I/O limits for disk groups
Since QEMU 2.4 >= there is a possiblity to group disks together. Instead of setting per DISK limits a VM can have a total amount of DISK IOPS it can use. Grouping disks together in a VM would make serveral use cases possible. Instead of restricting certain disks, a user can decide where the IOPS should go. It would also close the loophole of users creating many small disks, each with a (default) IOPS limit, and bundling those together to a big DISK (LVM / RAID).
relates to issue #5509