I already did dome research on this subject but I am not satisfied with the responses, so I am asking directly.
We have a traditional HDP cluster with admin node (1), master nodes (2) and worker nodes (15 and growing) (kerberos is enabled)
We also have some other nodes in our cluster like web application nodes (hadoop access), spring batch nodes (hadoop access), cache nodes (hadoop access) and some others without hadoop access. These nodes are not managed by Ambari (which makes the kerberos configuration a bit more complicated).
Would it be a good idea to handle these nodes like edge nodes and add them to Ambari (client config would be out of the box)? What does this mean in terms of RAM, CPU to be allocated for Ambari Agent (etc) for each "edge node"?