VMware vSAN should manage better VM snapshots compared with traditional storage and VMFS datastores. The reason is the new (v2) on-disk format in VSAN 6.0 and the new filesystem that is used: VirstoFS.
VirstoFS is the first implementation of technology that was acquired when VMware bought a company called Virsto a number of years ago.
Also there is a new sparse format called vsanSparse. These replace the traditional vmfsSparse format (redo logs).
So you can use VM snapshots safetly also in produtions?
In my option, you still have to take care on how snapshot are used and, very important, on how they grow! Recently I’ve got a huge problem with some VMs running on vSAN 6.6 where that snapshots where big (vmdk was 1,5TB and snapshot was more than 1 TB). Also the VMs were generating a medium load on I/O.
In this case snapshot cancellation was not working at all! The VM remain in a “Consolidate needed” state with still the running snapshot on the bigger disks!
Maybe one of the reason was also the change of the default policy after the attempt to remove the snapshot.
Consolidation was not working at all: after 8 hours it goes in a generic error.
The only working solutions was to power-off the VM and use the vmfstools to create a new vmdk object without the snapshot.
Let’s assume that the current vmdk snapshot was VMname_1-000001.vmdk (you can simple check the right name in the VM disk properties), you can consolidate it to a new file called VMname_2.vmdk using this command:
vmkfstools -i VMname_1-000001.vmdk -W vsan VMname_2.vmdk
At this point you can remove the old file (also from datastore browser) and from the VM settings. Run the snapshot consolidation (that will simple do nothing) and at this point you can attach the new disk!