RAID and Disaster Recovery (RAID1/RAID10/RAID5)

[bingboo]

Hi. I am trying to come up with a strategy for disaster recovery in an environment that I did not configure and I am posting in this forum because we have 3 servers and they are all RAID. My experience with RAID disaster recovery is fairly limited so I apologize for the newb questions.

The first Server (s1) using software RAID (2000 OS) --- has 4 drives. 2 for the OS that are RAID1 and 2 for user folders, also RAID1… From my understanding, if I have a drive failure, I just break the mirror swap the drive and reestablish the mirror… Hot Swap so I can do this while live. Which brings me another question, if I were to keep a disk offsite and rotate with one of the drives to work as a backup of the OS, I am assuming that I will be able to specify from which drive to mirror to the other….And I must always break the mirror before ripping a drive out yes? Good way to keep a secure copy of OS?

The second server (hardware RAID) has 4 drives at RAID 10 (both OS and data) which from my understanding is like RAID1 on crack…….. is it pretty much the same scenario as above or am I misunderstanding something? Do I have to break the mirror or can I just walk up and pull a drive out? Again I would like to keep a disk offsite that has a complete backup of OS (and in this case data because it is on the same drive).

Third server is hardware RAID 5 across 3 drives…. Because the data is spread across all drives I was thinking of using something like Acronis True Image 8

http://www.acronis.com/

to clone it to a solitary drive in order to keep a backup of OS.

All data is backed up on tape so my big priority is to not have to rebuild server OS/environment from scratch in the event of hardware/software/intrusion…. Sorry for such a long post and I very much appreciate any direction that is offered.

 

[technome]

Raid 10, with enough drives is raid 1 on crack... very expensive, fast and addictive, just like dope.


I have setup OS based raids of all kinds on practice machines but I highly recommend a hardware based solution. Os raid forms are slower and are susceptible to OS corruption. Good hardware raids are extremely reliable, and for added safety have hotspare drive implementation ability. I only use OS based raid1, duplexed mirrors, involving two drives for very small networks. I will not use the OS for any other raid types.

Recovery.....
Raid 1
If you created a mirror, sync it, copy the startup files in the root of the primary over to the secondary manually, break the mirror, remove the secondary drive as a backup, replace it with a new drive, bring the server up and create a mirror, you would have a backup with the removed secondary drive. You would have to place the backup (secondary) drive in the primary drives position to boot. No hot swap can occur, Os can not do it. With hardware raid, many adapters support a hotspare, not the same as a hot swap. With an SCA drive chassis, you could be quickly back in business, as a drives switch is so easy.

Raid 10
With this config you can not just go over and rip drives out, you need to know which drives are corresponding to which mirror, you would have to break the mirror before removing the drives. Same with hardware raid, only after breaking the mirror, you could probably remove the drives, but the SCA enclosure would probably go into an alarm state, balking at the drive removal, again the correct drives would need to be removed or your array will be useless. Not a good idea, one mistake and it is all over.

Raid 5..
I rely on tape for raid 5 disasters. From all the post I have read, some state they image all the time, and are successful. Then, there are countless people, who image and cry when the image does not work. Unless you can test an image restore, I would not trust it.

A little advice, mark all drives and cable as to there correct position in drive chassis and as to cable connections. NEVER move a drive in an array from one scsi channel to another, almost alway results in array failure upon startup. Not a good idea to move drives around to different slots on the same scsi channel, some adapters handle it, others do not.


If you have a workgroup server, a complete restore can be done in a couple of hours with a fast tape unit involved. Active directory is more complicated but is fairly quick if you have had practice on a lab machine.

 

[bingboo]

I appreciate your response --- unfortunately I just was able to take a look at it and I have to go into a meeting. Would it be possible for you to check out my follow up questions which I will post tomorrow? I greatly appreciate your help....kind of blind on this path.

 

[technome]

Will check for posts tommorrow

 

[bingboo]

Hi ---
after reading this I am thinking that it sounds like the whole idea of having a hard drive backup offsite might be a bad idea. I have a test server that I could try out the acronis images on...but what's this about using active directory?
Hope you are having a good day.

 

[technome]

If you have a workgroup server versus Active Directory, it is easy to have a drive offsite. If you have a good tape drive with a workgroup server it is easy to get the system running again in no time. AD complicates a restore.