Update readme to document rawdisk

[dalehamel]

@zchee readme update as requested for #169

[zchee]

@dalehamel Thanks! merged.

[alafon]

@zchee I won't open a new issue to ask a possible simple question but could you explain a bit more what using this feature implies ? It is said that there is a potential cost of data durability, what does that mean ? Can mounted volumes be affected/corrupted ? Or is it just about the containers inner data ?

Thanks in advance

[dalehamel]

@alafon since I added the support for this, I'll answer.

Basically, if you are running something where data integrity really matters (MySQL perhaps), then an unclean shutdown of the VM could result in some bytes being unwritten when using this driver.

Most likely though, since you're probably using this for development, this probably doesn't matter.

This is because the virtio-blk is much more "raw" than the ahci approach, which attempts to more closely mimic how an actual SCSI device would operate.

We use virtio-blk because it is faster and lower overhead, and we make sure everything is disposable anyways (this is kubernetes after all), so we don't really care if there are some edge cases where data could be corrupted.

[alafon]

Thanks for the answer, pretty clear :)