Raid Question

Possibly the wrong groups, but not too sure where to post this
question. A friend of mine has a Raid array, with three 146gb drives
in a compaq server, and then three 36gb drives, below that. The drives
were not re done, as the originals were the 36's, and they were just
swapped around in the cage they are in, the 146's up front at the top,
but the old ones at the bottom, still with the OS on them. The new
drives were configured as extra partitions.

Now my question is, with the 6 drives, between the partitioning and
setup (done by a professional company I am told) there is a missing
amount of 200gb between all the drives. Now I know that you loose a
certain amount with formatting and all that, but 200gb? Someone I know
who works in raids and just raids said that it doesnt appear to be
configured correctly, but without him seeing it (he doesn't live
anywhere near me or my other friend!) he cant be too sure, so I am
asking, does this sound right?

The company who set it up to begin with says the missing part is
normal, even up to that size. Just all seems a bit off to me.

TIA

-Ben

On Tue, 10 Mar 2009 09:21:54 -0700 (PDT), Wagg <(E-Mail Removed)>
wrote:

>Possibly the wrong groups, but not too sure where to post this
>question. A friend of mine has a Raid array, with three 146gb drives
>in a compaq server, and then three 36gb drives, below that. The drives
>were not re done, as the originals were the 36's, and they were just
>swapped around in the cage they are in, the 146's up front at the top,
>but the old ones at the bottom, still with the OS on them. The new
>drives were configured as extra partitions.
>
>Now my question is, with the 6 drives, between the partitioning and
>setup (done by a professional company I am told) there is a missing
>amount of 200gb between all the drives. Now I know that you loose a
>certain amount with formatting and all that, but 200gb? Someone I know
>who works in raids and just raids said that it doesnt appear to be
>configured correctly, but without him seeing it (he doesn't live
>anywhere near me or my other friend!) he cant be too sure, so I am
>asking, does this sound right?
>
>The company who set it up to begin with says the missing part is
>normal, even up to that size. Just all seems a bit off to me.

What level of RAID is it? Different types "waste" different quantities
of space.

In this case, I'd guess from your figures that you have two RAID5
arrays of three drives each. RAID5 allows you to lose one disk in the
array without ill effect, and when you replace the failed disk the
array rebuilds back to fault-tolerant. It does this by using
equivalent space to one whole disk as "parity" error checking codes.

Boot array: 3x36, of which a third is used for parity.
Data array: 3x146, of which a third is used to parity.

So you have 146+36 gig "wasted" as parity, which is 182gig - quite
close to the 200gig you say.

Common RAID types:

RAID0 (or "not a RAID): 2x36gig disks, gives 2x36gig space as one
virtual disk, and no safety
RAID1 (mirror): 2x36gig disks, gives 1x36gig space as one virtual disk
and 1 disk fail tolerance
RAID5 (parity): Nx36gig disks, gives (N-1)x36gig space as one virtual
disk and 1 disk fail tolerance

Also,
JBOD (also "not a RAID", Just a Bunch of Disks): N disks of any sizes,
gives the combined sum of diskspace as one virtual disk with no
safety.

Cheers - Jaimie
--
I like nonsense. It wakes up the brain cells. -- Dr. Seuss

In article <b6a82516-5d34-4b1c-8f29-(E-Mail Removed)
..com>, Wagg <(E-Mail Removed)> writes

>Now my question is, with the 6 drives, between the partitioning and
>setup (done by a professional company I am told) there is a missing
>amount of 200gb between all the drives.

Since the minimum number of drives needed for RAID 5 is three, and he
has three of each drive fitted, that strongly suggests RAID 5 is in use.

So if they are configured in RAID 5, that's okay. The usable capacity
for RAID 5 is (number of drives) - 1.

3 x 36GB in RAID 5 = 2 x 36GB = 72GB

3 x 146GB in RAID 5 = 2 x 146GB = 292GB

So where he would expect (3x36) + (3 x 146) = 546GB, he actually has (2
x 36) + (2 x 146) = 364GB, a difference of 182GB.

Deduct about another 10% from the usable capacity for formatting and
filesystem overhead and you have a yield of 364 - (10% of 364) = 327GB
for file storage.

Add the differences together: 182 + 36.4 = 218.4GB "missing" (it's not)

And that's not taking into account the difference between 1GB = 1000Mb
(what the drive makers say) and 1GB = 1024Mb (what computers say)...

