic to pods deployed to Amazon EC2 IP and instance targets or to Amazon Fargate IP targets. For more information, see Amazon Load Balancer Controller on GitHub.

Prerequisites

Before you can load balance network traffic using the Amazon Load Balancer Controller, you must meet the following requirements.

  • Have an existing cluster. If you don't have an existing cluster, see Getting started with Amazon EKS. If you need to update the version of an existing cluster, see Updating a cluster.

  • Have the Amazon Load Balancer Controller deployed on your cluster. For more information, see Installing the Amazon Load Balancer Controller add-on. We recommend version 2.4.1 or later for 1.19 or later clusters. If your cluster is earlier than 1.19, then we recommend using version 2.3.1.

  • At least one subnet. If multiple tagged subnets are found in an Availability Zone, the controller chooses the first subnet whose subnet ID comes first lexicographically. The subnet must have at least eight available IP addresses.

  • If you're using the Amazon Load Balancer Controller version v2.1.1 or earlier, subnets must be tagged as follows. If using version 2.1.2 or later, this tag is optional. You might want to tag a subnet if you have multiple clusters running in the same VPC, or multiple Amazon services sharing subnets in a VPC, and want more control over where load balancers are provisioned for each cluster. If you explicitly specify subnet IDs as an annotation on a service object, then Kubernetes and the Amazon Load Balancer Controller use those subnets directly to create the load balancer. Subnet tagging isn't required if you choose to use this method for provisioning load balancers and you can skip the following private and public subnet tagging requirements. Replace cluster-name with your cluster name.

    • Keykubernetes.io/cluster/cluster-name

    • Valueshared or owned

  • Your public and private subnets must meet the following requirements, unless you explicitly specify subnet IDs as an annotation on a service or ingress object. If you provision load balancers by explicitly specifying subnet IDs as an annotation on a service or ingress object, then Kubernetes and the Amazon Load Balancer Controller use those subnets directly to create the load balancer and the following tags aren't required.

    • Private subnets – Must be tagged in the following format. This is so that Kubernetes and the Amazon Load Balancer Controller know that the subnets can be used for internal load balancers. If you use eksctl or an Amazon EKS Amazon Amazon CloudFormation template to create your VPC after March 26, 2020, then the subnets are tagged appropriately when they're created. For more information about the Amazon EKS Amazon Amazon CloudFormation VPC templates, see Creating a VPC for your Amazon EKS cluster.

      • Keykubernetes.io/role/internal-elb

      • Value1

    • Public subnets – Must be tagged in the following format. This is so that Kubernetes knows to use only those subnets for external load balancers instead of choosing a public subnet in each Availability Zone (based on the lexicographical order of the subnet IDs). If you use eksctl or an Amazon EKS Amazon CloudFormation template to create your VPC after March 26, 2020, then the subnets are tagged appropriately when they're created. For more information about the Amazon EKS Amazon CloudFormation VPC templates, see Creating a VPC for your Amazon EKS cluster.

      • Keykubernetes.io/role/elb

      • Value1

    If the subnet role tags aren't explicitly added, the Kubernetes service controller examines the route table of your cluster VPC subnets to determine if the subnet is private or public. We recommend that you don't rely on this behavior, and instead explicitly add the private or public role tags. The Amazon Load Balancer Controller doesn't examine route tables, and requires the private and public tags to be present for successful auto discovery.

Considerations

  • The configuration of your load balancer is controlled by annotations that are added to the manifest for your service. Service annotations are different when using the Amazon Load Balancer Controller than they are when using the Amazon cloud provider load balancer controller. Make sure to review the annotations for the Amazon Load Balancer Controller before deploying services.

  • When using the Amazon EKS VPC CNI plugin, the Amazon Load Balancer Controller can load balance to Amazon EC2 IP or instance targets and Fargate IP targets. When using Alternate compatible CNI plugins, the controller can only load balance to instance targets. For more information about Network Load Balancer target types, see Target type in the User Guide for Network Load Balancers

  • If you want to add tags to the load balancer when or after it's created, add the following annotation in your service specification. For more information, see Amazon Resource Tags in the Amazon Load Balancer Controller documentation.

    service.beta.kubernetes.io/aws-load-balancer-additional-resource-tags
  • You can assign Elastic IP addresses to the Network Load Balancer by adding the following annotation. Replace the example-values with the Allocation IDs of your Elastic IP addresses. The number of Allocation IDs must match the number of subnets that are used for the load balancer. For more information, see the Amazon Load Balancer Controller documentation.

