823.418 of 64tb thin provisioned vmdks on a 8TB datastore - is that brave or nuts ?

[continuum]

While running some tests with recovery scripts I created a couple of 64TB thin provisioned vmdks on a 8TB datastore.
The amount of 64tb vmdk I could create without any warnings was a bit scary - in words- eight-hundred-thousand vmdks ....
The  hidden .sbc.sf file that is required to do this is massive: 320 GB.

To sum it up - VMFS 6 and ESXi 7 have no self respecting limits for the amount of vmdks that it allows to create.
The webinterface still tells me that from the 8TB of free space only 4 TBs are used.

Needless to mention that with a sbc.sf files of that size no third party tool is able to read this VMFS 6 volume.
Datastorebrowser itself crashes most of the times ....

Lesson learned: if you assume that ESXi warns you when you run datastores without mercy you are wrong.
Be careful yourself !!!

Ulli

[continuum]

Experiment finished ...
Created more and more files - not with a size of 64tb but just small textfiles.
ESXi expanded the metafiles very bravely - but finally after allowing  2 million files it gave up.
All data on the volume is lost - even UFSexplorer can no longer read the volume.

Most useful usecase ? - sabotage.
Create a script that writes files into a hidden directory and the VMFS volume will autodestruct sooner or later ...

Suggested solution: fix (remove) the expansion of metafiles on the fly. The procedure used by ESXi 7 is simply crazy ....

Ulli