On 22.07.2015 21:58, Scott Dowdle wrote:
"ext4 over ploop over ext4" wasting disk space as overhead.
That is the case for all disk-file-as-disk-image containers and not unique to ploop. You said if you can't use OpenVZ and ZFS together (in the future maybe) then you'd switch to KVM... at which point you'd probably being using qcow2. I'm just saying. :)
Situation is different. If I create block device on ZFS: # zfs create -s -b 4K -V 128G tank/disk This empty 128GiB disk will use only 64K on zfs pool: # zfs list NAME USED AVAIL REFER MOUNTPOINT tank 2.91G 3.53T 96K /tank tank/disk 64K 3.53T 64K - And later, if I mount this disk inside VM using XFS filesystem with "discard,noatime" options - unused blocks inside VM will be "discarded", and will no use storage space in the zfs pool at all. For example, if I write 4.5GiB ISO image to filesystem inside VM, and later delete this file inside VM - on the ZFS side this zvol firstly grow in size of used blocks (+4.5GiB) and later after file deletion - USED space decrease in size to previous values (-4.5GiB) so only used blocks are marked as really used, overhead is near 0%. And free space inside ZFS pool can be used by any dataset. In case of ploop - free space inside ploop can't be used any more, and from hardware node point of view - this disk space is wasted, even in case when it unused and marked as free inside ploop image.
Thanks for your considerate conversation / discussion. Much appreciated.
You are welcome! -- Best regards, Gena _______________________________________________ Users mailing list Users@openvz.org https://lists.openvz.org/mailman/listinfo/users