    service.beta.kubernetes.io/aws-load-balancer-eip-allocations: eipalloc-xxxxxxxxxxxxxxxxx,eipalloc-yyyyyyyyyyyyyyyyy
  • Amazon EKS adds one inbound rule to the node's security group for client traffic and one rule for each load balancer subnet in the VPC for health checks for each Network Load Balancer that you create. deployment of a service of type LoadBalancer can fail if Amazon EKS attempts to create rules that exceed the quota for the maximum number of rules allowed for a security group. For more information, see Security groups in Amazon VPC quotas in the Amazon VPC User Guide. Consider the following options to minimize the chances of exceeding the maximum number of rules for a security group:

    • Request an increase in your rules per security group quota. For more information, see Requesting a quota increase in the Service Quotas User Guide.

    • Use IP targets, rather than instance targets. With IP targets, you can share rules for the same target ports. You can manually specify load balancer subnets with an annotation. For more information, see Annotations on GitHub.

    • Use an ingress, instead of a service of type LoadBalancer, to send traffic to your service. The Amazon Application Load Balancer requires fewer rules than Network Load Balancers. You can share an ALB across multiple ingresses. For more information, see Application load balancing on Amazon EKS. You can't share a Network Load Balancer across multiple services.

    • Deploy your clusters to multiple accounts.

  • If your pods run on Windows in an Amazon EKS cluster, a single service with a load balancer can support up to 64 backend pods. Each pod has its own unique IP address. This is a limitation of the Windows OS on the Amazon EC2 nodes.

  • We recommend only creating new Network Load Balancers with the Amazon Load Balancer Controller. Attempting to replace existing Network Load Balancers created with the Amazon cloud provider load balancer controller can result in multiple Network Load Balancers that might cause application downtime.

Create a network load balancer

You can create a network load balancer with IP or instance targets.

IP targets

You can use IP targets with pods deployed to Amazon EC2 nodes or Fargate. Your Kubernetes service must be created as type LoadBalancer. For more information, see Type LoadBalancer in the Kubernetes documentation.

To create a load balancer that uses IP targets, add the following annotations to a service manifest and deploy your service. The external value for aws-load-balancer-type is what causes the Amazon Load Balancer Controller, rather than the Amazon cloud provider load balancer controller, to create the Network Load Balancer. You can view a sample service manifest with the annotations.

service.beta.kubernetes.io/aws-load-balancer-type: "external" service.beta.kubernetes.io/aws-load-balancer-nlb-target-type: "ip"
Note

If you're load balancing to IPv6 pods, add the following annotation. You can only load balance over IPv6 to IP targets, not instance targets. Without this annotation, load balancing is over IPv4.

service.beta.kubernetes.io/aws-load-balancer-ip-address-type: dualstack

Network Load Balancers are created with the internal aws-load-balancer-scheme, by default. For internal Network Load Balancers, your Amazon EKS cluster must be configured to use at least one private subnet in your VPC. Kubernetes examines the route table for your subnets to identify whether they are public or private. Public subnets have a route directly to the internet using an internet gateway, but private subnets do not.

If you want to create a Network Load Balancer in a public subnet to load balance to Amazon EC2 nodes (Fargate can only be private), specify internet-facing with the following annotation:

service.beta.kubernetes.io/aws-load-balancer-scheme: "internet-facing"
Note

The service.beta.kubernetes.io/aws-load-balancer-type: "nlb-ip" annotation is still supported for backwards compatibility. However, we recommend using the previous annotations for new load balancers instead of service.beta.kubernetes.io/aws-load-balancer-type: "nlb-ip".

Important

Do not edit the annotations after creating your service. If you need to modify it, delete the service object and create it again with the desired value for this annotation.

Instance targets

The Amazon cloud provider load balancer controller creates Network Load Balancers with instance targets only. Version 2.2.0 and later of the Amazon Load Balancer Controller also creates Network Load Balancers with instance targets. We recommend using it, rather than the Amazon cloud provider load balancer controller, to create new Network Load Balancers. You can use Network Load Balancer instance targets with pods deployed to Amazon EC2 nodes, but not to Fargate. To load balance network traffic across pods deployed to Fargate, you must use IP targets.

To deploy a Network Load Balancer to a private subnet, your service specification must have the following annotations. You can view a sample service manifest with the annotations. The external value for aws-load-balancer-type is what causes the Amazon Load Balancer Controller, rather than the Amazon cloud provider load balancer controller, to create the Network Load Balancer.

service.beta.kubernetes.io/aws-load-balancer-type: "external" service.beta.kubernetes.io/aws-load-balancer-nlb-target-type: "instance"

Network Load Balancers are created with the internal aws-load-balancer-scheme, by default. For internal Network Load Balancers, your Amazon EKS cluster must be configured to use at least one private subnet in your VPC. Kubernetes examines the route table for your subnets to identify whether they are public or private. Public subnets have a route directly to the internet using an internet gateway, but private subnets do not.

