Basically, the RAID5 rack fails and they need to get the data off. It's been created on Windows, but the theory behind RAID should mean that even with a missing drive we should be able to recreate it.

The solution involves Knoppix, a certain amount of trial and error and some simple programming to completely recreate the old RAID volume based on images taken of the disk.

Many of the principles here could just as easily have been carried out on any Unix system -- Knoppix really only features as an effective way of copying the original disks off (using dd) into an image to be used when recreating the volume.

I've done similar things - on all sorts of platforms - to recover data or repair machines. Some of the simpler things are recovering the text and formatting from a document that had been corrupted in Word - I opened it with OpenOffice, which didn't have the same hang-ups as Word did, resaved it as RTF and then opened it again in Word, and presto! Document back.

Using dd, I've stripped sections from documents or headers that were causing problems (a package courier here in the UK supplies a PDF file for the address label, but when you download it they include a 128-byte header of stuff you don't need which just confuses Acrobat), and I've lost track of the number of times I've used emacs to recover, repair or reformat something that was considered damaged beyond repair by a client or associate.

It's these types of lateral thinking that turn your good administrator into an excellent IT administrator.

Now this is not really a Linux story - and I'm not trying to dress it up as such - but Linux  (and Unix) comes with a lot of these low-level tools that enable you to work with information without protecting you from it all the time.

From an 'average joe' perspective, the protection is good, but when it comes to an administrator or IT expert repairing or resolving the issue, that low-level access and toolset becomes invaluable.

www.computerworld.com