wish for Linux MD mirrored raid types

wish for Linux MD mirrored raid types

[Keld Jørn Simonsen]

Hi List

based on the recent discussion, that showed lacking knowledge
on Linux MD RAID10 features, I have some thoughts:

It is really hard to disseminate information on "new" features
in MD RAID. RAID10 has been in the kernel since 2.6.9 - from 2004.
I have tried to give info on RAID10 at a number of web pages,
and still many people, even on our linux-raid list are not aware
of it.

Also many people are confused about Linux MD raid10 and RAID1+0.

So I think we shopuld rather name things in another way.

I would like linux MD raid10 functionality to be part of the Linux MD
RAID1 module, and be called raid1. This is in accordance with the
use of the RAID1 term as standadized by SNIA. In fact the RAID10-offset
layout is an implementation of a SNIA RAID specification. The RAID10-near
layout is an implementation of a simple RAID layout. And the RAID10-far
layout is just another layout far a mirrored RAID.  So all these types
could just be defined as different RAID1 layouts.

I would then also wish for RAID10-far to be the default RAID1
layout. There is general agreement on this list that RAID10-far
is the best mirrored layout for most purposes. In the interest of giving
the best performance to the Linux RAID users, we should make
the defaults the best practise - users tend to choose defaults,
especially often they do not have much knowledge. 

We could still keep the RAID10 code for backwards compatibility,
or even let this new naming just be calls to the raid10 code
from the raid1 module.

best regards
Keld

Re: wish for Linux MD mirrored raid types

[Roman Mamedov]

[-- Attachment #1: Type: text/plain, Size: 765 bytes --]

On Fri, 6 May 2011 09:17:52 +0200
Keld Jørn Simonsen <keld@keldix.com> wrote:

> I would like linux MD raid10 functionality to be part of the Linux MD
> RAID1 module, and be called raid1. This is in accordance with the
> use of the RAID1 term as standadized by SNIA. In fact the RAID10-offset
> layout is an implementation of a SNIA RAID specification. The RAID10-near
> layout is an implementation of a simple RAID layout. And the RAID10-far
> layout is just another layout far a mirrored RAID.  So all these types
> could just be defined as different RAID1 layouts.

RAID1 is RAID1, RAID10 is RAID10.
RAID1 on 4 drives is very different from RAID10 on 4 drives.
Don't add confusion by trying to rename RAID10 to RAID1.

-- 
With respect,
Roman

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

Re: wish for Linux MD mirrored raid types

[David Brown]

On 06/05/2011 09:17, Keld Jørn Simonsen wrote:
> Hi List
>
> based on the recent discussion, that showed lacking knowledge
> on Linux MD RAID10 features, I have some thoughts:
>
> It is really hard to disseminate information on "new" features
> in MD RAID. RAID10 has been in the kernel since 2.6.9 - from 2004.
> I have tried to give info on RAID10 at a number of web pages,
> and still many people, even on our linux-raid list are not aware
> of it.
>
> Also many people are confused about Linux MD raid10 and RAID1+0.
>
> So I think we shopuld rather name things in another way.
>
> I would like linux MD raid10 functionality to be part of the Linux MD
> RAID1 module, and be called raid1. This is in accordance with the
> use of the RAID1 term as standadized by SNIA. In fact the RAID10-offset
> layout is an implementation of a SNIA RAID specification. The RAID10-near
> layout is an implementation of a simple RAID layout. And the RAID10-far
> layout is just another layout far a mirrored RAID.  So all these types
> could just be defined as different RAID1 layouts.
>
> I would then also wish for RAID10-far to be the default RAID1
> layout. There is general agreement on this list that RAID10-far
> is the best mirrored layout for most purposes. In the interest of giving
> the best performance to the Linux RAID users, we should make
> the defaults the best practise - users tend to choose defaults,
> especially often they do not have much knowledge.
>
> We could still keep the RAID10 code for backwards compatibility,
> or even let this new naming just be calls to the raid10 code
> from the raid1 module.
>

I mostly agree with you, but for a few points:

You say the various raid10 layouts match SNIA RAID1 specifications and 
layouts - does that also apply if you have more than two disks?  And 
what about weird things like raid10 over three disks?

