Configuring Databricks

Databricks is a unified, open analytics platform for building, deploying, sharing, and maintaining enterprise-grade data, analytics, and AI solutions. Databricks is used for ETL and for managing security, governance, data discovery among it's various other features.

The Lazsa Platform uses Databricks for various operations like data integration, data transformation, and data quality. After you save the connection details for Databricks, you can use it in any of the nodes in a data pipeline, as mentioned earlier.

Prerequisites

The following permissions are required for configuring Databricks:

  • s3:ListBucket

  • s3:PutObject

  • s3:GetObject

  • s3:DeleteObject

  • s3:PutObjectAcl

To access an Amazon S3 bucket from Databricks, you must create an instance profile with read, write, and delete permissions. For detailed instructions on how to create an instance profile, refer to the following link: instance-profile-tutorial.html

To configure the connection details of your Databricks, do the following:

  1. Sign in to the Lazsa Platform and click Configuration in the left navigation pane.
  2. On the Platform Setup screen, on the Cloud Platform, Tools & Technologies tile, click Configure.
  3. On the Cloud Platform, Tools & Technologies screen, in the Data Integration section, click Configure.
    Data Integration Tools
    4. (After you save your first connection details in this section, you see the Modify button here.)
  4. On the Databricks screen, do the following:

    1. In the Details section, provide the following details:

      FieldDescription
      NameGive a unique name to your Databricks configuration. This name is used to save and identify your specific Databricks connection details within the Lazsa Platform.
      DescriptionProvide a brief description that helps you identify the purpose or context of this Databricks configuration.

    1. In the Configuration section, provide the following information:

      FieldDescription
      Databricks URL

      Provide the URL for the configured Databricks instance.

      Select Cloud Provider

      Select the cloud provider on which the Databricks cluster is deployed.

      Note:

      You must select the same cloud service provider on which the Databricks cluster is deployed. For example, if Databricks is deployed on Azure, do not select AWS as the cloud service provider.

    2. Depending on the cloud provider on which the Databricks cluster is deployed and how you want to retrieve the credentials to connect to Databricks, do one of the following:

      FieldDescription
      Connect using Lazsa Orchestrator Agent

      Enable this option to resolve your Databricks credentials within your private network via Lazsa Orchestrator Agent without sharing them with the Lazsa Platform.

      • If the Cloud Service Provider is AWS, then provide the following details:

        • Select the Lazsa Orchestrator Agent for AWS

        • Secret Name - provide the name with which the secret is stored in AWS Secrets Manager.

      Databricks Configuration AWS Secrets Manager

      • If the Cloud Service Provider is Azure, then provide the following details:

        • Select the Lazsa Orchestrator Agent for Azure

        • Vault Name - provide the name of the vault where the secret is stored in Azure Key Vault.

      Databricks Configuration Azure Key Vault

      Select Secret Manager
      • Select Lazsa, the user credentials are securely stored in the Lazsa-managed secrets store.
        • Databricks API Token Key - provide the key for Databricks API token.

        Databricks Configuration No Orchestrator Lazsa- managed secrets

      • Select AWS Secrets Manager.
        • In the Secret Management Tool dropdown list, the AWS Secrets Manager configurations that you save and activate in the Secret Management section on the Cloud Platform, Tools & Technologies screen are listed for selection. Select the configuration of your choice.
        • Secret Name - Provide the secret name, with which the secrets for Databricks are stored.

          Databricks Configuration No Orchestrator AWS Secrets Manager
      Databricks API TokenProvide the Databricks API token.
      Provisions user access using Azure AD Group 
      Test ConnectionClick Test Connection to verify the provided details and ensure that the Lazsa Platform is able to successfully connect to the configured Databricks instance.

    3. In the Databricks Resource Configuration section, provide the following information:

      FieldDescription
      Organization ID

      Provide the organization ID. When you log into your Databricks workspace, observe the URL in the address bar of the browser. The numbers following the o= in the URL make up the organization ID.

      For example if the URL is https://abcd-teste2-test-spcse2.cloud.databricks.com/?o=2281745829657864#, the organization ID is 2281745829657864.

