Kinesis Connector for Structured Streaming

Implementation of Kinesis Source Provider in Spark Structured Streaming. SPARK-18165 describes the need for such implementation. More details on the implementation can be read in this blog

Downloading and Using the Connector

The connector is available from the Maven Central repository. It can be used using the --packages option or the spark.jars.packages configuration property. Use the following connector artifact

Spark 3.0: com.qubole.spark/spark-sql-kinesis_2.12/1.2.0-spark_3.0
Spark 2.4: com.qubole.spark/spark-sql-kinesis_2.11/1.2.0-spark_2.4

Developer Setup

Checkout kinesis-sql branch depending upon your Spark version. Use Master branch for the latest Spark version

Spark version 3.0.x

git clone git@github.com:qubole/kinesis-sql.git
git checkout master
cd kinesis-sql
mvn install -DskipTests

This will create target/spark-sql-kinesis_2.12-*.jar file which contains the connector code and its dependency jars.

How to use it

Setup Kinesis

Refer Amazon Docs for more options

Create Kinesis Stream

$ aws kinesis create-stream --stream-name test --shard-count 2

Add Records in the stream

$ aws kinesis put-record --stream-name test --partition-key 1 --data 'Kinesis'
$ aws kinesis put-record --stream-name test --partition-key 1 --data 'Connector'
$ aws kinesis put-record --stream-name test --partition-key 1 --data 'for'
$ aws kinesis put-record --stream-name test --partition-key 1 --data 'Apache'
$ aws kinesis put-record --stream-name test --partition-key 1 --data 'Spark'

Example Streaming Job

Refering $SPARK_HOME to the Spark installation directory.

Open Spark-Shell

$SPARK_HOME/bin/spark-shell --jars target/spark-sql-kinesis_2.11-2.2.0.jar

Subscribe to Kinesis Source

// Subscribe the "test" stream
scala> :paste
val kinesis = spark
	.readStream
	.format("kinesis")
	.option("streamName", "spark-streaming-example")
   	.option("endpointUrl", "https://kinesis.us-east-1.amazonaws.com")
    .option("awsAccessKeyId", [ACCESS_KEY])
    .option("awsSecretKey", [SECRET_KEY])
    .option("startingposition", "TRIM_HORIZON")
	.load

Check Schema

scala> kinesis.printSchema
root
|-- data: binary (nullable = true)
|-- streamName: string (nullable = true)
|-- partitionKey: string (nullable = true)
|-- sequenceNumber: string (nullable = true)
|-- approximateArrivalTimestamp: timestamp (nullable = true)

Word Count

// Cast data into string and group by data column
scala> :paste

	 kinesis
    .selectExpr("CAST(data AS STRING)").as[(String)]
    .groupBy("data").count()
	.writeStream
	.format("console")
    .outputMode("complete") 
	.start()
	.awaitTermination()

Output in Console

+------------+-----+
|        data|count|
+------------+-----+
|         for|    1|
|      Apache|    1|
|       Spark|    1|
|     Kinesis|    1|
|   Connector|    1|
+------------+-----+

Using the Kinesis Sink

// Cast data into string and group by data column
    scala> :paste
    kinesis
    .selectExpr("CAST(rand() AS STRING) as partitionKey","CAST(data AS STRING)").as[(String,String)]
    .groupBy("data").count()
    .writeStream
    .format("kinesis")
    .outputMode("update") 
    .option("streamName", "spark-sink-example")
    .option("endpointUrl", "https://kinesis.us-east-1.amazonaws.com")
    .option("awsAccessKeyId", [ACCESS_KEY])
    .option("awsSecretKey", [SECRET_KEY])
    .start()
    .awaitTermination()

Kinesis Source Configuration

Option-Name	Default-Value	Description
streamName	-	Name of the stream in Kinesis to read from
endpointUrl	https://kinesis.us-east-1.amazonaws.com	end-point URL for Kinesis Stream
awsAccessKeyId	-	AWS Credentials for Kinesis describe, read record operations
awsSecretKey	-	AWS Credentials for Kinesis describe, read record operations
awsSTSRoleARN	-	AWS STS Role ARN for Kinesis describe, read record operations
awsSTSSessionName	-	AWS STS Session name for Kinesis describe, read record operations
awsUseInstanceProfile	true	Use Instance Profile Credentials if none of credentials provided
startingPosition	LATEST	Starting Position in Kinesis to fetch data from. Possible values are "latest", "trim_horizon", "earliest" (alias for trim_horizon), or JSON serialized map shardId->KinesisPosition
failondataloss	true	fail the streaming job if any active shard is missing or expired
kinesis.executor.maxFetchTimeInMs	1000	Maximum time spent in executor to fetch record from Kinesis per Shard
kinesis.executor.maxFetchRecordsPerShard	100000	Maximum Number of records to fetch per shard
kinesis.executor.maxRecordPerRead	10000	Maximum Number of records to fetch per getRecords API call
kinesis.executor.addIdleTimeBetweenReads	false	Add delay between two consecutive getRecords API call
kinesis.executor.idleTimeBetweenReadsInMs	1000	Minimum delay between two consecutive getRecords
kinesis.client.describeShardInterval	1s (1 second)	Minimum Interval between two ListShards API calls to consider resharding
kinesis.client.numRetries	3	Maximum Number of retries for Kinesis API requests
kinesis.client.retryIntervalMs	1000	Cool-off period before retrying Kinesis API
kinesis.client.maxRetryIntervalMs	10000	Max Cool-off period between 2 retries
kinesis.client.avoidEmptyBatches	false	Avoid creating an empty microbatch job by checking upfront if there are any unread data in the stream before the batch is started

Kinesis Sink Configuration

Option-Name	Default-Value	Description
streamName	-	Name of the stream in Kinesis to write to
endpointUrl	https://kinesis.us-east-1.amazonaws.com	The aws endpoint of the kinesis Stream
awsAccessKeyId	-	AWS Credentials for Kinesis describe, read record operations
awsSecretKey	-	AWS Credentials for Kinesis describe, read record
awsSTSRoleARN	-	AWS STS Role ARN for Kinesis describe, read record operations
awsSTSSessionName	-	AWS STS Session name for Kinesis describe, read record operations
awsUseInstanceProfile	true	Use Instance Profile Credentials if none of credentials provided
kinesis.executor.recordMaxBufferedTime	1000 (millis)	Specify the maximum buffered time of a record
kinesis.executor.maxConnections	1	Specify the maximum connections to Kinesis
kinesis.executor.aggregationEnabled	true	Specify if records should be aggregated before sending them to Kinesis
kniesis.executor.flushwaittimemillis	100	Wait time while flushing records to Kinesis on Task End

Roadmap

We need to migrate to DataSource V2 APIs for MicroBatchExecution.
Maintain Per Micro-Batch Shard Commit state in Dynamo DB

Acknowledgement

This connector would not have been possible without reference implemetation of Kafka connector for Structured streaming, Kinesis Connector for Legacy Streaming and Kinesis Client Library. Structure of some part of the code is influenced by the excellent work done by various Apache Spark Contributors.

RapidAPI/kinesis-sql