[mythtv-users] Storage space question

[Marc]

> -----Original Message-----
> From: mythtv-users-bounces at mythtv.org [mailto:mythtv-users-
> bounces at mythtv.org] On Behalf Of Ismo Tanskanen
> Sent: Sunday, January 20, 2008 4:39 AM
> To: Discussion about mythtv
> Subject: [mythtv-users] Storage space question
> 
> I have four 750Gb disks in my backend server.
> 
> I'm wondering, how to set them up. I want them as media storage, not
> for
> recordings. I have one separate big disk for recordings.
> My goal is to have one, big consistent space, 3TB if possible.
> 
> I could go with LVM, but how it handles disk error? I don't really need
> RAID safety, but I think that in LVM, if one disk fails, I lose whole
> filesystem. That does not sound good.
> 
> Is any filesystem capable of join disk spaces together, but still keep
> them "separately", that if one disk fails, other three still retains
> data they hold?
> 
> Of cource I could mount disks separately under medias-folder, but it
> not
> flexible enough since I have lot of folders there.
> 
> Thanks,
> -Kane

Kane,

You said you do not need the redundancy of Raid but then mention that you do
not want to lose data if a disk fails. Which is it? Lol

The 2 ways I know of that would give you one large volume are LVM and Raid
0. Both essentially so the same thing but in a different way. LVM merges all
the drives into one. Raid 0 does this too but how it does it is different.
It spans the data across all 4 disks so that a part of the file is on disk
1, 2, 3, and 4. Raid 0 would offer better read/write speeds than LVM would
because of this spanning. Both of these configurations however mean that if
one drive fails all the data is lost.

The only other way to not lose all your data if one failed would be to not
have them be "one big consistent space" as you called it. What this would
mean mounting the 4 drives individually at different points. For instance
/media1, /media2, /media3 and /media4. You could name them anything. Then
you would setup 4 storage groups, default1 through default4, again could be
named anything.

What this would then mean is that, when it was time to record, Myth would
check all of the storage groups and pick one to record on. My understanding
of this is that it will pick the drive that has the most room and record
there. The next time it would pick another drive, and another, etc. The only
problem with this type of setup is that if one drive fails you can lose 25%
or more of your recordings.

Since you have more than 3 drives this would be a perfect candidate for Raid
5. This would give you one big drive with redundancy that would allow you to
lose 1 drive without loss of data as well as give you some of the speed
benefits of a Raid 0 configuration. However, you would lose approximately
30% of your total storeage of the drives to give you this redundancy. Raid 5
would write the data across all 4 disks similar to the following, Data would
be written to disk 1 then 2, 3, and 4. Then once you looped back to disk 1
it would write a parity bit. Then it would write data one 2, 3, 4, and then
1, and write another parity bit on drive 2. Then start on drive 3 and begin
the loop again.
I am using Raid 5 for my media drive. I have 4 320GB drives, total storage
would be approx. 1.25 TB. In Raid 5 configuration I have approx. 860GB of
usable storage. Yes I lost about 33% but I know that if I lose a drive I
will not lose my video.

Marc