If you want to create an Network Load Balancer in a public subnet to load balance to Amazon EC2 nodes, specify internet-facing with the following annotation:

service.beta.kubernetes.io/aws-load-balancer-scheme: "internet-facing"
Important

Do not edit the annotations after creating your service. If you need to modify it, delete the service object and create it again with the desired value for this annotation.

(Optional) Deploy a sample application

Prerequisites

To deploy a sample application

  1. If you're deploying to Fargate, make sure you have an available private subnet in your VPC and create a Fargate profile. If you're not deploying to Fargate, skip this step. You can create the profile by running the following command or in the Amazon Web Services Management Console using the same values for name and namespace that are in the command. Replace the example values with your own.

    eksctl create fargateprofile \ --cluster my-cluster \ --region region-code \ --name nlb-sample-app \ --namespace nlb-sample-app
  2. Deploy a sample application.

    1. Create a namespace for the application.

      kubectl create namespace nlb-sample-app
    2. Save the following contents to a file named sample-deployment.yaml file on your computer.

      apiVersion: apps/v1 kind: Deployment metadata: name: nlb-sample-app namespace: nlb-sample-app spec: replicas: 3 selector: matchLabels: app: nginx template: metadata: labels: app: nginx spec: containers: - name: nginx image: public.ecr.aws/nginx/nginx:1.21 ports: - name: tcp containerPort: 80
    3. Apply the manifest to the cluster.

      kubectl apply -f sample-deployment.yaml
  3. Create a service with an internet-facing Network Load Balancer that load balances to IP targets.

    1. Save the following contents to a file named sample-service.yaml file on your computer. If you're deploying to Fargate nodes, remove the service.beta.kubernetes.io/aws-load-balancer-scheme: internet-facing line.

      apiVersion: v1 kind: Service metadata: name: nlb-sample-service namespace: nlb-sample-app annotations: service.beta.kubernetes.io/aws-load-balancer-type: external service.beta.kubernetes.io/aws-load-balancer-nlb-target-type: ip service.beta.kubernetes.io/aws-load-balancer-scheme: internet-facing spec: ports: - port: 80 targetPort: 80 protocol: TCP type: LoadBalancer selector: app: nginx
    2. Apply the manifest to the cluster.

      kubectl apply -f sample-service.yaml
  4. Verify that the service was deployed.

    kubectl get svc nlb-sample-service -n nlb-sample-app

    Example output:

    NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE sample-service LoadBalancer 10.100.240.137 k8s-nlbsampl-nlbsampl-xxxxxxxxxx-xxxxxxxxxxxxxxxx.elb.region-code.amazonaws.com.cn 80:32400/TCP 16h
    Note

    The values for 10.100.240.137 and xxxxxxxxxx-xxxxxxxxxxxxxxxx will be different than the example output (they will be unique to your load balancer) and cn-north-1 may be different for you, depending on which Amazon Web Services Region that your cluster is in.

  5. Open the Amazon EC2 Amazon Web Services Management Console. Select Target Groups (under Load Balancing) in the left navigation pane. In the Name column, select the target group's name where the value in the Load balancer column matches a portion of the name in the EXTERNAL-IP column of the output in the previous step. For example, you'd select the target group named k8s-default-samplese-xxxxxxxxxx if your output were the same as the output above. The Target type is IP because that was specified in the sample service manifest.

  6. Select the Target group and then select the Targets tab. Under Registered targets, you should see three IP addresses of the three replicas deployed in a previous step. Wait until the status of all targets is healthy before continuing. It might take several minutes before all targets are healthy. The targets might be in an unhealthy state before changing to a healthy state.

  7. Send traffic to the service replacing xxxxxxxxxx-xxxxxxxxxxxxxxxx and cn-north-1 with the values returned in the output for a previous step for EXTERNAL-IP. If you deployed to a private subnet, then you'll need to view the page from a device within your VPC, such as a bastion host. For more information, see Linux Bastion Hosts on Amazon.

    curl k8s-default-samplese-xxxxxxxxxx-xxxxxxxxxxxxxxxx.elb.region-code.amazonaws.com.cn

    Example output:

    <!DOCTYPE html> <html> <head> <title>Welcome to nginx!</title> ...
  8. When you're finished with the sample deployment, service, and namespace, remove them.

    kubectl delete namespace nlb-sample-app