--
(\__/)
(='.'=) Bunny says Windows 7 is Vi$ta reloaded.
(")_(") http://imgs.xkcd.com/comics/windows_7.png

On Tue, 10 Mar 2009 18:05:52 -0000, "Synapse Syndrome"
<(E-Mail Removed)> wrote:
>Jaimie Vandenbergh <(E-Mail Removed)> wrote:
>>
>> Also,
>> JBOD (also "not a RAID", Just a Bunch of Disks): N disks of any sizes,
>> gives the combined sum of diskspace as one virtual disk with no
>> safety.
>
>I've never seen the point of JBOD. Do people actually use it?

I've been using one for a couple of years now.

>If one drive
>fails, you lose everything (without having to go through a recovery
>process - and what happens to files that span different physical drives?)

It's not striped, so you'll lose everything from the failed drive
forward - how fatal that is depends on which disk pops and how
resilient your filesystem is.

>and there is no advantage as far as I can see. It would be much better to
>extend a drive using volume mount points or symbolic links, no?

Here's my use case: I've got a NAS box with 3x1Tb in, RAID5 so 2Tb
disk space. I like to back it up, because RAID isn't backup of course.

So it's backed up nightly to a 2Tb JBOD. The 4x500gig that used to be
in the NAS are now in a four bay USB enclosure (Edgestor DAS400) that
has a "JBOD" switch on the back.

Nice and simple, completely transparent to the host (which is the NAS,
so it needs to be transparent), and without any risk of individual
folder trees running short of space. And if a disk pops, who cares?
The original is still RAID5 protected, so bang in a fresh disk,
resync, done.


I'd never recommend them for anything else except second or later
level backups, obviously.

(previously, when the 4x500gig RAID5 = 1.5Tb were in the NAS, I backed
that up to a 4x320gig JBOD that was the original content of the NAS -
recycling at work! I upgraded when the JBOD ran out of space)

Cheers - Jaimie
--
If it's worth doing, it's worth overdoing.

On Tue, 10 Mar 2009 18:29:27 -0000, "Synapse Syndrome"
<(E-Mail Removed)> wrote:

>Jaimie Vandenbergh <(E-Mail Removed)> wrote:
>>
>>> If one drive
>>> fails, you lose everything (without having to go through a recovery
>>> process - and what happens to files that span different physical drives?)
>>
>> It's not striped, so you'll lose everything from the failed drive
>> forward - how fatal that is depends on which disk pops and how
>> resilient your filesystem is.
>
>
>BTW, when I said "drives that span different physical drives", I meant that
>I assumed that some large files may have some data on one drive, and the
>rest on the next drive in the JBOD. I didn't mean that I thought the data
>is striped through all the drives.

That's what I thought you meant, but I decided to cover the stripe
possibility since a file that's half on a dead drive is clearly half
dead. When trying to recover stuff, the best bet by far is to leave
the disks in the JBOD array while you do - otherwise as you suspect
files split across two good drives will be impossible to reconstitute.

But you should never put yourself in the situation where your only
hope is to recover from a broken JBOD. That's as crazy as not having
backups at all.

Cheers - Jaimie
--
I have seen the fuschia, and it's ... er ... sort of pink and mauve
with bits hanging out. -- Richard Robinson, urs

On Mar 10, 6:38*pm, Jaimie Vandenbergh <jai...@sometimes.sessile.org>
wrote:
> On Tue, 10 Mar 2009 18:29:27 -0000, "Synapse Syndrome"
>
>
>
> <syna...@NOSPAMsyndrome.me.uk> wrote:
> >Jaimie Vandenbergh <jai...@sometimes.sessile.org> wrote:
>
> >>> If one drive
> >>> fails, you lose everything (without having to go through a recovery
> >>> process - and what happens to files that span different physical drives?)
>
> >> It's not striped, so you'll lose everything from the failed drive
> >> forward - how fatal that is depends on which disk pops and how
> >> resilient your filesystem is.
>
> >BTW, when I said "drives that span different physical drives", I meant that
> >I assumed that some large files may have some data on one drive, and the
> >rest on the next drive in the JBOD. *I didn't mean that I thought the data
> >is striped through all the drives.
>
> That's what I thought you meant, but I decided to cover the stripe
> possibility since a file that's half on a dead drive is clearly half
> dead. When trying to recover stuff, the best bet by far is to leave
> the disks in the JBOD array while you do - otherwise as you suspect
> files split across two good drives will be impossible to reconstitute.
>
> But you should never put yourself in the situation where your only
> hope is to recover from a broken JBOD. That's as crazy as not having
> backups at all.
>
> * * * * Cheers - Jaimie
> --
> I have seen the fuschia, and it's ... er ... sort of pink and mauve
> with bits hanging out. * * * * * * * * * * -- RichardRobinson, urs

I think it is a Raid 5, but will get him to check. Making a bit more
sense now! I will find out and post back if needed, if that doesnt
answer it all!

