Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

My company is buying a couple servers. I don't have a sysadmin background and neither does anyone at my company but I'm the "computer guy" so it falls to me to spec them out.

We were given the option of putting RAID on our servers but I declined since we already have RAID backups going at our office. My boss didn't understand why I declined the RAID and, honestly, neither do I.

My thinking is that if a server's hard drive fails, we can restore it from backup (to a different, clean server) and it wouldn't be a huge deal (we'd practice these recoveries, of course).

My boss's thinking (as I understand it) is that having RAID on our servers would mean no downtime (or at least significantly less downtime) because if the primary disk fails, the server can just switch to one of the other disks without missing a beat.

It's important to note that this server will be very low-traffic. In fact, it won't even be connected to the internet (except possibly to grab OS updates). It will mostly just be running automated jobs.

Since neither myself not my boss know a whole lot about this "RAID-on-a-server" idea, can someone shed some light on this for us? And to clarify, this isn't a me-vs-my-boss situation. We're just trying to find the right solution together.

share|improve this question
    
Never count lower than 2. You need >2 disks, >2 NICs >2 Routers, Firewalls, Switches, Power supplies. You need at least 2 of everything in your server/datacentre to give you enough redundancy to sleep at night. Or at least that's how I feel about it. I'd never spec a server without some level of RAID. Hell, my workstation has a pair of 1TB disks in RAID 1. –  Tom O'Connor Oct 15 '10 at 16:56

3 Answers 3

up vote 6 down vote accepted

RAID is not backup, it's redundancy. Not installing atleast RAID1 into production servers is something I would strongly recommend against.

Have you tried restoing a server from backup? Do you test it regulary? If so; then you know how long time it actually takes to get a server back up on it's feet in working condition.

Let's say it takes 8 hours to bring a server back up on it's feet from tape backup. Do you have a spare disk lying around? If not: add 20 hours. Can your company produce the same amount of $$ while missing the server for 28 hours?

RAID is a very cheap option to have some resilliency against disk failures, you should really implement this.

share|improve this answer
    
RAID is about availability. If a drive dies, you don't lose access to the server until you get a new hard disk. –  Bart Silverstrim Oct 15 '10 at 16:28
    
But the OP is right in that he should have the backup for restore. Oh, and avoid RAID 5 now. Go RAID 10 or mirroring (RAID 1) if it's a low-usage server. Prefer RAID 10 though. And hardware RAID card, preferably with hardware to show blinkies and status of failed disks and allow hot-swapping drives. –  Bart Silverstrim Oct 15 '10 at 16:29
    
@Bart, how do you feel about RAID 6 as an alternative to 5? –  Tom O'Connor Oct 15 '10 at 16:54
1  
@Tom: Less common, can have overhead on write operations so it depends on what the server is used for, but it can survive two drive failures. From Wiki: RAID 6 provides protection against data loss during an array rebuild, when a second drive is lost, a bad block read is encountered, or when a human operator accidentally removes and replaces the wrong disk drive when attempting to replace a failed drive. - Therefore, I wouldn't mind it. Basically any RAID that can survive two drive failures is better in my book. –  Bart Silverstrim Oct 16 '10 at 15:48

If you set up your server's disks to be mirrored (RAID 1) then you'll be all but guaranteed to never need downtime on the server due to a disk failure.

Personally, I'd say it's a good idea to have a mirror even on low-traffic servers just for the safety net it'll give you against hardware failures. It only costs one extra drive (plus the knowledge and time to configure the RAID) and saves you the time it would take to build out a new server from your backups.

share|improve this answer

Adding RAID to a server won't add that much to the cost of the hardware, shouldn't add anything significant to the cost of setup & maintenance, and has the potential to prevent downtime while you reinstall, restore backups, etc.

If you're basing it on cost, what you're really comparing is the cost of the extra hardware vs. the cost of your time for doing recovery when the drive fails. If it's a low-priority server doing automated processing, it's likely going to be in service until it dies - why not extend that time up front?

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.