Migrating from Storm to Flink

After understanding the differences and similarities of Storm and Flink, you can migrate your Storm topologies to Flink data pipelines. When your Flink application is ready, you need to submit your Flink job to your Cloudera Streaming Analytics (CSA) cluster. Using the Flink Dashboard and Atlas you can monitor your application and job metadata.

The following high-level steps summarize the migration process from Storm to Flink:
  1. Identify the Flink application sources and sinks in your Storm topology.
  2. Identify the Flink application business logic from your Storm topology.
  3. Map your Storm topology to the Flink data pipeline.
    1. If you are using Trident, you need to consider how Flink handles state in itself. For more information, see State handling in Flink and the Stateful Tutorial.
  4. Build your Flink application project.

After creating the Flink application, you only need to set up the CSA cluster, and submit your Flink application to your cluster.