Typically, I create two containers on my clusters. Why? Some background first.



I assembled a mostly self service instance of VMware vCenter on top of Nutanix. Users can deploy OVAs and have access to VMs in their resource pools and folders. I group users by team and create resource pools and folders with specific access levels. We give access to AD users to either the folder or both the folder and resource pool depending on if they can create VMs (or deploy an OVA). Most users just get VM Console Access but some do get more of a power user privilege.



Because users can't follow process or procedures, we often get thick provisioned VM. While thick provisioned VMs don't actually get created in their full size instead Nutanix reserved (Explicit Reserved) the space even though the vdisk is thin provisioned on storage. In the past, this was not obvious and was not well reflected in free logical space.



So, I create two containers. Periodically, I use VMware storage vMotion to move VMs between the primary container and the administrative container and change to thin provisioned in the process. I then move them all back to the primary container. I also get to take advantage of compression in cases where compression didn't occur on the initial VM creation. Also, I can ensure fingerprinting for new features like Erasure Coding though it will happen automatically, eventually.



Another usage is when I have multiple VMware clusters residing on a single Nutanix cluster. While this is not advisable (due to the risk of concurrent maintenance actions on multiple host in different clusters without realizing the impact of a single Nutanix cluster) I do it to have unique datastores for the different clusters. When I split clusters, it was easier to move the VMs to the new Nutanix cluster (exploiting snapshots in a PD domain).



So, while multiple containers is frowned upon, it has it's uses.



Joe