Data Recovery Panic
Unfortunately, many of us don't have our Come-to-Jesus™ moment until we're desperate to rescue a failing hard drive, and that's when we realize we should have been doing something smarter all along. We run recovery software and cross our fingers, we try ridiculous methods to coax our data back to life, or we pay exorbitant fees with absolutely no guarantee of success.
I have come to terms with this: All of my hard drives are going to die. So am I for that matter, but one thing at a time. Many of my students admit that they've had a drive fail in the past, and that they're aware that their current drive will fail, but few have a plan in place to prevent data loss. Why?
I think it has to do with details and permission.
We get bogged down in the wrong details, and we don't give ourselves absolute authority over our own data. We each need to own this statement — “My data is my problem,” — and step up to the plate.
What Drive Should I Buy?
This is where many people get hung up. Ultimately it's a necessary step, but also one of the least interesting details. Let tackle it.
Interface: USB 3.0 or Thunderbolt?
These are the only interfaces I would consider (besides a gigabit NAS). Most of my drives are USB 3.0, because although Thunderbolt is technically faster, the speed difference is not significant for my use, and USB 3.0 is more widely supported and ultimately cheaper.
Brand: Western Digital, Seagate or Hitachi?
I own many Seagate drives, and several Western Digital drives. After reading this report by Backblaze, I will no longer sneer at Hitachi drives, and they now may become my brand of choice.
Hard Drive or SSD?
SSDs are wicked fast compared to traditional hard drives, but their higher cost makes them untenable for large scale backups. Stick to the spinning disks for your backup drives for now.
As a side note, if you have an SSD in your computer, make extra sure you're backing it up regularly. When an SSD dies, it dies quickly and hard. Many times I've been able to hear when a spinning hard drive is starting to go, but apparently this is not the case with an SSD.
This is the main hangup, because it generally keeps us from doing what we know we should do. Most of us think we're being smart if we spend as little money as possible to solve a given problem. This ain't GEICO. If I lost my wedding photos, how much money would I spend to get them back? How about the photos and videos from the first 5 years of my children's lives? Stop worrying about +/– $50. Get the drive that makes sense.
2, 3 or 4TB?
I put this item after price, because although it affects price, it also potentially creates another problem. Like the observer effect in quantum physics, by trying to address the problem of hard drive space, I can create another problem for myself — Whatever size drive I buy, I must be able to back it up.
I don't like the idea of spending $100–200 on a new, speedy hard drive and then using it simply for backup. If it's bigger/faster than my other drives, I'm tempted to make it my primary drive, and then relegate my older, smaller, slower drive to be the backup. That can work, until it doesn't. By definition, a smaller drive cannot (indefinitely) back up a larger drive.
If I can only afford a single drive, and I stretch my budget a bit and get that 4TB model, I know I'll use the space, and I'm getting the best bang for my buck. But what happens when I realize I've almost filled that 4TB drive, and I don't have it backed up?
Buy 2, Not 1
I shouldn't buy a drive that's too big for me to back up. If I cannot afford two 4TB drives, then I should just get two 3TB drives, or two 2TB drives, or whatever I can afford two of.
What should I do with my older, slower, smaller drive? I could use it for a system backup (a la Time Machine. Otherwise I'll format it, zero its data, then recycle or sell/give it away. I try to eliminate unnecessary (undersized) drives from the backup equation.
Manual Copying ≠ Backup
I can't rely on myself to manually run a program or script, or click some button or (heaven forbid) manually copy files in order to back up my data. If it doesn't happen on a regular schedule, automatically without my thinking about it, it's not backup.
Apple provides a backup solution in Time Machine, but it took me a while to understand how far I could trust (throw) it. I only use Time Machine to back up my system drive, not my media.
I have a small (320GB), bus-powered USB drive plugged into a USB 3.0 hub, and the hub gets plugged into my laptop every time I come home. Time Machine sees the drive, backs itself up to it regularly, and that's the extent of my reliance upon Time Machine.
Buy 3, Not 2
So, my hard drive crashes. But, I've got a backup. (Yay me!) I've got one backup. (Yay.) I've now got just one copy of my data. (Yay?) Which means I now have no backup. (Hmm.)
So I go out and buy a new hard drive to replace the now-dead model, and I begin copying everything over to the new drive. My old backup drive is now working its butt off to copy every file to the new backup drive, and in the process, it dies. And now I have nothing.
If my data does not exist in three places, it's not backed up. I need more than one backup.
Speaking of which, what happens if I've done my due diligence and have 3 copies of everything in my house, and then my house burns down? Or there's a flood, tornado, electrical freakout, or theft? All my beautiful, shiny eggs were in one big, vulnerable basket.
Bank of Mom
Backup #3 could live at my parents' house. That's good for a short period of time, but what happens 6 or 12 months out? I should buy another drive, make a new backup (#4), and swap it out every time I visit.
Or I could get a safety deposit box and leave a drive there. Or maybe leave a drive in my desk at work. Any of these options is viable, but it's work. If this was my plan, I would need to set a calendar/alarm and just do it.
Let's be realistic though — Backing up to the cloud is sloooooooow. I've been using Backblaze for about 18 months and I only have about 3.5TB uploaded. However, I prioritized so that my most important data got uploaded first, and I've gotten great peace of mind knowing that my most important data is in really good hands, away from my house.
Lots of folks would probably smirk at using optical media (DVD or Blu-Ray) to back up data nowadays. Still, stored in my basement I have probably a dozen spindles holding 100 CDs or DVDs each, and I've had pretty good luck with them being readable for many years. Several times, I've had clients contact me and sheepishly ask if I had a project that we worked on 5+ years ago. Sure enough, after digging through my stacks, I've found the projects.
I still have CD-R backups of all of my undergrad work (1995-99). I do not have (nor would I want) any hard drive from this time period.
RAID ≠ Backup
A RAID enclosure uses multiple drives that act as a single drive. Using RAID 1 (mirroring), a drive is automatically backed up by another drive without any user intervention. This sounds great, until the RAID enclosure fails to do its job, and all of the drives are unreadable. RAID can still be a good idea, but I now think of it as only 1/2 of a solution, not a full backup.
Although they're not technically RAID, several years ago I bought a 1st and 2nd gen Drobo. They were incredibly slow, but I'm sure their current crop of products are much faster. Unfortunately, I trusted a Drobo to be my only backup for my Midwesternal series of Photoshop documents, and it ate them. Lesson learned.
A NAS (network attached storage) is like a RAID that can be accessed over a network. Modern NAS units can be grown over time by replacing smaller drives with larger drives as they become available and cheaper. I've enjoyed this capability, until I realized that I had created a device that I wasn't really able to back up. Beware Frankenstein's monster.
About 6 months ago I purchased a Synology 412+, and I'm more than satisfied with its speed and capabilities. It has the ability to back itself up with multiple, directly attached USB 3.0 drives.
Some day, I will have to put my backup strategy to the test and do at least a partial restore. I still dread that day, but I hope that I will have tested my system enough to be confident that my paranoia will have paid off.