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.

Earlier this week, one drive in our office's 2-year-old RAID5 array (4x 750GB Seagate drives, HP ProLiant 110 with Smart Array controller, running Windows Server 2003) failed. From what I know, RAID5 should be able to handle a single drive failure, but evidently critical OS files were stored on the failed drive (we kept getting No Operating System found errors) so we had to re-install the server OS and send the other drives to a data recovery center. We had lots of critical research data on those drives, and I want to know how we can re-setup this server to avoid problems like this in the future. And yes, we've been appraised about the importance of off-site backups. Please don't rub salt in our wounds! :)

Here's what I was thinking as a setup:

Visual of RAID setup

Can someone tell me...

  1. Is the configuration I'm suggesting even possible? I don't know much about servers. I'd like to go with RAID10 because our data is really important and performance isn't as much as an issue.
  2. Is there a way to save on space? It seems ridiculous having two 250GB partitions for the OS, which will only be used for the OS and nothing else, but I don't know what else to do since the RAID10 drives would have to be the same size.

Let me know if I can provide any more details about the setup.

share|improve this question
    
Do you know the model of the RAID controller in the system? Also, which generation ML110 is this? G6? G7? –  ewwhite Aug 24 '12 at 18:30
1  
It sounds like you really need to read the RAID levels question before proceeding... –  voretaq7 Aug 24 '12 at 18:41
4  
A RAID 5 can handle a single drive failure. If your OS wouldn't boot after a drive failure that means that you either had two simultaneous failures, or your array was already degraded when you had your most recent one. "but evidently critical OS files were stored on the failed drive" - That's not how a RAID5 works. Files and their parity data are split evenly across all disks in the array. No disks hold a complete copy of any file. In the event of a single disk failure, the missing data can be calculated from the data remaining on the unfailed disks. –  MDMarra Aug 24 '12 at 19:53
add comment

3 Answers

Well, the good thing is that the HP Smart Array controllers allow you to divide a physical array (a collection of disks) into one or more logical drives (virtual disks of a specific RAID level).

You can take four drives and create an array. You could then create a logical drive of say, 72GB, for the Operating system and define it as a RAID 1+0. You could then create another RAID 1+0 logical drive for the data. You could also create a logical drive of RAID 5 or RAID 0 (don't, but it's possible).

At the operating system level, the aforementioned logical drives will show up as separate drive letters.

This functionality is available with all/most HP Smart Array RAID controllers. However, in order to create this configuration, you'll need to boot the server with the Easy Setup CD or download the HP SmartStart CD.

share|improve this answer
add comment

Ditto ewwhite's comments. You say that your data is critical. First step, dig out the HP Quickspecs for your ProLiant and determine which RAID adapters it can support. In your situation, I'd be wanting to have a controller with battery backed cache, or even better, flash memory (not sure if any controllers for your server will have flash memory). Next requirement on the list, is see if your controller, or a controller that your server supports, will provide HP Advanced Data Guard (ADG) or RAID6. This allows two drives to fail. Next, install a hot spare. In fact, if your data has a significant £££ value, install two hot-spares. Finally, and most importantly, make sure that you are monitoring your server, either through HP Systems Insight Manager, a third party SNMP management platform, or good-old Windows WMI (you'll need the ProLiant WMI providers installing for this last one). As for the disk layout, create one big array, and carve out RAID 6 logical drives. Simples.

Oh, and take off-site backups ;-)

share|improve this answer
add comment

The proposal is possible, and decent. However, I'd like to suggest something slightly different, which may or not appeal to you.

First of all, let me say that it's good you guys learned your lesson about backups. Off-site, good. Any at all: absolutely necessary. However, the other thing you guys don't seem to be considering is that SATA is not really enterprise-grade, and the thought of storing "critical research data" on a SATA array just feels all kinds of wrong to me. Granted, the risk of catastrophic data loss is largely mitigated with proper backups (not completely, even backups can fail or get corrupted), but if I were responsible for this data, I'd absolutely insist that it be put on what I consider and enterprise-grade server on SCSI drives.

All the servers I manage that I have any control over use at least 6 SCSI drives. 2 OS drives in mirror RAID (small capacity, but higher speed drives - such as a 146GB, 15000 RPM) and a data RAID array (or arrays) in RAID 10, usually consisting of 4 higher capacity, but slower drives.

It seems to me that you have an opportunity here to get the buy-in to do this right and get the budget for a proper, enterprise-grade server, rather than putting new drives into the same old thing that failed you last time, so I'd definitely seize the opportunity while it lasts.

share|improve this answer
add comment

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.