Friday, April 24, 2009

The Idiot’s Guide to Restoring a Failed RAID 1 Hard Drive

In my last post I recommended you use a RAID 1 drive like those offered by Western Digital and LaCie as the primary storage for your photos.

Taking that simple step will offer a huge measure of protection to most photographers.

But buying the drive is only part of the battle. You have to know what to do when it fails, because some day a light on that drive is going to start blinking, warning you that one of the drives in the RAID has failed, and you are back to having your files on one drive. When that happens, you need to take immediate action to restore protection of your files.

Taking Action

If you read the manuals, most RAID 1 drives have options to replace the faulty disk with a new drive, then let the RAID automatically rebuild. This means you have to open up your RAID case, find and replace the faulty drive, and then have the RAID rebuild itself. If you select the wrong drive, or make one error, it could destroy your entire RAID structure, thus compromising all of your files.

This solution is technically possible, but I think for most users it’s a process fraught with huge risks. I believe you are more likely to mess something up than fix it. That’s because you’re stepping into a mine field, and the manual is not a good enough map to get you through. One misstep means your data goes kaboom! Even experienced IT professionals can make costly mistakes here.

Furthermore, if the “good” drive fails during rebuild, you will lose all of your files.

I think there is a far simpler solution that minimizes risks, almost guarantees success, and is still cost effective. I call it “The Idiot’s Guide to Restoring a Failed RAID 1 Hard Drive”

First, do a backup so you have a current copy of the good drive. If you have multiple backup sets, update as many of them as you can. It’s cheap insurance.


Take the simplest, safest approach:

1. Shut the drive down. This is the best way to prevent further drive failures.

2. Buy a new RAID 1 drive.

3. Copy all of your data from the old failed RAID to the new RAID 1.

The cost of a new RAID 1 drive is nothing compared to the cost of having a drive recovered by a professional data recovery service. Simply copying your data to a new RAID 1 drive is so simple that any digital photographer should be able to do it correctly 99.999% of the time.

If you can’t wait a 24 hours for a new RAID 1 drive to be delivered, then keep a spare on hand. A $220 drive waiting on your shelf is cheap insurance compared to the value of your files and the cost in time and money to restore them in a worst-case scenario.

That’s it! It’s really that simple. This is a solution that you, the average photographer can just comes down to making the right choices.

This leaves you with one question: What do I do with my old RAID drive? Stay tuned for that in a future post...

Labels: , , , ,

Monday, April 06, 2009

Systems for Automatic Duplication

So I’ve convinced you to store your files on a volume that has automatic duplication the question is, which one should you buy?

There are more solutions available than ever, and fortunately there are some great options for the home user that don’t require an IT genius to set up.

My recommendation is for a RAID 1 Mirroring setup.

I think RAID 1 offers the best protection from the most common hard drive failures. That’s because most RAID 1 products allow you to access your data after a single hard drive failure. That allows you to make an immediate backup of your files to ensure you have an up-to-date copy, and it gives you many options for restoring your redundant copy.

There are many vendors for RAID 1 products, but Western Digital and LaCie stand out as having solutions that are priced for the home user, and are easy to set up and maintain.

Western Digital offers the My Book Studio Edition II. LaCie offers several solutions, with the Hard Drive MAX being the most attractively priced.

Between the two, the Western Digital product looks like the more attractive choice to me.

My prime motivator is price...the Western Digital product is about $100 cheaper than similar capacity LaCie models, which adds up quick if you need a lot of storage.

But I also like the fact that Western Digital is a larger company that is likely selling more units...and I think a larger user base for a product is always a good thing. I have no way of knowing if that really will make it a better’s just a gut feeling. I haven’t used either product, so I’m just going by specs and company reputation.

Before you buy any storage solution, you should run it through my Storage Litmus Test:

Do you know how an successfully operate it successfully 100% of the time?

Read the manuals online before you buy and ask yourself if you know you can successfully operate it. If you don’t feel confident that you can, then don’t buy that solution.(And if you already have a solution, read the manuals now and see if it passes or fails the test.)

Bad Solutions and Non-Solutions

In our poll a few weeks back, several people said they are using RAID 0 Striping to protect their files. That’s bad news because RAID 0 provides less protection than a single hard drive. RAID 0 works by splitting up a file so half of it goes on one drive, and the other half goes to a second drive. If one hard drive fails, you lose all of your files. Even worse, if the data that tracks how the files are split up becomes corrupted, you lose access to all your files. RAID 0 is about speed, not protection. Don’t use it for protection!

RAID 5 is another solution some people are using. It’s been the darling of the IT industry for a long time, but I think it’s playing with fire.

About a year ago, a client had a RAID 5 fail. After about 4 weeks without access to their drives and considerable expense, they were able to recover all their data...and this was with a top-end system. They got lucky...they could have lost it all...and luck is not a protection strategy. Last I talked to them, they were investigating a RAID 1 system.

If you want to get techie about it, check out Why RAID 5 stops working in 2009.

The Drobo is also subject to the many of RAID 5’s shortcommings since it typically uses parity data to create a virtual duplicate. That means that if a second drive fails before a rebuild is complete, you can lose all of your data. I did one tweet recommending the Drobo, but I’ve since become jaundiced on it, and I’ve removed the tweet. I think RAID 1 is a much safer solution than the Drobo.

If you are using RAID 5 or a Drobo, I’d replace it with a RAID 1 immediately. Not having to rebuild a disk from parity data is a good thing.

What does Rich do?

What do I do? I use a custom RAID 1 solution that I’ve assembled to work on Mac servers. I use SoftRAID to make mirrored copies to two separate hard drives. SoftRAID doesn’t use proprietary formats, so both copies of a volume are readable on any computer...I don’t need a RAID card or enclosure to read them in event of a failure. I use Sonnet DP500 SATA enclosures with Sonnet SATA cards, and various brands of SATA hard drives. The servers run 24/7/365 and are hooked up to a battery backup to protect against short power outages. In this configuration, with dozens of hard drives, I have a drive or two fail every year, but I have not lost any data, and I have been able to keep accessing my files while the mirrored drives rebuilt. It is very fault tolerant, very simple, and has proven very reliable.

My solution does basically the same thing as the Western Digital and LaCie products, just on a larger scale.

Whatever solution you choose, remember that an automatic file duplication system like a RAID 1 is NOT a backup! It will not protect against catastrophic failures, i.e. when your hard drive is lost, stolen, or destroyed. It is meant to protect you as you are working, which is the time in between backups. It also makes it less likely you’ll need to use a backup to restore a hard drive, which can be a time consuming process (And you typically lose some files, since backups are almost always out-of-sync with your hard drive).

Buy one today, before your single hard drive fails.

Labels: , , , ,