Tilbakemeldinger Rediger

Table copy operations on Azure Cosmos DB Cassandra API from Spark

  • Artikkel
  • 2 minutter å lese

APPLIES TO: Cassandra API

This article describes how to copy data between tables in Azure Cosmos DB Cassandra API from Spark. The commands described in this article can also be used to copy data from Apache Cassandra tables to Azure Cosmos DB Cassandra API tables.

Cassandra API configuration

import org.apache.spark.sql.cassandra._
//Spark connector
import com.datastax.spark.connector._
import com.datastax.spark.connector.cql.CassandraConnector

//if using Spark 2.x, CosmosDB library for multiple retry
//import com.microsoft.azure.cosmosdb.cassandra

//Connection-related
spark.conf.set("spark.cassandra.connection.host","YOUR_ACCOUNT_NAME.cassandra.cosmosdb.azure.com")
spark.conf.set("spark.cassandra.connection.port","10350")
spark.conf.set("spark.cassandra.connection.ssl.enabled","true")
spark.conf.set("spark.cassandra.auth.username","YOUR_ACCOUNT_NAME")
spark.conf.set("spark.cassandra.auth.password","YOUR_ACCOUNT_KEY")
// if using Spark 2.x
// spark.conf.set("spark.cassandra.connection.factory", "com.microsoft.azure.cosmosdb.cassandra.CosmosDbConnectionFactory")

//Throughput-related...adjust as needed
spark.conf.set("spark.cassandra.output.batch.size.rows", "1")
//spark.conf.set("spark.cassandra.connection.connections_per_executor_max", "10") // Spark 2.x
spark.conf.set("spark.cassandra.connection.remoteConnectionsPerExecutor", "10") // Spark 3.x
spark.conf.set("spark.cassandra.output.concurrent.writes", "1000")
spark.conf.set("spark.cassandra.concurrent.reads", "512")
spark.conf.set("spark.cassandra.output.batch.grouping.buffer.size", "1000")
spark.conf.set("spark.cassandra.connection.keep_alive_ms", "600000000")

Note

If you are using Spark 3.0 or higher, you do not need to install the Cosmos DB helper and connection factory. You should also use remoteConnectionsPerExecutor instead of connections_per_executor_max for the Spark 3 connector (see above). You will see that connection related properties are defined within the notebook above. Using the syntax below, connection properties can be defined in this manner without needing to be defined at the cluster level (Spark context initialization).

Warning

The Spark 3 samples shown in this article have been tested with Spark version 3.0.1 and the corresponding Cassandra Spark Connector com.datastax.spark:spark-cassandra-connector-assembly_2.12:3.0.0. Later versions of Spark and/or the Cassandra connector may not function as expected.

Insert sample data

val booksDF = Seq(
   ("b00001", "Arthur Conan Doyle", "A study in scarlet", 1887,11.33),
   ("b00023", "Arthur Conan Doyle", "A sign of four", 1890,22.45),
   ("b01001", "Arthur Conan Doyle", "The adventures of Sherlock Holmes", 1892,19.83),
   ("b00501", "Arthur Conan Doyle", "The memoirs of Sherlock Holmes", 1893,14.22),
   ("b00300", "Arthur Conan Doyle", "The hounds of Baskerville", 1901,12.25)
).toDF("book_id", "book_author", "book_name", "book_pub_year","book_price")

booksDF.write
  .mode("append")
  .format("org.apache.spark.sql.cassandra")
  .options(Map( "table" -> "books", "keyspace" -> "books_ks", "output.consistency.level" -> "ALL", "ttl" -> "10000000"))
  .save()

Copy data between tables

Copy data between tables (destination table exists)

//1) Create destination table
val cdbConnector = CassandraConnector(sc)
cdbConnector.withSessionDo(session => session.execute("CREATE TABLE IF NOT EXISTS books_ks.books_copy(book_id TEXT PRIMARY KEY,book_author TEXT, book_name TEXT,book_pub_year INT,book_price FLOAT) WITH cosmosdb_provisioned_throughput=4000;"))

//2) Read from one table
val readBooksDF = sqlContext
  .read
  .format("org.apache.spark.sql.cassandra")
  .options(Map( "table" -> "books", "keyspace" -> "books_ks"))
  .load

//3) Save to destination table
readBooksDF.write
  .cassandraFormat("books_copy", "books_ks", "")
  .save()

//4) Validate copy to destination table
sqlContext
  .read
  .format("org.apache.spark.sql.cassandra")
  .options(Map( "table" -> "books_copy", "keyspace" -> "books_ks"))
  .load
  .show

Copy data between tables (destination table does not exist)

import com.datastax.spark.connector._

//1) Read from source table
val readBooksDF = sqlContext
  .read
  .format("org.apache.spark.sql.cassandra")
  .options(Map( "table" -> "books", "keyspace" -> "books_ks"))
  .load

//2) Creates an empty table in the keyspace based off of source table
val newBooksDF = readBooksDF
newBooksDF.createCassandraTable(
    "books_ks", 
    "books_new", 
    partitionKeyColumns = Some(Seq("book_id"))
    //clusteringKeyColumns = Some(Seq("some column"))
    )

//3) Saves the data from the source table into the newly created table
newBooksDF.write
  .cassandraFormat("books_new", "books_ks","")
  .mode(SaveMode.Append)
  .save()

//4) Validate table creation and data load
sqlContext
  .read
  .format("org.apache.spark.sql.cassandra")
  .options(Map( "table" -> "books_new", "keyspace" -> "books_ks"))
  .load
  .show

The output-

+-------+------------------+--------------------+----------+-------------+
|book_id|       book_author|           book_name|book_price|book_pub_year|
+-------+------------------+--------------------+----------+-------------+
| b00300|Arthur Conan Doyle|The hounds of Bas...|     12.25|         1901|
| b00001|Arthur Conan Doyle|  A study in scarlet|     11.33|         1887|
| b00023|Arthur Conan Doyle|      A sign of four|     22.45|         1890|
| b00501|Arthur Conan Doyle|The memoirs of Sh...|     14.22|         1893|
| b01001|Arthur Conan Doyle|The adventures of...|     19.83|         1892|
+-------+------------------+--------------------+----------+-------------+

import com.datastax.spark.connector._
readBooksDF: org.apache.spark.sql.DataFrame = [book_id: string, book_author: string ... 3 more fields]
newBooksDF: org.apache.spark.sql.DataFrame = [book_id: string, book_author: string ... 3 more fields]

Next steps