There are times when it is important that a standard raid1 element is 
also accessible as a normal disk (with metadata 0.90).  Examples include 
booting and sometimes data recovery or transferring the disk to another 
machine.

There are things you can do with raid1 that you cannot do with raid10 at 
the moment, such as re-shaping and re-sizing.  It wouldn't make sense to 
classify raid10 as a type of raid1 until it also has this capability.

It is also not clear how adding an extra drive to a raid10,far layout 
should work.  If you add an extra drive to a raid1 set, you get a 
three-way mirror.  If you add an extra drive to a raid10,far drive, 
should it directly mirror one of the existing drives?  Should it reshape 
to a raid10,far3 arrangement?  Should it turn into a raid10,far2 and 
re-balance across the disks?  Any of these might be valid choices.


I support your "campaign for raid10,far awareness", but I'm not sure 
that making it the default for raid1 is appropriate at the moment.  Once 
raid10 re-shapes and re-sizes are fully supported, and once all main 
distros have moved to a new enough version of grub (which supports 
booting from raid10, and different metadata versions), then there will 
be few reasons for anyone to choose "standard" raid1 rather than raid10,far.


What would make a bigger difference is to get better raid support into 
the distro's installers.  Most of these, when faced with two disks, will 
just ask you which drive you want to use - if the support raid at all 
during installation, it is accessed through "advanced" and "manual 
partition" screens.  And getting grub onto both disks is very much a 
post-install manual operation, for those that know that it needs to be 
done.  Getting distro installers to set up raid10,far by default would 
be a much bigger step.


--
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: wish for Linux MD mirrored raid types

[Keld Jørn Simonsen]

On Fri, May 06, 2011 at 01:31:59PM +0600, Roman Mamedov wrote:
> On Fri, 6 May 2011 09:17:52 +0200
> Keld Jørn Simonsen <keld@keldix.com> wrote:
> 
> > I would like linux MD raid10 functionality to be part of the Linux MD
> > RAID1 module, and be called raid1. This is in accordance with the
> > use of the RAID1 term as standadized by SNIA. In fact the RAID10-offset
> > layout is an implementation of a SNIA RAID specification. The RAID10-near
> > layout is an implementation of a simple RAID layout. And the RAID10-far
> > layout is just another layout far a mirrored RAID.  So all these types
> > could just be defined as different RAID1 layouts.
> 
> RAID1 is RAID1, RAID10 is RAID10.
> RAID1 on 4 drives is very different from RAID10 on 4 drives.
> Don't add confusion by trying to rename RAID10 to RAID1.

How are they different?
Say what is the difference between a Linux MD RAID1 with 4 disks, and
the default Linux MD RAID10 with 4 disks? (in the near layout)?

Best regards
keld
--
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: wish for Linux MD mirrored raid types

[Jonathan Tripathy]

On 06/05/2011 10:03, Keld Jørn Simonsen wrote:
> On Fri, May 06, 2011 at 01:31:59PM +0600, Roman Mamedov wrote:
>    
>> On Fri, 6 May 2011 09:17:52 +0200
>> Keld Jørn Simonsen<keld@keldix.com>  wrote:
>>
>>      
>>> I would like linux MD raid10 functionality to be part of the Linux MD
>>> RAID1 module, and be called raid1. This is in accordance with the
>>> use of the RAID1 term as standadized by SNIA. In fact the RAID10-offset
>>> layout is an implementation of a SNIA RAID specification. The RAID10-near
>>> layout is an implementation of a simple RAID layout. And the RAID10-far
>>> layout is just another layout far a mirrored RAID.  So all these types
>>> could just be defined as different RAID1 layouts.
>>>        
>> RAID1 is RAID1, RAID10 is RAID10.
>> RAID1 on 4 drives is very different from RAID10 on 4 drives.
>> Don't add confusion by trying to rename RAID10 to RAID1.
>>      
> How are they different?
> Say what is the difference between a Linux MD RAID1 with 4 disks, and
> the default Linux MD RAID10 with 4 disks? (in the near layout)?
>
>    
RAID1 is traditionally a mirror only setup (ok, some RAID 
implementations may do some load-balancing of some sort). So a RAID1 
with 4 disks is one data set copied onto 4 disks. Bandwidth is roughly 
the same as a single disk (ignoring any load balancing).
RAID10 is mirror and stripe. A RAID10 with 4 disks is similar to a 2 
disk RAID0 (double bandwidth with data split in half across both disks), 
but with each disk having a mirror (which brings the total up to 4 drives).

