FAQ Search Today's Posts Mark Forums Read
» Video Reviews

» Linux Archive

Linux-archive is a website aiming to archive linux email lists and to make them easily accessible for linux users/developers.


» Sponsor

» Partners

» Sponsor

Go Back   Linux Archive > Debian > Debian User

 
 
LinkBack Thread Tools
 
Old 03-11-2009, 11:11 AM
Sam Leon
 
Default RAID 5 data structures

I have a funny question. I have been playing with a 3 disk raid 5 setup
for my desktop. I guess I don't fully understand how the "stripe" is
managed or even what it is. I know the stripe is made up of a chunk
from each disk. Now I always thought of the stripe in raid the same as
a block in ext3 or a cluster in ntfs. Meaning if I have a 1k file that
I write to an ext3 filesystem with 4k blocks, my 1 k file will take up
one block thus wasting 3k of space.


Now I thought the stripe in raid followed the same principle. Meaning
if I have a 3 disk array with 64k chunks then my data stripe is the
number of disks minus one drive because of parity so in this 3 disk
array I would have a data stripe of 128k. So how much of that space
would my 1k file take up? Would it take up the whole 128k stripe or
just one chunk leaving the other chunk free for something else?


When I migrated my root drive over to the raid5 array I made, I was
afraid that it would use alot more space on the raid array since it is
full of very small files but to my surprise df -h reported the same
values for / on the array and / on the single disk that I had copied the
data from. So what is going on here?


Thanks,
Sam


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
 
Old 03-11-2009, 11:41 AM
Mark Allums
 
Default RAID 5 data structures

Sam Leon wrote:
I have a funny question. I have been playing with a 3 disk raid 5 setup
for my desktop. I guess I don't fully understand how the "stripe" is
managed or even what it is. I know the stripe is made up of a chunk
from each disk. Now I always thought of the stripe in raid the same as
a block in ext3 or a cluster in ntfs. Meaning if I have a 1k file that
I write to an ext3 filesystem with 4k blocks, my 1 k file will take up
one block thus wasting 3k of space.


Now I thought the stripe in raid followed the same principle. Meaning
if I have a 3 disk array with 64k chunks then my data stripe is the
number of disks minus one drive because of parity so in this 3 disk
array I would have a data stripe of 128k. So how much of that space
would my 1k file take up? Would it take up the whole 128k stripe or
just one chunk leaving the other chunk free for something else?


When I migrated my root drive over to the raid5 array I made, I was
afraid that it would use alot more space on the raid array since it is
full of very small files but to my surprise df -h reported the same
values for / on the array and / on the single disk that I had copied the
data from. So what is going on here?


Thanks,
Sam




RAID adds a bit of virtuality to the storage system. Also, RAID adds
some redundancy. So, the information on your drive are spread out.
Have you Googled or Wikipedia-ed the subject? (not really a Debian
question.)


Mark Allums


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
 
Old 03-11-2009, 07:21 PM
Sam Leon
 
Default RAID 5 data structures

Mark Allums wrote:

Sam Leon wrote:
I have a funny question. I have been playing with a 3 disk raid 5
setup for my desktop. I guess I don't fully understand how the
"stripe" is managed or even what it is. I know the stripe is made up
of a chunk from each disk. Now I always thought of the stripe in raid
the same as a block in ext3 or a cluster in ntfs. Meaning if I have a
1k file that I write to an ext3 filesystem with 4k blocks, my 1 k file
will take up one block thus wasting 3k of space.


Now I thought the stripe in raid followed the same principle. Meaning
if I have a 3 disk array with 64k chunks then my data stripe is the
number of disks minus one drive because of parity so in this 3 disk
array I would have a data stripe of 128k. So how much of that space
would my 1k file take up? Would it take up the whole 128k stripe or
just one chunk leaving the other chunk free for something else?


When I migrated my root drive over to the raid5 array I made, I was
afraid that it would use alot more space on the raid array since it is
full of very small files but to my surprise df -h reported the same
values for / on the array and / on the single disk that I had copied
the data from. So what is going on here?


