Setting Cassandra-specific properties 

Use the Spark Cassandra Connector options to configure DataStax Enterprise Spark.

Spark integration uses the Spark Cassandra Connector 1.2 under the hood. You can use the configuration options defined in that project to configure DataStax Enterprise Spark. Spark recognizes system properties having the spark. prefix and adds the properties to the configuration object implicitly upon creation. You can avoid adding system properties to the configuration object by passing false for the loadDefaults parameter in the SparkConf constructor.

You pass settings for Spark, Spark Shell, and other DSE Spark built-in applications using the intermediate application spark-submit, described in Spark 1.2 documentation.

Configuring the Spark shell

You pass Spark configuration arguments using the following syntax:
dse spark [submission arguments] [application arguments]
  • Submission arguments:
    • --properties-file <path-to-properties-file>

      The location of the properties file having the configuration settings. By default, Spark loads the settings from conf/spark-defaults.conf.

    • --executor-memory <memory>.

      How much memory to allocate on each machine for the application. You can provide the memory argument in JVM format using either the k, m, or g suffix.

    • --total-executor-cores <cores>

      The total number of cores the application uses

    • --conf name=value

      An arbitrary Spark option to the Spark configuration prefixed by spark.

    • --help

      Shows a help message that displays all options except DataStax Enterprise Spark shell options.

    • -- jars <additional-jars>

      A comma-separated list of paths to additional jar files.

    • --verbose

      Displays which arguments are recognized as Spark configuration options and which are forwarded to the Spark Shell.

  • Spark shell application arguments:
    • -i <file>

      Runs a script from the specified file.

Configuring Spark applications

You pass the Spark submission arguments using the following syntax:
dse spark-submit [submission arguments] <application file> [application arguments]
  • All the submission arguments listed in the previous section, and additionally:
    • -- class <class-name>

      The full name of the application main class

    • -- name <name>

      The application name as displayed in the Spark web-app

    • -- py-files <files>

      A comma-separated list of the .zip, .egg or .py files, which will be set on PYTHONPATH for Python applications

    • -- files <files>

      A comma-separated list files that are distributed among the executors and available for the application.

    • -- master <master URL>

      The URL of the Spark Master.

  • Application file, a JAR or .py file that contains the application being run.

    Passed without any control argument; acts as a separator between Spark configuration arguments and custom application arguments.

In general, Spark submission arguments are translated into system properties -Dname=value and other VM params like classpath. The application arguments are passed directly to the application.

Spark configuration object 

Use the com.datastax.bdp.spark.DseSparkContext class to create a Spark context object to connect to DSE clusters. The DseSparkContext class is functionally the same as org.apache.spark.SparkContext.

import com.datastax.bdp.spark.DseSparkContext 
import org.apache.spark.SparkConf 

object ConfigurationExample extends App { 

  def createSparkContext() = { 
    val conf = new SparkConf()
     /* set the app name here or by using the --name option when
        you submit the app */
      .setAppName("Configuration example") 

    new DseSparkContext.apply(conf) 

  val sc = createSparkContext() 

  // ... 


Property list 

The following key Cassandra-specific properties are recognized:

The default keyspace for Spark SQL.
Default = 9042. Port for native client protocol connections.
Default = 9160. Port for thrift connections.
The host name or IP address to which the Thrift RPC service and native transport is bound. The rpc_address property in the cassandra.yaml, which is localhost by default, determines the default value of this property.

Read properties

Default = 100000. Approximate number of rows in a single Spark partition. The higher the value, the fewer Spark tasks are created. Increasing the value too much may limit the parallelism level.
Default = 1000. Number of rows being fetched per roundtrip to Cassandra. Increasing this value increases memory consumption. Decreasing the value increases the number of roundtrips.
Default = LOCAL_ONE. Consistency level to use when reading.

Write properties

You can set the following properties in SparkConf to fine tune the saving process.


Default = auto. Number of bytes per single batch. The default, auto, means the connector adjusts the number of bytes based on the amount of data.

Default = LOCAL_ONE. Consistency level to use when writing.
Default = 5. Maximum number of batches executed in parallel by a single Spark

Default = 64K. The maximum total size of the batch in bytes.

Connector 1.2 documentation describes additional, low-level properties.