Red Hat Linux USB backup solution.

The problem might be performance...

Is the USB drive of roughly the same type as your internal drive? When writing to a mirrored pair (mirrorred by the kernel, not by rsync & the like)
a write succeeds when both writes are done, so if you have f.e. a 10K rpm disk internaly and a 5400 rpm disk in the USB driver that might
create a bottleneck.

Otherwise mirroring can be done between any 2 equal sized partitions. jast have a look at mdadm etc.

Well I have not even bought the USB drive yet. I am just looking for solutions to create a mirror image of my current drive, as easy as possible. I am no expert at linux and I would appreciate if the solution was easy to implement.

Right now the web server is on production so it's hard for me to open up the computer and install drives etc. Even more, I have never done this on a  Unix Machine so bare with me and try to explain as much as possible. If you think there are other easier or more reasonable solution to my problems and I am open for suggestions.

Eitherway I am looking for a fail safe way to replace my webserver in 10 minutes or less in case it crashes.

>Any recommendations?

Don't use USB!

Use something like this:

http://www.maplin.co.uk/Module.aspx?TabID=1&ModuleNo=37674&doy=17m7

It fits into an open 5.25in enclosure, and the hard drive can be pulled out manually without opening up your system. If you have 2 of them, then your main hard drive is configured as a primary (and your CD as the secondary) on one interface, and the second enclosure can contain your backup. to which you image the original. In the event of a drive failure, you shut down, replace the original with the image, and put a new hard drive in for imaging.

Similar enclosures exist for SATA drives. (or indeed if you need better performance, then you might start to look at scsi RAID with 'hot swap'.

(   (()
(`-' _\
 ''  ''

OK,

first look into the time it takes to replace parts...,

You'll probably want to go for some rackmounted disks that can be hotswapped. (ie. replaced without powerdown).
That are not tightly screwed inside your system.

Then place 2 (or if expenses allow) 3 disk in that rack, 2 will be mirrored the 3rd will be hot standby, that means the
lowlevel drivers will pull that disk in place of the failing one. (Of course you can also put 2 -> 3 disks inside your system)
You can sleep on to notice next morning you have to order a replacement disk if one fails.
Replace the defective one with the new arrival and configre that one as a spare. When you don't have a diskrack,
then you can at least plan your downtime to replace a defective disk).
BTW Most Diskracks are SCSI, although firewire also exists.
Prefereably members of a mirrorset are on different controllers so you also cover controller failure as well as disk failure,
and gain some performance benefits for being able to do update to both devices at the same time.

Info about mirroring:
http://gentoo-wiki.com/HOWTO_Gentoo_Install_on_Software_RAID
http://gentoo-wiki.com/HOWTO_setup_evms

These are gentoo oriented but should be applicable to a wider audience. ( emerge is the gentoo package manager,
equivalent to rpm + up2date on RedHat) If the docs tell you to emerge a package you should be able to find an equivalent one.

If you have more specific questions fire ahead

Noci...

Ok So how hard would it be to set up this software RAID solution? I just need to install another HD on my computer plug it in, boot, and configure it?
I need minimal downtime because this server is live and bringing it down is terrible. That's why I was looking at USB because this way I don't have to turn it off.

Also as said above, the easier the better, because I truly don't have much linux experience.

best regards.

FIRST: MAKE A BACKUP!!!!!!!

Easiest would be to first create the raid1 set(s) needed on a spare set of disks.
and copy all your data over.

First choose your options for the hard drives:
  External rack or the encosures pjedmond described.
  preferably in a external rack. (you would then have down time for installing the controller(s). and plugging them in.
  look for controllers that are bootable.

partition types are not (83/82) ext2/swap but fd (raidmember)

Starting a raidset from scratch is relatively easy
  create the partitions and define them as raid partitions,
  combine them to raid volumes,
  copy data over,  (part can be done during production, static data), parts that can change during production
  have to be done during downtime.
  see that the grub/lilo loader is on BOTH members bootsectors!!! [possibly the hotspare too]
    (this is not mirrorred at all)
    grub is a little easier on the maintenance as the grub.conf file resides within the (/boot) filesystem.
  reboot from your raid rack.
  optionaly add old disk as a spare member.
 
Cheapest on disks:
More trouble some:
  insert a new disk into your system,
  partionit, as you would do for a raid set,  but with your new disk configured corectly and a nonexistent! device for second disk
  copy all data over
  boot from the single member raid1 set and replace the bogus member with the old drive
  when a disk fails
 

To be honest I don't think I have enough linux knowledge or am confortable enough with linux to pull that solution off.
Right now I am looking at some of these solutions which seem to be fairly easy and documented: http://howtoforge.com/taxonomy_menu/1/34

well noci for your efforts you take the 300 points... even though I am still looking for options.
; )

I'll stay subscribed,
but now I am on holiday for a few weeks...

Please ask along if you need more info