0

Hi, I had a catastrophic RAID failure and was hoping for some advice on next steps.

A couple of years ago, I purchased an Addonics mini-tower with a built-in hardware raid controller and filled it with hard drives and set it up as a RAID 1 array. I used this as my primary backup device for all of our home computers, and was quite happy for a while. It worked well and I thought I was pretty well protected against loss.

One day, the array failed to mount when I attached it to my laptop. I tried another computer and another, yet the array failed to mount.

When I called the manufacturer, they told me it was most likely either a bad eSATA cable or a bad eSATA to USB adapter. I replaced both, it didn't fix the problem.

Not knowing what the cause could be, I took one of the hard drives out, put it inside a drive enclosure and took a look. Most of the data was still on the drive, however, much of the data could not be opened or retrieved. I took out the other drives, and found the same problem with each of those.

When I called the manufacturer back, they told me that it might then be a problem with the disk controller -- that the disk controller could have failed, and corrupted the data on all the drives.

Egad! I had no idea that the RAID array I had set up thinking it would be so bulletproof has such an important single point of failure. Now, I'm in a worse position that I was before the array -- since some things I had put into the array for "safekeeping" was no longer on any computer or other backup.

Now I'm at a loss on what to do next. I'd like to still possess all of my backups, but I no longer trust that a RAID will keep my information safe. Is there a way I can set up a RAID with multiple disk controllers? I'm willing to put some money into a setup that will enable me to ensure that my information is safe, is there something safer than RAID?

Any advice is appreciated. Thanks!

2
Contributors
3
Replies
4
Views
7 Years
Discussion Span
Last Post by rubberman
1

RAID just protects from drive failure. Not controller failure, unless you also have redundant controllers! In any case, RAID-1 is just mirroring - no redundancy. You have two drives with the same data. If one failes, the other contains the data. However, some other types of problems can corrupt both drives. RAID-5 is better, since it incorporates striping and distributed parity blocks. IE, in an array of 4 drives, for example, any single drive can fail, and the others can be used to reconstruct it. It is more resilient to other sorts of failures. So, in your case, if you had 4 drives, you would have 2 + 2 drives mirrored. If you want complete mirroring with recovery possible, then I recommend RAID-10, which is mirroring with striping and distributed parity - sort of double redundancy.

Here is the wikipedia article on RAID: http://en.wikipedia.org/wiki/RAID

Votes + Comments
Well written and accurate.
0

RAID just protects from drive failure. Not controller failure, unless you also have redundant controllers! In any case, RAID-1 is just mirroring - no redundancy. You have two drives with the same data. If one failes, the other contains the data. However, some other types of problems can corrupt both drives. RAID-5 is better, since it incorporates striping and distributed parity blocks. IE, in an array of 4 drives, for example, any single drive can fail, and the others can be used to reconstruct it. It is more resilient to other sorts of failures. So, in your case, if you had 4 drives, you would have 2 + 2 drives mirrored. If you want complete mirroring with recovery possible, then I recommend RAID-10, which is mirroring with striping and distributed parity - sort of double redundancy.

Here is the wikipedia article on RAID: http://en.wikipedia.org/wiki/RAID

Thanks, I'll read more about RAID-10. Question though, if I set up RAID 10, could a faulty drive controller still potentially corrupt the data on all of the RAID 10 drives?

Is there a way to set up a raid with redundant drive controllers?

Thank you!

0

Sorry to take so long to get back to you on this... :-)

Not to delve too deeply into fault-tolerant computing architectures, to my mind, RAID is primarily intended to provide a higher degree of fault-resiliance than just a bunch of discs (JBOD), but it will not guarantee that all failure scenarios will leave your data intact. In you have JBOD, then when a disc fails, you lose everything on the disc. If you have a RAID-5 array, then the data is striped over all the discs, as is the parity data, so even if one disc fails, the array can continue to operate in "degraded mode" until the disc is replaced and the new disc is rebuilt from the data on the others. RAID-1, mirroring, works much the same way, except that the discs are copies, so the rebuild cycle is just copying the data from the good drive to the new one, but you can keep using the array in degraded mode until the new disc is synced with the other. Generally, either of these work well. However, something like a catastrophic controller failure can (depending upon the design of the controller) possibly corrupt the array, resulting in your data becoming inaccessible.

So, what to do? As many have found over the years, just having RAID arrays does NOT mean you can just stop making backups of your critical data. Single disc failure (or more discs, depending upon your hardware) == continued operation and online recovery. Multiple disc failure == SOL (Sorry, Out of Luck). Controller failure == SSOL (Sometimes Sorry Out of Luck).

Anyway, for a more robust solution, a RAID-10 array that is basically dual RAID-5 arrays that are mirrored and that each is using a separate controller, is much more reliable because you truly do not have a single point of failure in the data path, other than the computer itself. However, many times people configure RAID-10 arrays on a single controller, which is no more reliable to my mind than a single RAID-5 array, unless you are using dual enclosures so are protected from a power failure on one of the enclosures.

This topic has been dead for over six months. Start a new discussion instead.
Have something to contribute to this discussion? Please be thoughtful, detailed and courteous, and be sure to adhere to our posting rules.