Admin Mailing Lists wrote:
>
> I run 5 linux boxes and 2 NT servers on a network. All servers have 9G
> scsi drives in them. Currently we back up the most important data to DDS3
> tape..about 11G. We would like a better solution to this for full
> redundancy. We thought about putting another 9G in each server and doing
> software raid1 (on linux at least, dunno how we could do this in NT)
>
> So we thought, ok that solves us if we have a hardware (drive) problem,
> but what if it's software that gets corrupted along the way, or something
> gets installed that on reboot will break it. Then you got 2 exact copies
> of bad data. For example..our 1 NT server was up for like a month and a
> half, I rebooted it one day and it never came back up. It wasn't a
> hardware problem, but something with windows. Now if i was doing raid 1,
> it wouldn't matter, i'd still have a windows that wont boot on both
> drives.
>
> Any help or advice would be greatly appreciated.
We had a similar situation here...small LAN with 4 WindowsNT machines
and 8 Linux machines. We decided *against* mirroring each drive for the
very reasons you pointed out. Our solution to the problem was to *build*
a "backup server", consisting of a 100GB UDMA Raid-5 array, Seagate
4x50GB Sidewinder AIT autoloader, 12bay case w/ dual hot swappable power
supplies (each connected to a different UPS, connected to a different
circuit), 500Mhz P-III w/ 128MB memory, running RedHat 6.0. The Raid-5
array was set up using 5x25GB IBM UDMA disks in a RaidZone Smartcan and
raidtools-0.90. Total cost was around 10k. We could probably have gotten
by for about half that, but we decided to build-in extra capacity for
future growth. We created a mirror image of each client machine on Raid
filesystem, and then run an "rsync" every hour to synchronize the backup
server with each client machine. We do a full backup of the "mirror"
every friday night, and an incremental backup of the "mirror" every
night *except* friday...
With this system, if a user deletes a file or directory, they can easily
replace it with a backup copy that is no more than 1 hour old. Should
disaster strike and all the filesystems on a machine get destroyed, all
we have to do is boot off an emergency diskette, repartition the disks,
remake the filesystems, and "rsync" a copy of each of the partitions
from the backup server...
John
--
John Burton, Ph.D.
Senior Associate GATS, Inc.
[EMAIL PROTECTED] 11864 Canon Blvd - Suite 101
[EMAIL PROTECTED] (personal) Newport News, VA 23606
(757) 873-5920 (voice) (757) 873-5920 (fax)
