Document Deployment of a Flink and Spark Clusters with Portable Beam

The Apache Beam vision has been to provide a framework for users to write and execute pipelines on the programming language of your choice, and the runner of your choice. As the reality of Beam has evolved towards this vision, the way in which Beam is run on top of runners such as Apache Spark and Apache Flink has changed.

These changes are documented in the wiki and in design documents, and are accessible for Beam contributors; but they are not available in the user-facing documentation. This has been a barrier of adoption for other users of Beam.

This project involves improving the Flink Runner page[1] to include strategies to deploy Beam on a few different environments: A Kubernetes cluster, a Google Cloud Dataproc cluster, and an AWS EMR cluster. There are other places in the documentation that should be updated in this regard[4][5].

After working on the Flink Runner, then similar updates should be made to the Spark Runner page[2], and the getting started documentation[3].

[1] https://beam.apache.org/documentation/runners/flink/
[2] https://beam.apache.org/documentation/runners/spark/
[3] https://beam.apache.org/get-started/beam-overview/
[4] https://beam.apache.org/documentation/sdks/python-streaming/
[5] https://beam.apache.org/documentation/sdks/python-streaming/#unsupported-features