[mythtv-users] drive configuration dilemmna

Jay Ashworth jra at baylink.com
Tue Mar 29 21:56:23 UTC 2011


----- Original Message -----
> From: "Mark Lord" <mythtv at rtr.ca>

> A problem I have with RAID, is that it multiplies the odds of a failure
> causing data loss. Eg. 6 drives = 6X the likelihood of a failure that
> affects all data. Sure, perhaps the array will tolerate a single drive
> failure (which is kind of the whole point), but then you're running on
> the edge for a day or three until that drive is replaced and the RAID
> resync'd. I know of several people who "lost everything" during such an
> interval.

The second half of that confuses the issue, but you're doing your math wrong.

You say "multiplies the odds of a failure causing data loss".  Let's go
to the blackboard.

For easy to follow numbers, let us say that we have 2 drives in a RAID1 mirror,
and that the instantaneous odds of one drive failing (which will not lose) data
are 10%.

So, what are the odds we'll have 2 failures?

10% = 0.1

0.1 x 0.1 = 0.01, or *one percent*.  IOW, no, the odds don't go up, they go
down.  Certainly, there's a window where a second failure will kill you,
but if you *didn't* have the RAID, *the first failure -- with much higher
probability -- would have killed you.

As to multiple failures in the same RAID group, someone pointed out on 
here last week that there are ways to get around this, including but not 
limited to making sure your drives have different batch nuumbers on them,
or serial numbers with different prefixes.

All that said, yes, as drive capacities outstrip interface speeds, we're 
going to see a lot of the same problems we saw as drive capacities
outstripped tape capacities (y'know, *real* backup, which RAIDs are *not*.)

Cheers,
-- jra


More information about the mythtv-users mailing list