How not to run a server
Some time ago, a man saw an opportunity where he could help his fellow netizens express themselves, and he started a blogging host called Diary-X. Many people flocked to his banner and posted their thoughts, feelings and ramblings there for the world to see. Even the mighty Diary of a Wolfeman (yukyuk) got it’s humble beginnings as a Diary-X blog.
It was not meant to last forever, though, as Diary-X has suffered a fatal drive crash, and even DriveSavers (arguably the best in the biz) were unable to recover the drive. Bacchus at Eros Blog went off on a rant a while back about how it is almost criminal to remove a blog from the ‘sphere, so truly it is a tragedy that this has come to pass not for one, but for hundreds.
While I hate to kick a man while he is down, there is one thing that keeps coming back to me: there is only one drive mentioned. The man was hosting a server on one hard drive alone. This, my friends, is what happens when you let an amateur loose in the server room.
Y’see, there’s this thing called RAID. I’m sure you’ve heard of it. Redundant Array of Independent (or Inexpensive) Disks. RAID units are used for both performance and fault-tolerance, but the latter is usually the driving factor. There is more than one way to do it, but the gist of it is you take three or more hard drives and you spread the data across all of them in such a manner that either the data itself or the parity of the data exists on more than one drive at any given time. Thus, if a disk fails, the data can be recovered from the remaining drives in the array. In the better systems, you simply pull the failed drive, insert the replacement, and the RAID automatically rebuilds itself.
The worst part about all this is the commonality of RAIDs these days – it’s difficult to find a new motherboard nowdays that doesn’t have a RAID controller built-in. Controller add-in cards are pretty cheap, and I just picked up a pair of 100GB SATA drives for a touch over $110. For 3 bills USD, the man could have implemented a RAID in any computer he wanted and avoided the whole situation. Now, his clients have to suffer the consequences of his lack of foresight.
Folks, it’s not “if” a drive will fail, it’s WHEN a drive will fail. If you use a computer, data is your life, and backup routines are your religious observance. Don’t forget to pray. Ever.
End Rant.
February 28th, 2006 at 9:01 am
I think it was a ‘got what he paid for’ deal… he was on a co-loc’ machine halfway across the country, if I understand things correctly, so it’s not like HE could plunk any hardware he wanted into the device.
As for RAID: What if the RAID controller dies? Often, you can’t get one controller to resurrect the array from a previous device (we’ve had this happen!), so RAID isn’t necessarily a catch-all.
It’s not the drive scheme, it’s the backups, ’cause ALL devices fail some time.