Running PySparkling on Databricks Azure Cluster

Sparkling Water, PySparkling and RSparkling can be used on top of Databricks Azure Cluster. This tutorial is the PySparkling.

For Scala Sparkling Water, please visit Running Sparkling Water on Databricks Azure Cluster and for RSparkling, please visit Running RSparkling on Databricks Azure Cluster.

To start Sparkling Water H2OContext on Databricks Azure, the steps are:

  1. Login into Microsoft Azure Portal

  2. Create Databricks Azure Environment

    In order to connect to Databricks from Azure, please make sure you have created the user inside Azure Active Directory and using that user for the Databricks Login.

  3. Add PySparkling dependency

    In order to create the Python library in Databricks, go to Libraries, select PiPy as the library source and type in: h2o_pysparkling_2.3 for the latest PySparkling for Spark 2.3.

    If you don’t want to train new models and just score with existing H2O-3, SW, Driverless AI MOJO models, you can use a smaller package called h2o_pysparkling_scoring_2.3 instead.

    The advantage of adding the library from PiPy, instead of uploading manually, is that the latest version is always selected and also, the dependencies are automatically resolved.

    Uploading PySparkling Library

    You can configure each cluster manually and select which libraries should be attached or you can configure the library to be attached to all future clusters. It is advised to restart the cluster in case you attached the library to the already running cluster to ensure the clean environment.

  4. Create the cluster

    • Make sure the library is attached to the cluster

    • Select Spark 2.3.4

    Example of configured cluster ready to be started
  5. Create a Python notebook and attach it to the created cluster. To start H2OContext, the init part of the notebook should be:

    from pysparkling import *
    hc = H2OContext.getOrCreate()
    
  6. And voila, we should have H2OContext running

    Running H2O Context
  7. Flow is accessible via the URL printed out after H2OContext is started. Internally we use open port 9009. If you have an environment where a different port is open on your Azure Databricks cluster, you can configure it via spark.ext.h2o.client.web.port or corresponding setter on H2OConf.