[vdr] Re: OT: 149 GB Maxtor disks.

[Emil Naepflein <Emil.Naepflein@philosys.de>]

On Fri, 19 Oct 2001 23:34:28 +0200, Lars Bensmann <lars@almosthappy.de>
wrote:

> You have good nerves. I have had enough trouble with hard disks already.
> I wouldn't bet the data on 7 disks on one disk failure. BTW, a Raid-5
> would not make a big difference as a disk failure frequently takes the
> whole IDE channel down. And when two drives are gone Raid-5 is dead as
> well.

This doesn't matter for me when the ide channel is down on a disk
failure. It isn't important to stay online on a disk failure, it is
important to preserve the data. After the failed disk is removed the
system will come up again in degraded mode and the data will still be
available. 

BTW, I *never* had any disk failure on any PC-system I use since 15+
years. And I have used a lot of them. I know, some people have a lot a
bad luck with disks, but may be the reason are the selected disks and
the way they are used. What can you expect from a high performance disk
running constantly at the limit regarding use and environment
(temperature, shock, ...)? 

> 
> Maybe you could install two Raid-5 devices with just one drive at one
> channel. E.g. all masters form a Raid-5 and all slaves form the other
> one. So when one channel goes down two Raid-5 would be running
> unsyncronized, but the data was fine. Of course for this you would have
> to sacrifice another disk for parity, so you would have "just"
> (3 * 80 GB) + (2 * 80 GB) = 400 GB left.
> 
> On the two Raid-5 devices you could run LVM to hide the physical
> seperation.

This would be a possibility, but IMHO it isn't worth to scarifice of 80
GB here. i don't need 100 % system availability, I only need 100 % data
availability in case of a single disk failure.

Emil