Our organization has a similar use case. We have more than 10 platform instances, each with hundreds of schemas. 1000+ application teams (program/scrum) manage dataset entities in these schemas. For example, within a schema, 50 tables are owned by application team 1, and another 75 tables are owned by application team 2.
We are exploring how to configure metadata management governance policies for this scenario. Here are some thoughts that have not yet been tested:
1. We will create a custom object/business term (glossary) of applications.
2. We will tag dataset entities with the appropriate application names.
3. We will create tag-based Ranger policies that leverage application names and platform instances.
If there are any other approaches, we would like to explore them.
References:
https://datahubproject.io/docs/how/configuring-authorization-with-apache-ranger/
https://cwiki.apache.org/confluence/pages/viewpage.action?pageId=61322361