      Cluster Name

      Do one of the following:

      Workspace Parent FolderProvide the name of the parent folder in the Databricks workspace that you are configuring.
      Auto Clone Custom Code to Databricks ReposEnable this option if you want your custom code to be automatically cloned to Repos in Databricks. If you disable this option, Lazsa uploads the custom code to the Repos in Databricks.
      Secret Scope

      A secret scope is a collection of secrets that is stored in an encrypted database owned and managed by Databricks Azure. It allows Databricks to use the credentials stored in it locally, eliminating the need to connect to an external secrets management tool, every time a job is run in a data pipeline. You can either use an existing secret scope from the ones already created in Databricks or create a new one. Do one of the following:

      • Select a secret scope from the dropdown list.

      • Click + New Secret Scope to create a new secret scope. Provide a name for the new secret scope and click Create.Create New Secret Scope

    4. Secure configuration details with a password
      To password-protect your Databricks connection details, turn on this toggle, enter a password, and then retype it to confirm. This is optional but recommended. When you share the connection details with multiple users, password protection helps you ensure authorized access to the connection details.

    5. Click Save Configuration. You can see the configuration listed on the Data Integration screen.

       

    Create or Edit Cluster Details

    To create a new cluster provide the following details:

    FieldDescription
    Cluster NameProvide a name for the Databricks cluster that you are creating.
    Databricks Runtime Version

    Select a version based on the type of job for which you want to use the Databricks cluster.

    Worker TypeSelect the processing capacity that you need. This is based on the use case you are trying to achieve.
    WorkersNumber of worker nodes. This depends on the kind of cluster you want to create. For example, a multi-node cluster requires a greater number of workers.
    Enable AutoscalingTurn on this toggle to control the infrastructure costs. If you enable this option, and provide values for Min Workers and Max Workers, the infrastructure costs begin at minimum workers and can scale up to maximum workers depending on the requirement.
    Terminate after below minutes of inactivityUse this option to control the infrastructure costs. Provide a value in minutes. If the Databricks cluster is inactive for the specified minutes, it is automatically terminated.
    Use latest Calibo UTIL version Enable this option to use the required utility.
    FieldDescription
    First on DemandLets you pay for the compute capacity by the second.
    AvailabilityAvailability zone for a cluster. The instance type you want may be available only in certain zones.
    ZoneSelect a zone from the available options.
    Instance Profile ARN Instance Profiles allow you to access your data from Databricks clusters. Specify the Instance Profile ARN.
    EBS Volume TypeDatabricks provisions EBS volumes by default, for efficient functioning of the cluster. You can specify the type of EBS volume, the EBS volume count and the EBS volume size.
    EBS Volume Count
    EBS Volume Size
    FieldDescription
    Spark Config

    You can fine-tune your spark jobs by providing custom Spark configuration properties. For more information see Spark configuration properties.

    Enter the configuration properties as one key-value pair per line.

    Environment Variables

    You can add the required environment variables.

    Environment variables are used to configure various aspects of the behavior of the Databricks cluster and environment. For example: The Python version to use, number of worker instances to start on each worker node, location fo the Apache Spark installation on the cluster nodes, the logging level to be set for the cluster and so on.

    Logging Path

    When you create a cluster, you specify the location where the logs for the Spark driver node, worker nodes, and events are stored.

    Select the destination from the following options:

    • DBFS

    • S3

    Log Level

    Select the level for logging that you want to set for this cluster. Choose from the following options:

    • ALL

    • DEBUG

    • ERROR

    • FATAL

    • INFO

    • OFF

    • TRACE

    • WARN

    • NONE

    Init Scripts

    Set the destination and the path of the location where the init scripts for the cluster are stored. Init scripts customize the initialization process of the cluster by executing specified commands or scripts when the cluster spins up. The scripts perform various tasks like configuring the cluster environment, installing additional software packages, setting environment variables and so on. Specify the destination and the actual path for the init scripts:

    • Destination - Select from the following options:

      • Workspace

      • DBFS

      • S3

    • Init script path - Specify the actual path on the selected destination.

    Type

    A table is created with the options that are selected in the previous step and the information is summarized as shown below:

    TypeFile PathRegionDelete
    dbfsdbfs:/logging-delete icon
    File Path
    Region
    Tags

    Tags allow you to monitor the cost of cloud resources used by various group within your organization. The tags that you add as key-value pairs are applied by Databricks to cloud resources like VMs, disk volumes and so on.

    To add a tag, provide the following information:

    • Tag - Provide a name for the tag.

    • Value - Provide a value for the tag.

 

Related Topics Link IconRecommended Topics

What's next? Cloud Platforms, Tools and Technologies