Re: Best way to back up a slave harddisk?

"Mike" <> wrote in message
news:...
>I have a slave 250Gb sata drive in my pc.
>
> I want to back this slave drive up as this is where i keep all my files,
> but
> dont fancy writing 250Gb to dvd's. Can anyone confirm one of these
> ghost/image softwares will do this?. Does anyone kow how many dvd's it
> will
> use?.


The expensive, but effortless and safe, way to backup would be to get a NAS
(network attached storage) box with a RAID1 array (two mirrored drives).
Backing up could be completely automated.

ss.

"John Jordan" <> wrote in message
news:...
>
> RAID 1 is not a backup :-)
>
> Seriously, cheap RAID 1 controllers will normally ignore reallocated
> sectors, which can make them less reliable in some common failure modes
> than a single drive.


I have been using RAID1 for a few years now, and I have not had any
problems. RAID1 has saved my life on a couple of occasions. If what you
say is true, why are people using RAID1 at all? There is no mention of this
on any explanations it either.
e.g. http://en.wikipedia.org/wiki/Standar..._levels#RAID_1

ss.

Bitstring <>, from the wonderful
person Synapse Syndrome <> said
>"John Jordan" <> wrote in message
>news:...
>>
>> RAID 1 is not a backup :-)
>>
>> Seriously, cheap RAID 1 controllers will normally ignore reallocated
>> sectors, which can make them less reliable in some common failure modes
>> than a single drive.
>
>
>I have been using RAID1 for a few years now, and I have not had any
>problems. RAID1 has saved my life on a couple of occasions. If what you
>say is true, why are people using RAID1 at all? There is no mention of this
>on any explanations it either.
>e.g. http://en.wikipedia.org/wiki/Standar..._levels#RAID_1

All RAID1 is not created equal. Professional implementations (expensive
controllers) do intelligent things when the two volumes mismatch. Cheap
controllers often don't - sometimes they do horrendous things, sometimes
they just hide the failure.

As long as both disks work fine RAID1 does nothing. If one disk starts
to fail, RAID1 needs to start working .. that's the point at which you
discover if your controller is useful, or just marketing-ware. If you
got it for free, suspect the worst.

Me, I'd rather have two separate drives (in separate PCs, in separate
rooms, on separate UPSs) with regular 'second copy' type backup. Yeah, I
might lose a few hours data, but if your RAID1 controller goes belly up
you might lose all of everything, if that's the only copy you had.

The people who use in it anger (me, at work, once upon a time) it
typically need 100% (or close to it) uptime, and are willing to pay for
it (well, are willing to pay for RAID5 anyway).

--
GSV Three Minds in a Can
8,013 Km walked. 1,453Km PROWs surveyed. 26.4% complete.

"John Jordan" <> wrote in message
news:...
> Synapse Syndrome wrote:
>> "John Jordan" <> wrote in message
>> news:...
>>
>>>RAID 1 is not a backup :-)
>>>
>>>Seriously, cheap RAID 1 controllers will normally ignore reallocated
>>>sectors, which can make them less reliable in some common failure modes
>>>than a single drive.
>>
>> I have been using RAID1 for a few years now, and I have not had any
>> problems. RAID1 has saved my life on a couple of occasions. If what you
>> say is true, why are people using RAID1 at all?
>
> Why do so many people use HD models that turn out to have serious
> reliability issues? Because they don't know. To my knowledge, there's
> never been a large scale HD reliability study that breaks results down by
> drive model with any kind of scientific standards. The same applies to
> RAID, at least with consumer hardware.
>
>> There is no mention of this on any explanations it either.
>> e.g. http://en.wikipedia.org/wiki/Standar..._levels#RAID_1
>
> The "RAID 1 failure rate" section there assumes that HDs die outright.
> IME, this is only true in around half of HD failures. Ideally a RAID
> controller would mark a disk as failed at the first sign of trouble
> (pending sectors etc.), but the cheap ones don't seem to do this.
>
> In all three RAID 1 disk failures I've personally worked on, the failing
> drive had caused filesystem damage on the working drive. Not a huge
> sample, but this is an effect that seems to be totally ignored in
> discussions of RAID 1.

Have you got any references at all for this? Would Intel southbridge RAID
controllers constitute a 'cheap' controller?

ss.