Face it. Death is a part of life. This is true of people. It’s true for hard drives,too. Those containers of mass data on high speed spinning disks have a life span just like we do. So how long will a hard drive last? Great question!
MTBF
The MTBF is touted as a feature in the specifications of most hard drive. What is it? MTBF stands for Mean Time Between Failures. Mean means average. So the MTBF of a drive is the average number of hours it will work before it fails. The operating word here is average. Since it is an average, some drive will fail before the MTBF and some will last longer. In other words, you just don’t know when the drive will fail. You do know that it will fail.
JBOD
Most photographers I know use the JBOD method of storing their work. JBOD is the acronym for this. It means Just a Bunch Of Drives. This is a risky form of storage because unless there are copies of each of the drives, data will be lost when a drive fails. Some photographers and a lot of videographers use JBOD by copying work to disks then putting them on a shelf. According to one of my sources a hard drive stored this way has a 25% chance of never spinning up again after a year. At the very least store bare naked hard drives in static free bags with a packet of silica gel to protect against static damage while handling and the “drive killer” humidity.
Rationalization
A lot of photographers I have spoken with trust their data on individual hard drives and even have backups of them stored on a shelf somewhere. They tell themselves that their data is safe. This is a rational lie.
It isn’t safe.
There are lots of reasons why this isn’t a good strategy.
- The backups have to be updated every time a file is changed.
- No one knows how long a drive can sit idle before it freezes
- Drives sitting around can get lost
- One out of four drives stored without being spun up regularly will fail
- It’s too much work to keep up with the updates
RAID
Redundant Array of Individual (or Inexpensive) Drives offer a reasonably robust solution. There are some things you want, OK, need to know. There are several flavors of RAID. The most common are RAID 0, RAID 1 & RAID 5.
- RAID 0 stripes the data across multiple drives. It’s very fast. If one drive fails all of the data is lost.
- RAID 1 mirrors the data on one drive onto another. This is very safe. It’s also much slower and more expensive. It requires two drive to store the data of one drive.
- RAID 5 distribute the data across multiple drives using parity so that any one drive can fail. After the failed drive is replaced the remaining drives remap the data. Typically RAID 5 uses four identically sized drives to store the data that would be held on three drives.
BeyondRAID
Several years ago Data Robotics introduced Drobo. These RAID 5 like systems don’t need the drives to be the same size. Lower terabyte drives can be replaced with larger ones without data loss. The Drobo and its accompanying software Drobo Dashboard monitor the health of individual drives and warn when a failure is imminent or when a drive is approaching its capacity.
Drive mapping
Drobo’s BeyondRAID technology provides two critical improvements over RAID 5. First traditional RAIDs require all of the installed hard drives be the same capacity. Drobos can use multiple drive sizes. As a matter of fact when a Drobo is running short on space, one of its existing drives can be replaced with a larger one. Best of all, and the second improvement is that BeyondRAID only has to remap the data stored on the replaced drive to the new one saving lots of time. Traditional RAIDs must rebuild the entire volume. This can take days.
Everything ends…
While true of hard drives, our data doesn’t have to end. A good backup strategy that includes three complete copies of your data along with technology like RAID 5 and Drobo’s BeyondRAID will keep it safe for, well, a lifetime. Maybe even longer…
Kevin is a commercial photographer from Atlanta. He works for fashion, architectural, manufacturing and corporate clients. When he’s not shooting, he contributes to Photoshop User magazine & writes for Photofocus.com.
http://kevinamesphotography.com
https://facebook.com/KevinAmesPhotography
Local backups, such as JBOD, Raid or DROBO have the advantage of allowing you recover from disaster more quickly than alternatives (though that JBOD solution would be maddening to restore from). Making them the ideal place to restore data from when there is a problem. However, the backup in these cases are pretty much reside in the same place as the original copy meaning the same thing that may have killed the primary copy of your data could easily destroy this backup too. The further apart the backups exists from the original, the likelihood that a single event will be… Read more »
Here’s what I have done, and what I now do. I have 2x 2TB drives and a 1TB drive, and a 128GB SSD in my machine. So, the SSD is my system drive. I have Steam, and some applications that won’t fit on the SSD on the 1TB. I had been putting my lightroom catalog on one of the 2TB drives; and then back up to the second disk. I also run crash plan to do a cloud backup. I started running out of space (down to about 100GB free), so what I have done now is just add the… Read more »
Even if backups are stored on another media other than disks, ex. DVD, those backups can be lost in disaster, such as fire or flooding, or theft. Off site storage is a way to minimize a single site disaster.
We’ve covered that in other articles… this is simply pointing out a fact that many ignore… hard drives fail.
I understand that you have some sort of sponsoring/collaboration with Drobo, but this sounds a lot like FUD. If you don’t spin up drives even once in a year, well then you aren’t doing your backups properly right? A RAID system is not the answer to a non-working backup plan. Drives will fail, and when they do in a RAID setup things can get even worse than in a JBOD workflow. RAID is good for speed and data access, like a server. (Our Drobo at the office fails on a regular basis, btw) A good backup plan (local and off-site… Read more »
Erik…. we encounter people all the time who don’t understand that drives fail.
This is the point of this article.
We have others about 3-21 backup including offsite.
we’re advocating for having data in 2 places locally (if one of those can experience drive failure without data loss great).
I’ve had drives fail in my Drobo… all drives fail.