@fahed The HDFS Service inside of the DataLake is supporting of the environment, and its services. For example: Atlas. Ranger, Solr, Hbase. It's size, is based on the environment scale.
You are correct in the assumption that your end user HDFS Service is part of Data Hubs deployed around the environment. You should not try to use the environment's HDFS Service for applications and workloads that would be part of deeper Data Hubs.