Use a python script to deploy a SQL Server big data cluster on Azure Kubernetes Service (AKS)
In this tutorial, you use a sample python deployment script to deploy SQL Server 2019 Big Data Clusters to Azure Kubernetes Service (AKS).
AKS is only one option for hosting Kubernetes for your big data cluster. To learn about other deployment options as well as how to customize deployment options, see How to deploy SQL Server Big Data Clusters on Kubernetes.
The default big data cluster deployment used here consists of a SQL Master instance, one compute pool instance, two data pool instances, and two storage pool instances. Data is persisted using Kubernetes persistent volumes that use the AKS default storage classes. The default configuration used in this tutorial is suitable for dev/test environments.
- An Azure subscription.
- Big data tools:
- Azure Data Studio
- SQL Server 2019 extension
- Azure CLI
Log in to your Azure account
The script uses Azure CLI to automate the creation of an AKS cluster. Before running the script, you must log in to your Azure account with Azure CLI at least once. Run the following command from a command prompt.
Download the deployment script
This tutorial automates the creation of the big data cluster on AKS using a python script deploy-sql-big-data-aks.py. If you already installed python for azdata, you should be able to run the script successfully in this tutorial.
In a Windows PowerShell or Linux bash prompt, run the following command to download the deployment script from GitHub.
curl -o deploy-sql-big-data-aks.py "https://raw.githubusercontent.com/Microsoft/sql-server-samples/master/samples/features/sql-big-data-cluster/deployment/aks/deploy-sql-big-data-aks.py"
Run the deployment script
Use the following steps to run the deployment script in a Windows PowerShell or Linux bash prompt. This script will create an AKS service in Azure and then deploy a SQL Server 2019 big data cluster to AKS. You can also modify the script with other environment variables to create a custom deployment.
Run the script with the following command:
If you have both python3 and python2 on your client machine and in the path, you have to run the command using python3:
When prompted, enter the following information:
Value Description Azure subscription ID The Azure subscription ID to use for AKS. You can list all of your subscriptions and their IDs by running
az account listfrom another command line.
Azure resource group The Azure resource group name to create for the AKS cluster. Azure region The Azure region for the new AKS cluster (default westus). Machine size The machine size to use for nodes in the AKS cluster (default Standard_L8s). Worker nodes The number of worker nodes in the AKS cluster (default 1). Cluster name The name of both the AKS cluster and the big data cluster. The name of your big data cluster must be only lower case alpha-numeric characters, and no spaces. (default sqlbigdata). Password Password for the controller, HDFS/Spark gateway, and master instance (default MySQLBigData2019). Username Username for the controller user (default: admin).
The default Standard_L8s machine size may not be available in every Azure region. If you do select a different machine size, make sure that the total number of disks that can be attached across the nodes in the cluster is greater than or equal to 24. Each persistent volume claim in the cluster requires an attached disk. Currently, big data cluster requires 24 persistent volume claims. For example, the Standard_L8s machine size supports 32 attached disks, so you are able to evaluate big data clusters with a single node of this machine size.
The SQL Server
saaccount is disabled during big data cluster deployment. A new sysadmin login is provisioned in SQL Server master instance the same name specified for Username input and the password corrsponding to the Password input. Same Username and Password values are used for provisioning a controller admin user. Only user supported for gateway (Knox) is root and the password is the same as above.
The script will start by creating an AKS cluster using the parameters you specified. This step takes several minutes.
Monitor the status
After the script creates the AKS cluster, it proceeds to set necessary environment variables with the settings you specified earlier. It then calls azdata to deploy the big data cluster on AKS.
The client command window will output the deployment status. During the deployment process, you should see a series of messages where it is waiting for the controller pod:
2018-11-15 15:42:02.0209 UTC | INFO | Waiting for controller pod to be up...
After 10 to 20 minutes, you should be notified that the controller pod is running:
2018-11-15 15:50:50.0300 UTC | INFO | Controller pod is running. 2018-11-15 15:50:50.0585 UTC | INFO | Controller Endpoint: https://18.104.22.168:30080
The entire deployment can take a long time due to the time required to download the container images for the components of the big data cluster. However, it should not take several hours. If you are experiencing problems with your deployment, see Monitoring and troubleshoot SQL Server Big Data Clusters.
Inspect the cluster
At any time during deployment, you can use kubectl or azdata to inspect the status and details about the running big data cluster.
Open a new command window to use kubectl during the deployment process.
Run the following command to get a summary of the status of the whole cluster:
kubectl get all -n <your-big-data-cluster-name>
If you did not change the big data cluster name, the script defaults to sqlbigdata.
Inspect the kubernetes services and their internal and external endpoints with the following kubectl command:
kubectl get svc -n <your-big-data-cluster-name>
You can also inspect the status of the kubernetes pods with the following command:
kubectl get pods -n <your-big-data-cluster-name>
Find out more information about a specific pod with the following command:
kubectl describe pod <pod name> -n <your-big-data-cluster-name>
For more details about how to monitor and troubleshoot a deployment, see Monitoring and troubleshoot SQL Server Big Data Clusters.
Connect to the cluster
When the deployment script finishes, the output notifies you of success:
2018-11-15 16:10:25.0583 UTC | INFO | Cluster state: Ready 2018-11-15 16:10:25.0583 UTC | INFO | Cluster deployed successfully.
The SQL Server big data cluster is now deployed on AKS. You can now use Azure Data Studio to connect to the cluster. For more information, see Connect to a SQL Server big data cluster with Azure Data Studio.
If you are testing SQL Server Big Data Clusters in Azure, you should delete the AKS cluster when finished to avoid unexpected charges. Do not remove the cluster if you intend to continue using it.
The following steps tears down the AKS cluster which removes the SQL Server big data cluster as well. If you have any databases or HDFS data that you want to keep, back that data up before deleting the cluster.
Run the following Azure CLI command to remove the big data cluster and the AKS service in Azure (replace
<resource group name> with the Azure resource group you specified in the deployment script):
az group delete -n <resource group name>
The deployment script configured Azure Kubernetes Service and also deployed a SQL Server 2019 big data cluster. You can also choose to customize future deployments through manual installations. To learn more about how big data clusters are deployed as well as how to customize deployments, see How to deploy SQL Server Big Data Clusters on Kubernetes.
Now that the SQL Server big data cluster is deployed, you can load sample data and explore the tutorials: