My point is that, with RAID 0, if one disk fails the data on all the  disks
is lost.  Also, MTBF is additive in this case because of what I previously
said.  You've got more than one set of platters and heads to fail and any
one of them failing blows away the data.  I realize the chances of problems
are very small but drives do still fail.  My day job is architecting and
administering very large database systems (1 Terabyte RAIDS) and I do see
drives fail.  I'd never use RAID 0 for one of these systems.  

For typical use though, you are right that it's very unlikely to be a
problem.  I just think it's kind of unnecessary unless there's a justified
need for performance.  Video editing would be a good example.

To keep this remotely on topic, I just ordered a Minolta Scan Multi Pro.  I
should have it in 2 weeks.

Paul Wilson

> -----Original Message-----
> From: Austin Franklin [mailto:[EMAIL PROTECTED]]
> Sent: Sunday, November 11, 2001 7:40 PM
> To: [EMAIL PROTECTED]
> Subject: RE: filmscanners: Best solution for HD and images
> 
> 
> > I just wanted to note that RAID 0 is, in most cases, a bad idea.
> > The reason
> > is that if you stripe your data across multiple disks and 
> one fails, you
> > lose all the data.  It's better to split the files up among 
> many, smaller
> > logical drives.  It's great from a performance standpoint 
> but that's about
> > it.  RAID 0+1 or RAID 5 are much better ideas.
> >
> > Paul Wilson
> 
> I disagree that it's a bad idea.  It's no more "unsafe" than 
> a single disk.
> MTBF is NOT additive.  RAID 0 IS the fastest, and if that's 
> what you need,
> then it's a good idea.
> 
> Also, if you are using it as a data store, which is typically 
> what RAID is
> used for, instead of a main system disk, then you SHOULD be 
> backing up.
> 
> I have dozens of hard disks in my multiple machines, and haven't had a
> failure in years, and they are on 24/7/365.  The MTBF of the 
> drives is far
> less than the next technology leap that I replace the disks 
> for anyway.
> 

Reply via email to