Deploying with Azure Container Instances (ACI)

Owned by Denis Pitul
Last updated: Jul 28, 2023 by Ciprian LUCACI
5 min read

 

Step 1. Create Resource Group

Create the resource group where all the LHM dependent resources will reside.

  • 1 Resource Group that contains the following:

    • 1 Azure SQL Server:

      • Recommended Type: S3

    • 1 Azure Key Vault:

      • Recommended Type: Standard

    • 1 Azure Storage Account

    • 1 Service Principal in Azure AD with “ID Token” enabled

 

Once all deployment steps are complete, the provided resource group will include the following resources:

Storage Account: used for storing all telemetry data from the Databricks workspaces and consumption/cost/usage detail data

KeyVault: used for storing the storage account access key (if enabled for the deployment), the Azure AD App Registration client secret (for Azure AD Single Sign On into the application and optionally for accessing the Azure Blob Storage account), and finally the SQL Server Login password.

SQL Server database: used for storing the output of the analyzer and consumption data processors, feeds all the data required by the reports and dashboards

 

Step 2. Create Storage Account

Inside that resource group create a storage account, with two file shares called

  • logs

  • keystore

 

Step 3. Create a KeyVault

 

Step 4. Create a Azure SQL Server with Azure SQL Database

Create a Azure SQL Server and inside it create a Azure SQL Database with SQL authentication enabled.

 

Step 5. Create Database secret

The password you will save as a secret in the Key Vault with the following secret name: mssql-password

 

Step 6. Create a Azure App Registration

Create a Azure App Registration and fill in the Redirect URI with the following value:

https://<dns label used below>.<location>.azurecontainer.io/login/oauth2/code/azure

 

Step 7. Edit App Roles

On the App Roles section in the Azure App Registration configuration panel, you can create the LHM admin and user roles which you can later use to control users access to the Lakehouse Optimizer.

 

Step 8. Create Client Secret

Create a client secret for your Application and save the secret in the KeyVault you’ve created with the following secret name msft-provider-auth-secret

 

Step 9. Create Managed Identity

Create a managed Identity that will be used by the Application to communicate to the keyvault and storage account.

 

Step 10. Provide storage access to the SP

In order for the app to use the Service Principal to access the storage account you’ll need to:

  1. Add the Contributor role to the Managed Identity you’ve created

    1. Open the Storage Account

    2. Go to Access Control (IAM)

    3. Click on Add and follow the instructions to add the Contributor role to the Managed Identity

  2. Add a permission mask on the container you’re going to be using in order for the Managed Identity to be able to write on it

    1. From the Storage Account view, open the Containers section

    2. Go to the container that will be used by LHM (the name of it is reflected within the STORAGE_AZURE_CONTAINER environment variable)

    3. Go to Manage ACL

    4. Click on Add principal and follow the steps to a mask for the Managed Identity

    5. Make sure, once added, that you check all checkboxes for this new entry: Read, Write, Execute in order to give it access to use the container

Step 11. Edit Keyvault Access policies

Add the necessary access policy on the Key Vault so the managed identity is able to list and read secrets from it.

  • In the Azure Portal, go to your KeyVault created in step 3.

  • In the left hand sidebar click on Access Policies

  • Click on Create

  • From the Secrets section check List and Get as the permissions to be added

  • Search for the Managed Identity created on step 9 and select it. Click Next.

  • Skip the Application step as it’s optional

  • Click Create

Step 12. Configure SSL Communication

For SSL communication you need to create a keystore containing your server private key and certificate. The keystore will be named bplm.p12 and have no password.

Note: If this is not convenient you can change it but you’ll need to change the environment variables below as well.

Command you can use to generate the keystore:

openssl pkcs12 -export -out bplm.p12 -inkey ssl.key -in ssl.crt

 

Step 13. Create YAML definitions

Create YAML definition for the containers

apiVersion: '2019-12-01' location: <location> name: <name of your container instance> properties: containers: - name: lakehouse-monitor properties: environmentVariables: - name: APPLICATION_LOG_HTTPHEADER value: false - name: APPSERVICE_URL value: <FQDN> - name: AZURE_KEYVAULT_ENABLED value: true - name: AZURE_KEYVAULT_TENANTID value: <KV tenantID> - name: AZURE_KEYVAULT_URI value: <KV URL> - name: AZURE_MANAGED_IDENTITY_ID value: <managed identity objectID> - name: AZURE_MANAGED_IDENTITY_APP_ID value: <managed identity clientID> - name: LOG_LEVEL value: info - name: LOG_LEVEL_APP value: info - name: LOG_LEVEL_HTTP_HEADERS value: error - name: USE_SP_FOR_BACKGROUND_PROCESSORS value: true - name: USE_SP_FOR_STORAGE_ACCOUNT value: true - name: MICROSOFT_PROVIDER_AUTHENTICATION_SECRET value: ${msft-provider-auth-secret} - name: SERVICE_PRINCIPAL_CLIENT_SECRET value: ${msft-provider-auth-secret} - name: SERVICE_PRINCIPAL_CLIENTID value: <sp clientID> - name: SERVICE_PRINCIPAL_OBJECTID value: <sp objectID> - name: SERVICE_PRINCIPAL_TENANTID value: <sp tenantID> - name: SQL_DATABASE value: <sql database name> - name: SQL_SERVER_HOST value: <sql server host> - name: SQL_USER value: <sql username> - name: SQL_PASSWORD value: ${mssql-password} - name: STORAGE_AZURE_ACCOUNT value: <storage account name> - name: STORAGE_AZURE_CONTAINER value: lakehouse-monitor - name: CLOUD_PROVIDER value: azure - name: AUTHENTICATION_PROVIDER value: active-directory - name: SERVER_SERVLET_SESSION_PERSISTENT value: true - name: SERVER_SSL_ENABLED value: true - name: SERVER_SSL_KEY_STORE value: /keystore/bplm.p12 # - name: SERVER_SSL_KEY_STORE_PASSWORD # value: bplm # - name: SERVER_SSL_KEY_ALIAS # value: bplm # - name: SERVER_SSL_KEY_PASSWORD # value: bplm - name: SERVER_SSL_KEY_STORE_TYPE value: PKCS12 - name: METRIC_PROCESSOR_ENABLED value: true - name: METRIC_PROCESSOR_DOCTOR_INITIAL_DELAY value: PT1M - name: CONSUMPTION_USE_PREFILTER value: false - name: ADMIN_APP_ROLE value: <admin role value> - name: EXECUTIVE_APP_ROLE value: <executive role value> - name: AUTHORIZATION_CACHE_TIMEOUT value: 1800 - name: SERVER_PORT value: 443 image: blueprint.azurecr.io/bpcs/lakehouse-optimizer:2.3 ports: - port: 443 - port: 80 resources: requests: cpu: 4 memoryInGB: 8 volumeMounts: - mountPath: /var/log name: logs - mountPath: /keystore name: keystore imageRegistryCredentials: - server: blueprint.azurecr.io username: <ACR Username> password: <ACR Password> osType: Linux restartPolicy: Always ipAddress: type: Public ports: - port: 443 - port: 80 dnsNameLabel: <dns label> volumes: - name: logs azureFile: sharename: <name of share for the logs> storageAccountName: <storage account name where the share is created> storageAccountKey: <storage account key> - name: keystore azureFile: sharename: <name of share for the logs> storageAccountName: <storage account name where the share is created> storageAccountKey: <storage account key> tags: {} type: Microsoft.ContainerInstance/containerGroups

 

Step 14. Create Containers

Create the containers:

  • recommended minimum 8 cores