RAID is not a backup solution except when it is
I’m not sure what Joel Hruska‘s article is pointing out at ExtremeTech. I’m approaching this article from the perspective a home user or an individual who is not running any business systems or infrastructure. According to Joel:
RAID is not a backup. RAID is not a backup, ever. Not RAID 1, not RAID 5, not RAID 6. The “R” in RAID refers to redundant, not “backup.” While it’s true that a basic RAID 1 will protect against the sudden failure of a single disk, it’s no protection against simultaneous failure, external disaster (flood, fire, electrical surge), or incremental sector failure that results in a corrupted write. If a malicious user accesses and wipes critical data on Disk 0, Disk 1 happily follows suit.
I completely agree with a slightly modified premise: RAID is not the final backup solution. RAID is however, a step in the right direction. I would argue that most people (non computer savvy people) probably backup the old-fashioned way using a CD/DVD disk (aka ‘the augmented floppy disk’). Others may use cloud services like DropBox, SkyDrive or even Google Drive. I know some people who use a USB external hard drive and call that effectively a “backup”. In the grand scheme of things, RAID (not RAID level 0) is a valuable form of redundancy because in reality, some form of redundancy is better than no redundancy at all.
Saying RAID isn’t backup isn’t a new thought. I think Joel’s article is a bit out of touch with reality in the sense that most people cannot afford the money and time to learn and implement a perfect backup solution. His three main points seem to highlight this blind spot:
Cloud services aren’t automatically safe
Ok, fine. So the cloud isn’t perfect. At least my data is somewhere else. If my house burns down at least I’ll have all my data somewhere else. It may not be 100% foolproof, but 99.99999% is good enough for most people. Nothing in life is automatically guaranteed safe or secure. How are cloud services any different?
Backup isn’t one-size-fits-all
Everyone has different needs. People who work with photos and photo editing or animation/rendering have large file sizes to contend with where as writers probably have a multitude of small documents which pale in terms of file size to that of photos. So naturally one-size-fits-all may or may not be right for certain people. File size, write/read speed and Internet bandwidth all play as major factors in any backup solution. This idea is just pointless.
So, backing up has to be a royal pain in the ass? If it isn’t relatively easy to backup data, what makes you think that anyone can do something that requires constant discipline without interruption? Furthermore, why wouldn’t you want to make something convenient? I would think if something was simple or required little to no intervention from me that I’d be more likely to do that rather than take a more hassle-prone approach. If you can’t bake convenience into a solution, you’re doing it wrong.
Joel briefly explains what the best solution is in two different sentences:
RAID arrays (other than RAID 0) should be treated as a way to ensure access to a backup set is minimally disrupted in the event of failure, not as a backup solution in and of itself.
Blending on-and-offsite solutions is the answer; we’ll explore the various ways to do that in future articles.
A hybird/heterogenous solution involving on-and-offsite technologies – I wholeheartedly agree with this idea. But we spent 95% reading how RAID was not the answer but apparently a blended solution is the answer. Doesn’t this negate the title/premise/thesis of the article? Shouldn’t the article be titled: RAID – the stepping stone to a solid backup infrastructure (or something to that sentiment).
I’ve been working on my own backup solution over the past few years. It’s not 100% up and running, but in the meantime it is working nonetheless. My solution is relatively complex, but once finished, it should require little to no attention or maintenance, accommodates all of my data needs from files, photos, videos, TV recordings and music, and all the while utilizing cloud services. Apparently I’ve broken every tenant he mentioned all the while doing exactly what he preached. I gather that Joel’s intentions, while noble, have unintentionally misplaced the theme and voice in his article. Maybe I’ll chip in my own two cents with a posting of my own solution, but in the meantime I’ll wait for these future articles.