You can now use Apache Flink 1.5.0 and Apache Livy 0.5.0 on Amazon EMR release 5.16.0. Flink 1.5.0 has several new features, including a new deployment model that supports dynamic resource allocation on YARN for efficient resource utilization, support for a broadcast state that allows replication of configuration rules and patterns to all parallel instances of a function, a new SQL CLI client to run ad-hoc queries on data streams, ability to store state information on local disks for faster and efficient failure recovery, and other performance and throughput updates for streaming jobs. Livy 0.5.0 brings in support for SparkSQL interpreter for running interactive SQL queries, ability to use multiple languages and share the SparkContext in a single session, ability to share variables across jobs, and other improvements. Additionally, you can now use the upgraded version of Apache Hadoop 2.8.4, Apache Spark 2.3.1, Presto 0.203, Apache Phoenix 4.14.0 and Apache MXNet 1.2.0.
You can create an Amazon EMR cluster with the release 5.16.0 by choosing the release label “emr-5.16.0” from the AWS Management Console, AWS CLI, or SDK. You can choose Flink, Livy, Hadoop, Spark, Presto, Phoenix, and MXNet to install these applications when you launch your EMR cluster. Please visit the Amazon EMR documentation for more information about the EMR release 5.16.0, Flink 1.5.0, Livy 0.5.0, Spark 2.3.1, Presto 0.203, Hadoop 2.8.4, Phoenix 4.14.0, and MXNet 1.2.0
Amazon EMR release 5.16.0 is now available in all supported regions for Amazon EMR.