Comment 6 for bug 1456631

Another complication is that some cinder drivers appear to over allocate when the storage backend is only capable of allocating space in fixed intervals of > 1GiB.

For example,

After requesting a 100 GiB volume via `nova volume-create 100` and attaching it to my instance, `lsblk` now reports

```
[centos@clusterhq-flocker-buildslave ~]$ sudo lsblk --bytes --output SIZE /dev/vdb
        SIZE
111669149696

```

Which is exactly 104 GiB

```
In [12]: Byte(111669149696).to_GiB()
Out[12]: GiB(104.0)
```

Which makes me think that GB vs GiB is less important than I thought.
Telling people GB and actually creating GiB (or more) is just another form of over allocation, right?

One complication is that if I now fill this block device I might to be able to move the data to another 100 GiB cinder block device on another cinder platform....but that platform may have a different allocation interval and its 100 GiB volume may not have room.