Additionally, a RAID1 disk (at least using MD) can be accessed just like 
a normal disk (good for recovery etc.) however a single disk out of a 
RAID10 array is next to useless.
--
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: wish for Linux MD mirrored raid types

[Keld Jørn Simonsen]

On Fri, May 06, 2011 at 09:51:45AM +0200, David Brown wrote:
> On 06/05/2011 09:17, Keld Jørn Simonsen wrote:
> >Hi List
> >
> >based on the recent discussion, that showed lacking knowledge
> >on Linux MD RAID10 features, I have some thoughts:
> >
> >It is really hard to disseminate information on "new" features
> >in MD RAID. RAID10 has been in the kernel since 2.6.9 - from 2004.
> >I have tried to give info on RAID10 at a number of web pages,
> >and still many people, even on our linux-raid list are not aware
> >of it.
> >
> >Also many people are confused about Linux MD raid10 and RAID1+0.
> >
> >So I think we shopuld rather name things in another way.
> >
> >I would like linux MD raid10 functionality to be part of the Linux MD
> >RAID1 module, and be called raid1. This is in accordance with the
> >use of the RAID1 term as standadized by SNIA. In fact the RAID10-offset
> >layout is an implementation of a SNIA RAID specification. The RAID10-near
> >layout is an implementation of a simple RAID layout. And the RAID10-far
> >layout is just another layout far a mirrored RAID.  So all these types
> >could just be defined as different RAID1 layouts.
> >
> >I would then also wish for RAID10-far to be the default RAID1
> >layout. There is general agreement on this list that RAID10-far
> >is the best mirrored layout for most purposes. In the interest of giving
> >the best performance to the Linux RAID users, we should make
> >the defaults the best practise - users tend to choose defaults,
> >especially often they do not have much knowledge.
> >
> >We could still keep the RAID10 code for backwards compatibility,
> >or even let this new naming just be calls to the raid10 code
> >from the raid1 module.
> >
> 
> I mostly agree with you, but for a few points:
> 
> You say the various raid10 layouts match SNIA RAID1 specifications and 
> layouts - does that also apply if you have more than two disks?  And 
> what about weird things like raid10 over three disks?

I don't know. But anyway, you can see it as an extension of SNIA
standard RAID1 layouts. And then if the SNIA standard does not cover it,
the standard should be enhanced. I have sent specs on the far layout to
SNIA twice, but witout any response.

> There are times when it is important that a standard raid1 element is 
> also accessible as a normal disk (with metadata 0.90).  Examples include 
> booting and sometimes data recovery or transferring the disk to another 
> machine.

yes, and then we should document how this is done. Also boot loaders are
becoming more intelligent, so one common case where the raid also need
to be seen as a normal disk is going away.

> There are things you can do with raid1 that you cannot do with raid10 at 
> the moment, such as re-shaping and re-sizing.  It wouldn't make sense to 
> classify raid10 as a type of raid1 until it also has this capability.

Well, theoretically Linux MD RAID10 is a mirrored RAID type - that is a
RAID1 in standard parlance. You could document that some RAID1 layouts
cannot be re-shaped or re-sized.  I generally like to be backward
compatible, and I think this can be esily done just by naming the
different layouts of RAID1, for example: classic, near, far, offset,
directly corresponding to the functionalities and possibly kernel
modules that we have today.


> It is also not clear how adding an extra drive to a raid10,far layout 
> should work.  If you add an extra drive to a raid1 set, you get a 
> three-way mirror. 

You need to be clear on how you add disks - if you want more copies, or
more space with the same number of copies. I think the default would 
be to add more space. It is quite uncommon to want more than 2 copies,
while a need for more space happens all the time.


> If you add an extra drive to a raid10,far drive, 
> should it directly mirror one of the existing drives?  Should it reshape 
> to a raid10,far3 arrangement?  Should it turn into a raid10,far2 and 
> re-balance across the disks?  Any of these might be valid choices.

