These instructions build on those provided [here](http://docs.docker.io/installation/windows/) and [here](https://github.com/boot2docker/boot2docker).

Before doing the following, create a new virtual disk and attach it to your VM. The size of the virtual disk just depends on what you intend to use/run and how many containers you intend to run. The default, 8 GB, is probably more than enough and is reasonable to use if in doubt since the default mode dynamically allocates the drive as needed.

Boot the VM and run just

    fdisk -l
    
To see the disk device name(s) and their size. In this case it's the `/dev/sda` shown here that has 8 GB capacity:

    Disk /dev/sda: 8589 MB, 8589934592 bytes

Run the following to setup a partition on that disk:

    sudo fdisk /dev/sda
    
Use the `n` option to create a new partition, choose a `p` (primary) partition and go with partition number 1. Accept the defaults for the cylinder choices to fill the entire disk. 

Use `w` to write the partition changes and exit.

If you get an error about ioctl not able to re-read, then reboot the VM:

    sudo reboot
    
Format the filesystem on that partition being sure to assign the label boot2docker expects for the data volume:

    sudo mkfs.ext4 -L boot2docker-data /dev/sda1
    
Reboot to allow the new disk and mount point to go into affect:

    sudo reboot
    
To make sure it did what we expected, run

    df
    
and confirm there is are `/mnt/sda1` and `/mnt/sda1/var/lib/docker` mounts (but with appropriate disk and partition identifiers for your setup). If you really want to connect all the dots, do

    ls -ld /var/lib/docker