Thanks,
Sam




RAID adds a bit of virtuality to the storage system. Also, RAID adds
some redundancy. So, the information on your drive are spread out. Have
you Googled or Wikipedia-ed the subject? (not really a Debian question.)


Mark Allums





Yes, not a debian question but there are some smart people on here. I
have looked all over google and can't find the specifics that I am
looking for


Sam


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
 
Old 03-12-2009, 12:06 AM
Alex Samad
 
Default RAID 5 data structures

On Wed, Mar 11, 2009 at 07:11:11AM -0500, Sam Leon wrote:
> I have a funny question. I have been playing with a 3 disk raid 5 setup
> for my desktop. I guess I don't fully understand how the "stripe" is
> managed or even what it is. I know the stripe is made up of a chunk
> from each disk. Now I always thought of the stripe in raid the same as
> a block in ext3 or a cluster in ntfs. Meaning if I have a 1k file that
> I write to an ext3 filesystem with 4k blocks, my 1 k file will take up
> one block thus wasting 3k of space.
>
> Now I thought the stripe in raid followed the same principle. Meaning
> if I have a 3 disk array with 64k chunks then my data stripe is the
> number of disks minus one drive because of parity so in this 3 disk
> array I would have a data stripe of 128k. So how much of that space
> would my 1k file take up? Would it take up the whole 128k stripe or
> just one chunk leaving the other chunk free for something else?

I am no expert at this, but my understanding is

the chunk size has to do with the data / parity blocks. The raid system
will work in 64K chunks, so your 1K file will take up 1k of data and
whatever of fs space. but the raid device will have to read write 64K
chunks of information from each disk. so for 3 disks it is 2 x 64k of
data + 1 64K of parity.

>
> When I migrated my root drive over to the raid5 array I made, I was
> afraid that it would use alot more space on the raid array since it is
> full of very small files but to my surprise df -h reported the same
> values for / on the array and / on the single disk that I had copied the
> data from. So what is going on here?
>
> Thanks,
> Sam
>
>
> --
> To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org with a
> subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
>
>

--
"I can't tell you what it's like to be in Europe, for example, to be talking about the greatness of America. But the true greatness of America are the people."

- George W. Bush
07/02/2001
Washington, DC
 
Old 03-12-2009, 09:35 PM
Sam Leon
 
Default RAID 5 data structures

Joe McDonagh wrote:

Sam Leon wrote:
I have a funny question. I have been playing with a 3 disk raid 5
setup for my desktop. I guess I don't fully understand how the
"stripe" is managed or even what it is. I know the stripe is made up
of a chunk from each disk. Now I always thought of the stripe in raid
the same as a block in ext3 or a cluster in ntfs. Meaning if I have a
1k file that I write to an ext3 filesystem with 4k blocks, my 1 k file
will take up one block thus wasting 3k of space.


Now I thought the stripe in raid followed the same principle. Meaning
if I have a 3 disk array with 64k chunks then my data stripe is the
number of disks minus one drive because of parity so in this 3 disk
array I would have a data stripe of 128k. So how much of that space
would my 1k file take up? Would it take up the whole 128k stripe or
just one chunk leaving the other chunk free for something else?


When I migrated my root drive over to the raid5 array I made, I was
afraid that it would use alot more space on the raid array since it is
full of very small files but to my surprise df -h reported the same
values for / on the array and / on the single disk that I had copied
the data from. So what is going on here?


Thanks,
Sam


IIRC, the 64k chunks are transparent to the FS. That's why you're still
using 4k blocks. Strip size, width, chunk size, these terms have
different meanings depending on who you talk to. But, if memory serves,
you're seeing this because the FS is at a different level of abstraction
than RAID.





So multiple files can fit into a chunk and when they need access the
array fetches the whole chunk?


Thanks
Sam


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
 

Thread Tools




All times are GMT. The time now is 02:16 AM.

VBulletin, Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Content Relevant URLs by vBSEO ©2007, Crawlability, Inc.
Copyright 2007 - 2008, www.linux-archive.org