libvirt users - Dec 2021 - qemu-img create with extended_l2=on, followed by virt-install

Hi!


I have read the following blog post:
https://blogs.igalia.com/berto/2020/12/03/subcluster-allocation-for-qcow2-images/
and watched the video displayed there.

1) To make qcow2 as fast as possible, does it make sense to do the following:

[#] qemu-img create -f qcow2 -o extended_l2=on,cluster_size=128k my_image.qcow2
1000GB

And then:
[#] virt-install ?name=MyVM --disk path=my_image.qcow2,size=1000 . . .

Will this overwrite the qcow2 file already created or will it keep its
characteristics?

Does it make sense?


2) I wont?t be moving the VM around much and have the space I need, so I wonder
if it will be faster to do the following:
[#] qemu-img create -o preallocation=full -f qcow2 my_image.qcow2 1000GB

And then:
[#] ] virt-install ?name=MyVM --disk
path=my_image.qcow2,size=1000,bus=scsi,sparse=false . . .
(Running bus=scsi so I can trim the image ).

I could use a dedicated disk for the VM instead of using qcow2, but I like the
flexibility of qcow2.

Thanks a lot for any information about this! :-)

Best regards,
Richard Taubo