Do we have best practice, documentation or articles around proper BlockPlacementPolicy and NodeGroup topology for like say a 20 node vm cluster?

[#1]

@dvillarreal

I would say divided nodes into 2 racks and update rack for each datanode from Ambari(configure rack topology). I don't have much insights on NodeGroup topology, are you referring to node labels? Please correct me if I'm wrong.

[#2]

Looks like we currently don't have this in our docs yet. Per engineering, I will file a documentation bug. https://issues.apache.org/jira/browse/HDFS-6261

[#3 · Accepted solution]

Hi, we have an official document for deploy HDP on VMs: https://hortonworks.com/wp-content/uploads/2014/02/1514.Deploying-Hortonworks-Data-Platform-VMware-v.... It has reference link on HVE (NodeGroup) feature which include details you may want to know.