Agree that any of these are valid. And thus all of them should be doable.
But implementation will most likely be piece wise. I see no problem with
architecture there. You can always say that some features are
implemented, and some are not. That is at least the current situation,
and this is a very normal situation.

> I support your "campaign for raid10,far awareness", but I'm not sure 
> that making it the default for raid1 is appropriate at the moment.  Once 
> raid10 re-shapes and re-sizes are fully supported, and once all main 
> distros have moved to a new enough version of grub (which supports 
> booting from raid10, and different metadata versions), then there will 
> be few reasons for anyone to choose "standard" raid1 rather than raid10,far.

Everything takes time. But we can set a goal, and then plan for
reaching the goal. I am sure if we do it right, we can get there in
a time less than what we have had since the introduction of the raid10
driver in 2004.

> just ask you which drive you want to use - if the support raid at all 
> during installation, it is accessed through "advanced" and "manual 
> partition" screens.  And getting grub onto both disks is very much a 
> post-install manual operation, for those that know that it needs to be 
> done.  Getting distro installers to set up raid10,far by default would 
> be a much bigger step.

I would like it to be done in distro installers, with the help of grub
et al. And work is underway in some distros to do that.

Best regards
keld
--
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: wish for Linux MD mirrored raid types

[Keld Jørn Simonsen]

On Fri, May 06, 2011 at 10:22:33AM +0100, Jonathan Tripathy wrote:
> 
> On 06/05/2011 10:03, Keld Jørn Simonsen wrote:
> >On Fri, May 06, 2011 at 01:31:59PM +0600, Roman Mamedov wrote:
> >   
> >>On Fri, 6 May 2011 09:17:52 +0200
> >>Keld Jørn Simonsen<keld@keldix.com>  wrote:
> >>
> >>     
> >>>I would like linux MD raid10 functionality to be part of the Linux MD
> >>>RAID1 module, and be called raid1. This is in accordance with the
> >>>use of the RAID1 term as standadized by SNIA. In fact the RAID10-offset
> >>>layout is an implementation of a SNIA RAID specification. The RAID10-near
> >>>layout is an implementation of a simple RAID layout. And the RAID10-far
> >>>layout is just another layout far a mirrored RAID.  So all these types
> >>>could just be defined as different RAID1 layouts.
> >>>       
> >>RAID1 is RAID1, RAID10 is RAID10.
> >>RAID1 on 4 drives is very different from RAID10 on 4 drives.
> >>Don't add confusion by trying to rename RAID10 to RAID1.
> >>     
> >How are they different?
> >Say what is the difference between a Linux MD RAID1 with 4 disks, and
> >the default Linux MD RAID10 with 4 disks? (in the near layout)?
> >
> >   
> RAID1 is traditionally a mirror only setup (ok, some RAID 
> implementations may do some load-balancing of some sort). So a RAID1 
> with 4 disks is one data set copied onto 4 disks. Bandwidth is roughly 
> the same as a single disk (ignoring any load balancing).
> RAID10 is mirror and stripe. A RAID10 with 4 disks is similar to a 2 
> disk RAID0 (double bandwidth with data split in half across both disks), 
> but with each disk having a mirror (which brings the total up to 4 drives).
> 
> Additionally, a RAID1 disk (at least using MD) can be accessed just like 
> a normal disk (good for recovery etc.) however a single disk out of a 
> RAID10 array is next to useless.

I think you are demonstrating some of my points about general knowledge
quite nicely. Don't worry, you are not alone.

best regards
keld
--
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: wish for Linux MD mirrored raid types

[Roman Mamedov]

[-- Attachment #1: Type: text/plain, Size: 1358 bytes --]

On Fri, 6 May 2011 11:41:02 +0200
Keld Jørn Simonsen <keld@keldix.com> wrote:
   
> > RAID1 is traditionally a mirror only setup (ok, some RAID 
> > implementations may do some load-balancing of some sort). So a RAID1 
> > with 4 disks is one data set copied onto 4 disks. Bandwidth is roughly 
> > the same as a single disk (ignoring any load balancing).
> > RAID10 is mirror and stripe. A RAID10 with 4 disks is similar to a 2 
> > disk RAID0 (double bandwidth with data split in half across both disks), 
> > but with each disk having a mirror (which brings the total up to 4 drives).
> > 
> > Additionally, a RAID1 disk (at least using MD) can be accessed just like 
> > a normal disk (good for recovery etc.) however a single disk out of a 
> > RAID10 array is next to useless.
> 
> I think you are demonstrating some of my points about general knowledge
> quite nicely. Don't worry, you are not alone.

