Alternative to RAID... No change to Data HDDs, only one simple "modulo"-Disk

Alternative to RAID... No change to Data HDDs, only one simple "modulo"-Disk

[Markus Müller]

Hi Linux RAID,

I don't want Linux Software RAID, because RAID 5 itself and/or the Linux 
Software RAID has unflexibility I don´t want to accept. These are especialy:

Implementation independet, logical restrictions:
- With only one working disk of a raid 5 I can´t do anything, not to 
mention to mount it alone
- I must trust the RAID algorithmus, because it needs and does do 
writing to handle the data itself
- All hdd must be have the same size, ok, you can put langer in it, but 
you´ll waste space
Linux Software RAID restrictions:
- Putting new Disk in the RAID is BETA and forces to dataloss
<http://dict.leo.org/?p=2Ib6..&search=mention>
The only real advantage is the increased read- and writespeed as well as 
a single partition.

And now the clou: I don´t care about this advantages! So I prefer the 
following solution:
<http://dict.leo.org/?p=2Ib6..&search=advantage>
You use your hdds as you wouldn´t want to use anything like RAID. Then 
you insert a new hdd, which have at least the size of the bigest hdd in 
computer. Then you make over all hdds for each byte, starting on every 
disk at byte 0, a modulo 255, and write it to this new disk. Each read 
access can be done as it would be done before, on every write to one of 
these involved hdds you update over all disks the modulo and write it to 
this new disk. If the disks size differs of the different involved hdds, 
you simply have on the higher bytes a modulo of the remaining disks. For 
example, you have a 60 GB, a 80 GB and a 120 GB disk. The new disk with 
the modulo data is also 120 GBs. Then from 0 byte to 60 GBs you have a 
modulo over 3 disks, from 60 to 80 you have a modulo over 2 disks and 
the rest is a 1:1 copy of the 120 GB drive on this new 120 GB disk.

-> If a disk dies, you can restore it using the safed modulo data on 
this new 120 GB disk.

The benefit? See the restrictions above. None of this applys on this 
solution! But you have the same security benefit from RAID 5! The little 
price for this solution: You loos is speed, and you do not a single 
partition. I have a data dump, not a high performance raid, so I don´t 
care about the speed. (Current hdds are fast enought, also for this 
solution!) And I see multiple partition as a benefit. (personal opinion!)

Special Benefits:
- REAL SECURITY: You can resize your hdd selection without having the 
fear of dataloss! This mechanismus don´t change your source hdds! It 
only does need read access! It CAN´T do anything wrong with your data!
- REAL SECURITY: You can recover with normal rescue programs!
- FLEXIBILITY:: You can take any disk out of your system and simply 
mount it!
- FLEXIBILITY: The size of the involved hdds doesn´t matter; the disk 
with the modulo just must have the same size than the bigest hdd or more 
of your hdd colletion.

And now the matter of price: Does anybody  have an idear how to 
implement this? I am appreciate for any hint or comment.

Thanks,
Markus Mueller
-
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

RE: Alternative to RAID... No change to Data HDDs, only one simple "modulo"-Disk

[Guy]

How do we read about this?
Your links don't seem to do much, just search.  I can't read any articles.

My guess is that you don't know what you are talking about!
Or this is some sort of SPAM for leo.org

"- REAL SECURITY: You can recover with normal rescue programs!"
Does this indicate I can use tar to do backups?  This is not better than
RAID5, RAID1 or RAID6!

I hope I am wrong, sorry if I am.

Guy

-----Original Message-----
From: linux-raid-owner@vger.kernel.org
[mailto:linux-raid-owner@vger.kernel.org] On Behalf Of Markus Müller
Sent: Friday, June 18, 2004 12:13 PM
To: linux-raid@vger.kernel.org
Subject: Alternative to RAID... No change to Data HDDs, only one simple
"modulo"-Disk

Hi Linux RAID,

I don't want Linux Software RAID, because RAID 5 itself and/or the Linux 
Software RAID has unflexibility I don´t want to accept. These are especialy:

Implementation independet, logical restrictions:
- With only one working disk of a raid 5 I can´t do anything, not to 
mention to mount it alone
- I must trust the RAID algorithmus, because it needs and does do 
writing to handle the data itself
- All hdd must be have the same size, ok, you can put langer in it, but 
you´ll waste space
Linux Software RAID restrictions:
- Putting new Disk in the RAID is BETA and forces to dataloss
<http://dict.leo.org/?p=2Ib6..&search=mention>
The only real advantage is the increased read- and writespeed as well as 
a single partition.