-Ben

In message
<528f4b1b-365d-4da1-b149-(E-Mail Removed)>,
Wagg <(E-Mail Removed)> writes
>I think it is a Raid 5, but will get him to check. Making a bit more
>sense now! I will find out and post back if needed, if that doesnt
>answer it all!
Hopefully he will have the HP array configuration utility loaded.
Opening that up will tell you exactly how the disks are configured and
make the lights on the disks flash to tell you which disk is in which
volume/RAID set when you select the controller, RAID set, volume etc...
>
>-Ben

--
Clint Sharp

In message <(E-Mail Removed)>, Synapse Syndrome
<(E-Mail Removed)> writes
>I've never seen the point of JBOD. Do people actually use it?
I've seen them a few times in commercial settings.
> If one drive
>fails, you lose everything (without having to go through a recovery
>process - and what happens to files that span different physical drives?)
Not to be used for anything that isn't transient or backed up elsewhere
but the 'you lose everything' may not actually be that important
depending on the use it's put to.
>,
>and there is no advantage as far as I can see.
It's large, cheap and relatively fast storage that can be used as a
buffer for backup purposes or used for replicated systems where there is
no reason to have multiple redundant RAID sets because all servers
always have full copies of the data on the individual JBODs.

Perversely, the more you spend on storage the more likely you are to see
JBOD arrays.
--
Clint Sharp

In article <b6a82516-5d34-4b1c-8f29-fe3431c8d8ca@
41g2000yqf.googlegroups.com>, (E-Mail Removed) says...
> Possibly the wrong groups, but not too sure where to post this
> question. A friend of mine has a Raid array, with three 146gb drives
> in a compaq server, and then three 36gb drives, below that. The drives
> were not re done, as the originals were the 36's, and they were just
> swapped around in the cage they are in, the 146's up front at the top,
> but the old ones at the bottom, still with the OS on them. The new
> drives were configured as extra partitions.
>
> Now my question is, with the 6 drives, between the partitioning and
> setup (done by a professional company I am told) there is a missing
> amount of 200gb between all the drives. Now I know that you loose a
> certain amount with formatting and all that, but 200gb? Someone I know
> who works in raids and just raids said that it doesnt appear to be
> configured correctly, but without him seeing it (he doesn't live
> anywhere near me or my other friend!) he cant be too sure, so I am
> asking, does this sound right?
>
> The company who set it up to begin with says the missing part is
> normal, even up to that size. Just all seems a bit off to me.

Depends which RAID mode, i.e. morroring or striped.

EG 2 80s in a mirror array would read as 80Gb (less a bit for
formatting). In striping mode would show as about 150Gb.
--
Regards
Jon

In article
<b6a82516-5d34-4b1c-8f29-(E-Mail Removed)>, Wagg
<(E-Mail Removed)> writes
>Possibly the wrong groups, but not too sure where to post this
>question. A friend of mine has a Raid array, with three 146gb drives
>in a compaq server, and then three 36gb drives, below that. The drives
>were not re done, as the originals were the 36's, and they were just
>swapped around in the cage they are in, the 146's up front at the top,
>but the old ones at the bottom, still with the OS on them. The new
>drives were configured as extra partitions.
>
>Now my question is, with the 6 drives, between the partitioning and
>setup (done by a professional company I am told) there is a missing
>amount of 200gb between all the drives. Now I know that you loose a
>certain amount with formatting and all that, but 200gb? Someone I know
>who works in raids and just raids said that it doesnt appear to be
>configured correctly, but without him seeing it (he doesn't live
>anywhere near me or my other friend!) he cant be too sure, so I am
>asking, does this sound right?
>
>The company who set it up to begin with says the missing part is
>normal, even up to that size. Just all seems a bit off to me.
>
>TIA
>
>-Ben


There are different types of RAID. However, 3 disks typically means RAID
5.

In RAID 5 a disks worth of storage in the RAID group is used by the RAID
to provide redundancy.

I.e. 3 x 146GB = about 265GB after formatting.

"loosing" 200Gb on a RAID of 3 x 146 and one of 3 x 36 sounds about
right.


BR
Don C