Deploy on Amazon Elastic Kubernetes Service (EKS)

What is Amazon EKS?

Amazon Elastic Kubernetes Service (Amazon EKS) is a managed service that eliminates the need to install, operate, and maintain your own Kubernetes control plane on Amazon Web Services (AWS). This deployment guide covers setting up a highly available SurrealDB cluster backed by TiKV on Amazon EKS.

Requirements

An account on Amazon Web Services (AWS) and also the necessary Amazon Identity and Access Management (Amazon IAM) permissions to programmatically create resources in your account.
kubectl To manage the Kubernetes cluster.
AWS CLI installed and configured with your user’s credentials
eksctl installed

COST CONSIDERATIONS
Provisioning the environment in your AWS account will create resources and there will be cost associated with them. The cleanup section provides a guide to remove them, preventing further charges.

Important
This guide was tested in eu-west-1 (Ireland region) and it follows TiKV best practices for scalability and high availability. It will provision up to 12 Amazon Elastic Compute Cloud (Amazon EC2) instances, several Amazon Elastic Block Storage (Amazon EBS) drives, and up to three Amazon Elastic Loadbalancers (Amazon ELB). The forecasted cost to run this guide is $5 USD per hour.

Building an EKS Cluster

This section outlines how to build a cluster by using the eksctl tool. The following is the configuration that will be used to build the cluster:

SURREALDB CLUSTER CONFIG
apiVersion: eksctl.io/v1alpha5
kind: ClusterConfig
metadata:
  name: ${CLUSTER_NAME}
  region: ${AWS_REGION}
  version: '1.27'
  tags:
    karpenter.sh/discovery: ${CLUSTER_NAME}
    env: ${CLUSTER_NAME}
iam:
  withOIDC: true
vpc:
  cidr: 10.0.0.0/16
  clusterEndpoints:
    privateAccess: true
    publicAccess: true
addons:
- name: vpc-cni
  resolveConflicts: overwrite
- name: aws-ebs-csi-driver
  wellKnownPolicies:      # add IAM and service account
    ebsCSIController: true

managedNodeGroups:

  - name: admin
    desiredCapacity: 1
    privateNetworking: true
    labels:
      dedicated: admin

  - name: pd-1a
    desiredCapacity: 1
    privateNetworking: true
    availabilityZones: ["${AWS_REGION}a"]
    instanceType: c5.xlarge
    labels:
      dedicated: pd
    taints:
    - key: "dedicated"
      value: "pd"
      effect: NoSchedule

  - name: pd-1b
    desiredCapacity: 1
    privateNetworking: true
    availabilityZones: ["${AWS_REGION}b"]
    instanceType: c5.xlarge
    labels:
      dedicated: pd
    taints:
    - key: "dedicated"
      value: "pd"
      effect: NoSchedule

  - name: pd-1c
    desiredCapacity: 1
    privateNetworking: true
    availabilityZones: ["${AWS_REGION}c"]
    instanceType: c5.xlarge
    labels:
      dedicated: pd
    taints:
    - key: "dedicated"
      value: "pd"
      effect: NoSchedule

  - name: tikv-1a
    desiredCapacity: 1
    privateNetworking: true
    availabilityZones: ["${AWS_REGION}a"]
    instanceType: r5b.2xlarge
    labels:
      dedicated: tikv
    taints:

    - key: "dedicated"
      value: "tikv"
      effect: NoSchedule

  - name: tikv-1b
    desiredCapacity: 1
    privateNetworking: true
    availabilityZones: ["${AWS_REGION}b"]
    instanceType: r5b.2xlarge
    labels:
      dedicated: tikv
    taints:
    - key: "dedicated"
      value: "tikv"
      effect: NoSchedule

  - name: tikv-1c
    desiredCapacity: 1
    privateNetworking: true
    availabilityZones: ["${AWS_REGION}c"]
    instanceType: r5b.2xlarge
    labels:
      dedicated: tikv
    taints:
    - key: "dedicated"
      value: "tikv"
      effect: NoSchedule

  - name: default
    desiredCapacity: 2
    minSize: 2
    maxSize: 4
    instanceType: m5.large
    privateNetworking: false

Based on this configuration eksctl will:

Create a VPC across three availability zones
Create an EKS cluster
Create an IAM OIDC provider
Create the necessary node groups for TiKV according to best practices
Configure the VPC CNI addon to manage networking
Configure the EBS addon to manage storage volumes

Save the above configuration in a file named surrealdb-cluster.yml and apply the configuration file like so:

DEPLOY EKS CLUSTER
export CLUSTER_NAME=surrealdb-cluster
export AWS_REGION=eu-west-1

envsubst < surrealdb-cluster.yml | eksctl create cluster -f -

aws eks update-kubeconfig --name $CLUSTER_NAME

Note
The deployment of the cluster should take about 30 minutes.