So what is your proposal: people do not know they can do X, but commonly do Y,
so let's rename X to Y so that both things are called Y and they 'know how to
do it'?

RAID10 is an established term and means "stripe of mirrors", period.
RAID1 means just a mirror. All the rest is just implementation details.
Trying to redefine 2x2 to be 5 as some 'educational project' is definitely
misguided.

-- 
With respect,
Roman

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

Re: wish for Linux MD mirrored raid types

[Jonathan Tripathy]

>>> RAID1 is traditionally a mirror only setup (ok, some RAID
>>> implementations may do some load-balancing of some sort). So a RAID1
>>> with 4 disks is one data set copied onto 4 disks. Bandwidth is roughly
>>> the same as a single disk (ignoring any load balancing).
>>> RAID10 is mirror and stripe. A RAID10 with 4 disks is similar to a 2
>>> disk RAID0 (double bandwidth with data split in half across both disks),
>>> but with each disk having a mirror (which brings the total up to 4 drives).
>>>
>>> Additionally, a RAID1 disk (at least using MD) can be accessed just like
>>> a normal disk (good for recovery etc.) however a single disk out of a
>>> RAID10 array is next to useless.
>>>        
>>      
Just so I can sleep at night, is my understanding of RAID10 and RAID1 
above correct?

Re: wish for Linux MD mirrored raid types

[David Brown]

On 06/05/2011 12:05, Jonathan Tripathy wrote:
>
>>>> RAID1 is traditionally a mirror only setup (ok, some RAID
>>>> implementations may do some load-balancing of some sort). So a RAID1
>>>> with 4 disks is one data set copied onto 4 disks. Bandwidth is roughly
>>>> the same as a single disk (ignoring any load balancing).
>>>> RAID10 is mirror and stripe. A RAID10 with 4 disks is similar to a 2
>>>> disk RAID0 (double bandwidth with data split in half across both
>>>> disks),
>>>> but with each disk having a mirror (which brings the total up to 4
>>>> drives).
>>>>
>>>> Additionally, a RAID1 disk (at least using MD) can be accessed just
>>>> like
>>>> a normal disk (good for recovery etc.) however a single disk out of a
>>>> RAID10 array is next to useless.
> Just so I can sleep at night, is my understanding of RAID10 and RAID1
> above correct?
>

It's mostly right (assuming, of course, that /I/ am correct here...).

RAID1 is traditionally a two-way mirror on two disks (or possibly on two 
other raid sets, as in raid0+1).  So if you talk about RAID1 with 4 
disks, you should probably qualify it more precisely - otherwise people 
will wonder what you mean, or think it is impossible (many other RAID1 
solutions, hardware or software, don't support more than two-way 
mirrors).  /I/ would certainly say that a 4-disk RAID1 is a four-way 
mirror as you described - but some people might think of a standard 
layout RAID1+0.

As you say, RAID10,near on four disks is pretty much identical to 
RAID1+0 - i.e., a stripe of two normal RAID1 pairs.

A single md RAID1 disk can be accessed like a normal disk, /if/ it uses 
metadata format 0.90 which is put at the end of the drive.  If you have 
later metadata formats that are at the beginning, then that will cause 
trouble if you try to view the disk without using md.  A single disk 
from a RAID10 is, as you say, useless without md.

However, assuming your recovery PC supports md raid, then you can 
assembly your single RAID10 disk as a degraded RAID10 array.  After all, 
it wouldn't be very redundant if you only had access to your data when 
all the disks were working!