And now the clou: I don´t care about this advantages! So I prefer the 
following solution:
<http://dict.leo.org/?p=2Ib6..&search=advantage>
You use your hdds as you wouldn´t want to use anything like RAID. Then 
you insert a new hdd, which have at least the size of the bigest hdd in 
computer. Then you make over all hdds for each byte, starting on every 
disk at byte 0, a modulo 255, and write it to this new disk. Each read 
access can be done as it would be done before, on every write to one of 
these involved hdds you update over all disks the modulo and write it to 
this new disk. If the disks size differs of the different involved hdds, 
you simply have on the higher bytes a modulo of the remaining disks. For 
example, you have a 60 GB, a 80 GB and a 120 GB disk. The new disk with 
the modulo data is also 120 GBs. Then from 0 byte to 60 GBs you have a 
modulo over 3 disks, from 60 to 80 you have a modulo over 2 disks and 
the rest is a 1:1 copy of the 120 GB drive on this new 120 GB disk.

-> If a disk dies, you can restore it using the safed modulo data on 
this new 120 GB disk.

The benefit? See the restrictions above. None of this applys on this 
solution! But you have the same security benefit from RAID 5! The little 
price for this solution: You loos is speed, and you do not a single 
partition. I have a data dump, not a high performance raid, so I don´t 
care about the speed. (Current hdds are fast enought, also for this 
solution!) And I see multiple partition as a benefit. (personal opinion!)

Special Benefits:
- REAL SECURITY: You can resize your hdd selection without having the 
fear of dataloss! This mechanismus don´t change your source hdds! It 
only does need read access! It CAN´T do anything wrong with your data!
- REAL SECURITY: You can recover with normal rescue programs!
- FLEXIBILITY:: You can take any disk out of your system and simply 
mount it!
- FLEXIBILITY: The size of the involved hdds doesn´t matter; the disk 
with the modulo just must have the same size than the bigest hdd or more 
of your hdd colletion.

And now the matter of price: Does anybody  have an idear how to 
implement this? I am appreciate for any hint or comment.

Thanks,
Markus Mueller
-
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html


-
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

RE: Alternative to RAID... No change to Data HDDs, only one simple "modulo"-Disk

[Mark Hahn]

> How do we read about this?

afaikt, he's just looking for an incremental raid system,
which is understandably attractive.  keeping track of which 
blocks are on N disks and which have been migrated to N+1
might be tricky, though.  certainly not impossible.

I often wonder whether it's sensible to consider putting 
raid-like reliability features into the filesystem, rather than
restricting ourselves to block-level raid...

Re: Alternative to RAID... No change to Data HDDs, only one simple "modulo"-Disk

[Lars Marowsky-Bree]

On 2004-06-21T09:22:41,
   Mark Hahn <hahn@physics.mcmaster.ca> said:

> afaikt, he's just looking for an incremental raid system,
> which is understandably attractive.  keeping track of which 
> blocks are on N disks and which have been migrated to N+1
> might be tricky, though.  certainly not impossible.

It's the volume manager's job to do that. I can easily allocate blocks
from all disks for whatever RAID level you want etc. I believe EVMS can
actually do that, even. If not, it should be added ;-)


Sincerely,
    Lars Marowsky-Brée <lmb@suse.de>

-- 
High Availability & Clustering	      \ ever tried. ever failed. no matter.
SUSE Labs			      | try again. fail again. fail better.
Research & Development, SUSE LINUX AG \ 	-- Samuel Beckett

-
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html