Deploy TiKV

The following instructions will install TiKV operators in your EKS cluster.

CREATE TIKV CLUSTER
kubectl create namespace tidb-cluster

curl -O https://raw.githubusercontent.com/pingcap/tidb-operator/v1.6.1/examples/aws/tidb-cluster.yaml

kubectl apply -f tidb-cluster.yaml -n tidb-cluster

Install ALB Controller

The following instructions will install the AWS Load Balancer Controller.

The AWS Load Balancer Controller provisions and manages the necessary AWS resources when Kubernetes creates an Ingress or a LoadBalancer

Formerly known as AWS ALB Ingress Controller, it’s an open-source project on GitHub.

INSTALL ALB CONTROLLER
export ACCOUNT_NUMBER=$(aws sts get-caller-identity --query "Account" --output text)

curl -O https://raw.githubusercontent.com/kubernetes-sigs/aws-load-balancer-controller/v2.5.4/docs/install/iam_policy.json

aws iam create-policy \
    --policy-name AWSLoadBalancerControllerIAMPolicy \
    --policy-document file://iam_policy.json

eksctl create iamserviceaccount \
  --cluster=$CLUSTER_NAME \
  --namespace=kube-system \
  --name=aws-load-balancer-controller \
  --role-name AmazonEKSLoadBalancerControllerRole_${CLUSTER_NAME} \
  --attach-policy-arn=arn:aws:iam::$ACCOUNT_NUMBER:policy/AWSLoadBalancerControllerIAMPolicy \
  --approve

helm repo add eks https://aws.github.io/eks-charts
helm repo update eks

helm install aws-load-balancer-controller eks/aws-load-balancer-controller \
  -n kube-system \
  --set clusterName=$CLUSTER_NAME \
  --set serviceAccount.create=false \
  --set serviceAccount.name=aws-load-balancer-controller

Install SurrealDB

The following script will install SurrealDB on your EKS cluster backed by TiKV with a public endpoint exposed via an Application Load Balancer (ALB).

INSTALL SURREALDB
helm repo add surrealdb https://helm.surrealdb.com
helm repo update

TIKV_URL=tikv://basic-pd.tidb-cluster:2379

helm install \
    --set surrealdb.path=$TIKV_URL \
    --set surrealdb.auth=false \
    --set service.type="NodePort" \
    --set ingress.enabled=true \
    --set ingress.className="alb" \
    --set ingress.annotations."alb\.ingress\.kubernetes\.io/healthcheck-path"=/health \
    --set ingress.annotations."alb\.ingress\.kubernetes\.io/load-balancer-name"="ingress-${CLUSTER_NAME}" \
    --set ingress.annotations."alb\.ingress\.kubernetes\.io/scheme"="internet-facing" \
    --set ingress.annotations."alb\.ingress\.kubernetes\.io/target-type"="ip" \
    --set ingress.annotations."meta\.helm\.sh/release-name"="surrealdb-tikv" \
    --set ingress.annotations."meta\.helm\.sh/release-namespace"="default" \
    surrealdb-tikv surrealdb/surrealdb

Test your SurrealDB Installation

You can get the endpoint to use with your SurrealDB client as follows:

Test installation
export SURREALDB_ENDPOINT=$(kubectl get ingress surrealdb-tikv -o json | jq '.status.loadBalancer.ingress[0].hostname' | xargs)
echo $SURREALDB_ENDPOINT

# sample output
> ingress-27v2-902764750.eu-west-1.elb.amazonaws.com

Test your connection with the following command:

Connection test
surreal sql -e https://$SURREALDB_ENDPOINT
>

Cleanup

Cleanup can be performed with the following commands.

Cleanup
kubectl delete -f tidb-cluster.yaml -n tidb-cluster
helm uninstall surrealdb-tikv
helm -n kube-system uninstall aws-load-balancer-controller
helm -n tidb-admin uninstall tidb-operator

eksctl delete cluster \
    --force \
    --disable-nodegroup-eviction \
    --parallel 10 \
    --name $CLUSTER_NAME

The default cleanup behaviour is to preserve resources such as EBS volumes that were previously attached to your cluster. If this is not what you want, and in order to prevent you from incurring in additional charges related to the usage of these block storage devices, navigate to the AWS console and manually delete all volumes that were attached to your cluster, as shown in figure.

Edit this page on GitHub

Deploy on Kubernetes Deploy on Google GKE

SurrealDB