Intel raid 1, better than a backup. Reliable?

JonJ678

JonJ678

JonJ678

Soldato
Joined
22 Dec 2008
Posts
10,369
Location
England
Hey. Currently, i have two hard drives. I periodically copy all the data from one to the other. This is tedious, and will be more so when one fails.

Instead, i'd like to set them as a raid 1 pair, let them sync, switch off the computer, and unplug one. That should leave me with two identical drives, either bootable. It's the next bit i'm unsure about.

1/ if i leave one drive unplugged for a week, then reconnect it, the array should resync. Will this take a sane length of time, and will it always leave me with two copies of the newer data?

2/ say i use one drive of the degraded array for a week. Then something goes wrong, so i use the other for a few days (swap the cable over). What happens when i plug both in at once?

3/ the system is multiboot, currently xp, xp, debian, debian testing, using grub 2. Will this play nicely with intel's ich9r?

I'd like to be able to plug in the other drive & keep working if one corrupts. It would be good for the backup (raid syncing) to occur whichever os i'm in, without scripting. However i have doubts regarding how resilient the scheme would be, as spending 90% of the time degraded isn't the obvious use of raid. Also, i suspect the onboard raid will be inherently unreliable. Hence this thread, how reliable would the new scheme be?

Cheers
 
depends how you defines 'backup'?

before i answer questions, why would you need to unplug one of the raided hard drive? you only replace one of them when they break down.
 
I think I'd define backup as a secondary set of data that will survive destruction of the primary.

Two reasons in favour of unplugging the hard drive. The big one is that it provides a way to easily "recover" files I've foolishly deleted or accidentally corrupted. If windows falls over, or I absentmindedly break debian, I can plug the other drive in and continue as if the error never happened. A live raid would tend to propagate mistakes to the other set of data.

The other attraction is that the unplugged drive will survive psu/motherboards blowing up. This is desirable. I might go so far as to keep the second hard drive outside of the case, and connect it via esata when needed.

Thanks for the question, I think the above clarifies matters.

p.s. having some minor editing problems as a result of typing this from a phone.
 
Last edited:
hi

then raid 1 is not what you want.

esata or USB external hard drive is what you want.

fooling about with OS's, software etc etc - virtualbox is what you want :)
 
Sadly virtual box isn't comparable to a native install, otherwise snapshots would pretty much solve one concern. The performance hit is too great.

An external disk + robocopy or rsync backs up raw data well. I have not been able to maintain a bootable clone that way however. The closest i've come is a full disk image using dd. This usually works if you give it a very long time to run, and have the normal os offline for the copy. Raid 1, normally used degraded, looks like a better way to backup the data, os and programs than this.
 
I don't see the relevance of the PC spec. Vbox < native performance, whatever the spec is. As it happens, the PC in question is a reasonably quick one, used for some geometric modelling, some fea work, and a lot of matlab. A browser or text editor within a virtual machine is no bother, but I'm not going to deliberately slow down code that takes a significant length of time to execute at the best of times.

It's an x58 system, i7 920 @ stock (chasing stability problems somewhere in the memory subsystem), 12gb corsair, gigabyte UD5 and an evga 8800gt running under the mistaken impression that it's a FX3700 for the sake of driver optimisation.

I can't see a way to choose which of the two disks is cloned across under different circumstances without nuking the mbr beforehand, which would mean the array will take ages to rebuild each time. The compromise is probably to set up the OS as desired, clone the drive, put the clone somewhere safe and backup any accumulated data to another system in the current fashion. If the main OS collapses, I'll lose some settings and have to copy the data back, but at least I wont have to reinstall windows, ansys and so forth at an inconvenient time.

I should try to find out if rsync can deal with ntfs permissions intelligently enough to restore windows entirely to an earlier point in time or not too. That might be an improvement, store a validated copy somewhere on the network and rsync --delete from the validated copy if the working one breaks.
 
RAID is not backup.

It is redundancy.

End of.

Nothing wrong with taking a RAID 1 disk out of the array as a 'base' to store and use later if a failure occurs, restore data back afterwards etc. That is about as far as I would take it - a Disaster Recovery option, not ongoing backup.
 
Sin_Chase said:
RAID is not backup.

It is redundancy.

End of.

Nothing wrong with taking a RAID 1 disk out of the array as a 'base' to store and use later if a failure occurs, restore data back afterwards etc. That is about as far as I would take it - a Disaster Recovery option, not ongoing backup.
Click to expand...

This basically - and I'm sure you already know this from your project in your Sig :)

Let's face it, if your running dual boot? with a debian based linux install ... you are going to break it at some point. no matter what, especially if your tinkering with files and tweaking it up.

Have you considered moving your linux install to a partition on the 2nd drive with a shared data partition. At least then you can recover into windows with a simple mbr repair.

OR

Setup a scheduled nightly/weekly backup image of the main drive to the second drive mounted in a eSATA/USB external hotswap caddie as mentioned above. granted this is more of a traditional backup solution but more sensible surely?

As already mentioned the safest route is virtualisation but you've already ruled that out. But, if your dabbling with the unknown or a non stable OS, you may have to re evaluate if your actually causing more problems than really necesary by not using it?

Running a degraded RAID 1 array does not seem, to me anyway, the way to go.
 
You must log in or register to reply here.
Back
Top Bottom