Google Cloud Dataflow is a fully managed service for batch and streaming data processing. Pipelines are written with the open-source Apache Beam SDKs for Java, Python and Go and run on Dataflow, which allocates worker VMs, scales automatically and removes the VMs afterwards. Organisations use it for data movement, ETL into warehouses such as BigQuery, real-time analysis and log processing. You build pipelines with Beam code, prebuilt templates or JupyterLab notebooks and monitor them in the Google Cloud console. Streaming provides exactly-once processing by default.
Dataflow is operated by Google, a US company, and exists only as a service within Google Cloud, not for self-hosting. Beam code is portable to other runners such as Apache Flink or Spark. You choose a region when launching a job; the available locations include Europe (among them Belgium and Frankfurt), the Americas and Asia Pacific. An EU region determines where workers run, but as a US provider Google remains subject to US law, including the CLOUD Act. European organisations should weigh this.