| 7:34 pm on Nov 21, 2005 (gmt 0)|
RAID will get you drive failure backup, so if one drive goes down, it should keep working and you lose no data.
If you only have one drive, if it dies, you are SOL.
Always go with more.
| 8:26 pm on Nov 21, 2005 (gmt 0)|
i was thinking raid 1 so no backup.
| 2:58 am on Nov 22, 2005 (gmt 0)|
Don't confuse RAID mirroring with a backup. The RAID 1 setup gives you redundancy. System backups are a different thing altogether.
| 8:05 pm on Nov 22, 2005 (gmt 0)|
I've also read that raid 0 for redundancy can sometimes be flaky, depending on what happens (i.e. you end up with two corrupted hard drives). I'm running that right now but next time I do a server install I'm going to forget it and stick to striping for speed. For backup/redundancy I'll stick to copying all my data to another hard drive on the same machine a few times a day. Not quite complete redundancy, but again I've read that raid 1 isn't as ideal when it comes to crunch time as we'd like to believe (I stand to be corrected, I've never had a drive fail on my servers).
| 9:59 am on Dec 5, 2005 (gmt 0)|
Also dont confuse redundancy and failover. Your file copy gives no realtime failover.
|stand to be corrected, I've never had a drive fail on my servers |
Me neither, but risk mananegemnt is about Probability and Impact. So Its a low probability (higer on drives 2+ years old) but a very high Impact if it happens.
| 8:49 pm on Dec 5, 2005 (gmt 0)|
I agree. I've taken the position that I *know* if a drive fails I'm going to lose a few hours. But I also know I can recover from something like that fast. (i.e. I've accepted the risk that if a HD fails the worst case scenario I lose a couple of hours). I know I can do a bar metal reinstall in that amount of time. I don't need 100% gtd. uptime though I strive for it. My data isn't that important that a few hour old backup would be quite fine.(course I also went with scsi drives as they're reputed to be more reliable than most of the alternatives. And I've gone with Dell instead of a whitebox for the same reasons.).
If a power supply blows that would take longer (no extra power supplies kicking around) so I've gone redundant on that.
If losing a couple of hours of data or that amount of downtime is really going to affect your business (and in many business, I don't think that's really the case) then I'd probably be looking at something like a seperate failover server.
In short, look at every piece of hardware and ask yourself - if it dies, how soon can I get it replaced? what's the likelhood of it failing? and how much am I willing to pay to lower that time? For me, if the entire server disappeared, I'd be back online likely the same day.
I can speak from experience when I say the greater danger is plain old user error deleting files :).
| 8:37 am on Dec 6, 2005 (gmt 0)|
One thing I learned recently - dont rely on Windows full system backup (.bkf) for a bare metal recovery - unless the hardware is identical it dont work!
| 4:50 am on Dec 7, 2005 (gmt 0)|
They go hand in hand, if the site is Ecom and you are down, then you arent making any money. Ask youself, how much will I lose if Im down for 1 day? Then how much does an extra Harddrive cost for the Raid 1 setup, typically you'll find that the extra drive cost is minimal compared to a days earnings.
I always use Raid 1, and backup offsite aswell. Redundantcy, Redundantcy, Redundantcy. I live by that and survive because I do. Raid 1 is a great idea, and wouldnt think twice about it, always do it.
As far as Raid 0 goes for the speed, I reboot my Server, and start clicking away and compared to being up for like 2 months the difference is minimal. If you look in your Images folder and see the size, its what maybe 100mb? That can easilly be cached in the memory, same for quiries etc. On a static site I wouldnt see any noticable speed increase from Raid 0, ofcourse this all depends on file sizes and many many other factors.