Solved

Snapshot within stun time

  • 4 August 2020
  • 5 replies
  • 1120 views

Hi Cloud builder

While I was reading the NTNX Bible they said that the VM snapshot being taked by freezing the VM, unlikely with ESXi where the VM is stunned » Delta created for I/Os and then VMDK is consolidated

My question while on NTNX Platform how/when the I/O takes place while VM is being snapshot’ed while its freeze.

icon

Best answer by Alona 7 August 2020, 07:41

View original

This topic has been closed for comments

5 replies

Userlevel 3
Badge +4

Hi @VMwarer 

I request to review the below post, I hope this helps

Nutanix Protection Domain and VMware VM snapshots — can they co-exist?

Userlevel 6
Badge +5

Hi Ilyas,

I believe the below is the part of the Nutanix Bible to which you are referring? Let me know if I misunderstood, please.

Eliminating ESXi Stun

ESXi has native app consistent snapshot support using VMware guest tools. However, during this process, delta disks are created and ESXi "stuns" the VM in order to remap the virtual disks to the new delta files which will handle the new write IO. Stuns will also occur when a VMware snapshot is deleted.

During this stun process the VM its OS cannot execute any operations and is essentially in a "stuck" state (e.g. pings will fail, no IO). The duration of the stun will depend on the number of vmdks and speed of datastore metadata operations (e.g. create new delta disks, etc.)

By using Nutanix VSS we completely bypass the VMware snapshot / stun process and have little to no impact to performance or VM / OS availability.

 

Hi Ilyas,

I believe the below is the part of the Nutanix Bible to which you are referring? Let me know if I misunderstood, please.

Eliminating ESXi Stun

ESXi has native app consistent snapshot support using VMware guest tools. However, during this process, delta disks are created and ESXi "stuns" the VM in order to remap the virtual disks to the new delta files which will handle the new write IO. Stuns will also occur when a VMware snapshot is deleted.

During this stun process the VM its OS cannot execute any operations and is essentially in a "stuck" state (e.g. pings will fail, no IO). The duration of the stun will depend on the number of vmdks and speed of datastore metadata operations (e.g. create new delta disks, etc.)

By using Nutanix VSS we completely bypass the VMware snapshot / stun process and have little to no impact to performance or VM / OS availability.

 

Thanks Alona and Neel

I will read it and update the thread :) 

Userlevel 6
Badge +5

Ok, so I do not have much for you and I apologise but I was able to find something.

You know how hypervisor level snapshots and storage-level snapshots work differently traditional (I am tempted to put that word in quotes) environment and thus have different implications on the in-gues applications? For example a VM level snapshot would require a freeze of the I/O but a volume level snapshot would come not noticeable or nearly not noticeable?

Similar thing here. Nutanix is an HCI platform so we can operate with the block map on the storage layer and skip the stun entirely.

That’s what the Nutanix Bible paragraph meant. I sincerely hope that helps.

Thanks Alona & Team

Sorry for being late a little, this helps me for sure to be understand.

I am rewiring my mind with NTNX still, but all is going well till now

Thanks again