1

A theoretical question (I do not have spare hardware to test for this).

Scenarios

If RAID 0 is used (no redundancy of data) with HFS Plus, and if a disk within the array is disconnected for a while whilst the file system is mounted by OS X, then I assume that:

  • the OS will at least present a red alert for an improper eject
  • worse behaviours may occur.

I'm not particularly interested in the behaviours of the Mac around the time of disconnection. Most interested in the aftermath …


Logically, I guess: switch off all hardware, replace the missing part of the RAID 0 array (reintroduce the missing disk) then restart the Mac.

The question

How likely is it that the end user – the person who found or made the disconnection of the disk – might recover data in the following situations?

  1. Where the RAID 0 is software-controlled (for example, created by Apple Disk Utility)
  2. where the RAID 0 is hardware-controlled.

I imagine that the end user would find recovery difficult or impossible; that the RAID-specific expert services of a company might be required.


Whilst the question is theoretical, I will greatly appreciate any experience-based answers.

1
  • 1
    You can test this quite easily with inexpensive USB drives connected to a hub. The mechanics and reliability of the filesystem can be exercised with very low cost hardware. Apple does a better job than most in making Disk Utility understandable to end users on how to recover data.
    – bmike
    Jan 22, 2014 at 16:06

1 Answer 1

3

In a RAID 0 set established by Disk Utility you, of course, have zero redundancy. The moment one of the drives is lost from the set the entire set becomes useless and you have a volume failure. If (and only if) no attempt was made by anything to write to the volume while the physical drive wasn't there you might be lucky and it will be OK but otherwise you will have a problem.

At that point any recovery is going to be difficult. Any attached Mac will just see the volume disappear. I've seen this happen a number of times and getting the data off was only ever partially successful (and a small part at that) and extremely expensive. My advice to a couple of customers who've had it happen is just give it up as a joke and go to the backups. Chances of good data recovery are so close to zero as makes no difference.

If you are extremely lucky and there was no attempt to write to the volume and the heavens smile upon you then it is conceivable that a volume repair might work. I've actually seen it happen - once.

For a hardware RAID 0 it's pretty much the same.

The only place I like to use RAID 0 is in a separate hardware drive array such as the Promise boxes for a user who really needs the speed for something like video editing.

If you really need the speed of RAID 0 then make sure you have really good backup policies.

1
  • 1
    This is solid, solid advice. Any time I'm tempted to deploy RAID, I have found I need to really commit to backup and testing a restore before putting the RAID into production. Not only is RAID not a backup, but making clear stakeholders understand how long it takes to restore 10 TB of data is critical.
    – bmike
    Jan 22, 2014 at 16:04

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .