both, in streaming pipelines, you always need GBK, so there is no problem. In batch pipelines you can actually create random ids and optimistically create a lock file on hdfs, so you'll always get unique ids. If you don't have hdfs in place, you can fall back to GBK.
[ Full content available at: https://github.com/apache/beam/pull/6306 ] This message was relayed via gitbox.apache.org for [email protected]
