[Gllug] RAID on RAID

[Rich Walker]

Christian Smith <csmith at micromuse.com> writes:

> On Wed, 3 Nov 2004, Rich Walker wrote:
>
>>
>>Since the list is quiet...

[warms hand over open flames]

>>
>>I was pondering disk allocation strategies the other afternoon, and I
>>suddenly realised I've allocated my root partition all wrong on the
>>server...
>>
>>At the moment, the root partition is the only un-raided partition. There
>>are 4 disks, on separate cables, with a bunch of partitions doing a
>>bunch of stuff.
>>
>>The idea that occurred to me was to allocate a ~5GB chunk of each disk
>>and then do
>>hda1 + hde1 => md0, RAID1
>>hdc1 + hdg1 => md1, RAID1
>>md0 + md1 => md2, RAID1
>>
>>and then mount md2 as /
>
>
> Urgh! If you want to make use of all four disks you'd be better off
> having a RAID5 configuration. 

I've got a RAID5 configuration elsewhere, thanks. This is for the
occasion when a large chunk of the machine is down for whatever reason.

> So long as your initrd and lilo/grub boot
> off a non-RAIDed boot partition you should be dandy to mount a RAID5
> root.

That's exactly the opposite of what I want to do. In the event of a
substantial failure, I want /, /boot, /etc, /lib, /bin and /sbin
available during the recovery process. So I want them on multiple disks
*UP*TO*DATE* in a bootable form. Hence RAID1-over-RAID1.


> Then you simply have multiple copies of /boot on different disks for
> recovery if /boot is lost.



>
>
>>
>>Now, clearly write will be slow :-> But write to / is rare - most writes
>>go to /home, /var, /tmp and some to /big.
>>
>>Reads should alternate between md0 and md1.
>>
>>If any one disk controller goes down, no problem.
>
>
> Controllers would be the last thing to fail. They are solid state
> electronic devices. You'd probably have problems with drives first, then
> cables.

Upgrades. Idiocy. Dropping the box on the floor. Power problems.

All of these I've seen break stuff.

And I'm not even going to tell you what work experience people can do to
hardware :->



>>If any three disks go down, no problem.
>
>
> If you are planning for redundency to survive three disk failures at any
> one point, you should have a new supplier you trust.

Does not compute. I am planning for redundancy because modern disk
capacity and pricing allows me to (Redundant Array of Inexpensive
Drives, remember?). When something bad happens to the server, greater
redundancy is better - it increases survivability...

>
>
>>
>>But will LILO or GRUB cope with booting from RAID-on-RAID? And have I
>>missed anything?
>>
>>cheers, Rich.
>>
>>
>
> -- 
>     /"\
>     \ /    ASCII RIBBON CAMPAIGN - AGAINST HTML MAIL
>      X                           - AGAINST MS ATTACHMENTS
>     / \
> -- 
> Gllug mailing list  -  Gllug at gllug.org.uk
> http://lists.gllug.org.uk/mailman/listinfo/gllug

-- 
rich walker         |  Shadow Robot Company | rw at shadow.org.uk
technical director     251 Liverpool Road   |
need a Hand?           London  N1 1LX       | +UK 20 7700 2487
www.shadow.org.uk/products/newhand.shtml
-- 
Gllug mailing list  -  Gllug at gllug.org.uk
http://lists.gllug.org.uk/mailman/listinfo/gllug