Re: wish for Linux MD mirrored raid types

[Keld Jørn Simonsen]

On Fri, May 06, 2011 at 03:50:59PM +0600, Roman Mamedov wrote:
> On Fri, 6 May 2011 11:41:02 +0200
> Keld Jørn Simonsen <keld@keldix.com> wrote:
>    
> > > RAID1 is traditionally a mirror only setup (ok, some RAID 
> > > implementations may do some load-balancing of some sort). So a RAID1 
> > > with 4 disks is one data set copied onto 4 disks. Bandwidth is roughly 
> > > the same as a single disk (ignoring any load balancing).
> > > RAID10 is mirror and stripe. A RAID10 with 4 disks is similar to a 2 
> > > disk RAID0 (double bandwidth with data split in half across both disks), 
> > > but with each disk having a mirror (which brings the total up to 4 drives).
> > > 
> > > Additionally, a RAID1 disk (at least using MD) can be accessed just like 
> > > a normal disk (good for recovery etc.) however a single disk out of a 
> > > RAID10 array is next to useless.
> > 
> > I think you are demonstrating some of my points about general knowledge
> > quite nicely. Don't worry, you are not alone.
> 
> So what is your proposal: people do not know they can do X, but commonly do Y,
> so let's rename X to Y so that both things are called Y and they 'know how to
> do it'?
> 
> RAID10 is an established term and means "stripe of mirrors", period.
> RAID1 means just a mirror. All the rest is just implementation details.
> Trying to redefine 2x2 to be 5 as some 'educational project' is definitely
> misguided.

This is the common understanding: RAID10 means RAID1+0 - a RAID0 over
two RAID1's. As you write. But in Linux MD, raid10 is something else.
Something which is characterized as RAID1 in the SNIA RAID standards.
raid10,offset is directly an implementation of one of the SNIA RAID1
variants.

I am proposing that we call our raid10 layout types for RAID1 - in
accordance with SNIA standards, and in accordance with the common
understanding as demonstrated by you and others, even here on the
linux-raid kernel mailing list.

Best regards
keld
--
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: wish for Linux MD mirrored raid types

[Mikael Abrahamsson]

[-- Attachment #1: Type: TEXT/PLAIN, Size: 510 bytes --]

On Fri, 6 May 2011, Keld Jørn Simonsen wrote:

> I am proposing that we call our raid10 layout types for RAID1 - in 
> accordance with SNIA standards, and in accordance with the common 
> understanding as demonstrated by you and others, even here on the 
> linux-raid kernel mailing list.

Just so we're clear, if I create a raid1 with 4 drives I want 4 identical 
copies of the same information, not RAID1+0.

How should this be handled with your suggestion?

-- 
Mikael Abrahamsson    email: swmike@swm.pp.se

Re: wish for Linux MD mirrored raid types

[Keld Jørn Simonsen]

On Fri, May 06, 2011 at 12:54:28PM +0200, David Brown wrote:
> On 06/05/2011 12:05, Jonathan Tripathy wrote:
> >
> >>>>RAID1 is traditionally a mirror only setup (ok, some RAID
> >>>>implementations may do some load-balancing of some sort). So a RAID1
> >>>>with 4 disks is one data set copied onto 4 disks. Bandwidth is roughly
> >>>>the same as a single disk (ignoring any load balancing).
> >>>>RAID10 is mirror and stripe. A RAID10 with 4 disks is similar to a 2
> >>>>disk RAID0 (double bandwidth with data split in half across both
> >>>>disks),
> >>>>but with each disk having a mirror (which brings the total up to 4
> >>>>drives).
> >>>>
> >>>>Additionally, a RAID1 disk (at least using MD) can be accessed just
> >>>>like
> >>>>a normal disk (good for recovery etc.) however a single disk out of a
> >>>>RAID10 array is next to useless.
> >Just so I can sleep at night, is my understanding of RAID10 and RAID1
> >above correct?
> >
> 
> It's mostly right (assuming, of course, that /I/ am correct here...).
> 
> RAID1 is traditionally a two-way mirror on two disks (or possibly on two 
> other raid sets, as in raid0+1).  So if you talk about RAID1 with 4 
> disks, you should probably qualify it more precisely - otherwise people 
> will wonder what you mean, or think it is impossible (many other RAID1 
> solutions, hardware or software, don't support more than two-way 
> mirrors).  /I/ would certainly say that a 4-disk RAID1 is a four-way 
> mirror as you described - but some people might think of a standard 
> layout RAID1+0.
> 
> As you say, RAID10,near on four disks is pretty much identical to 
> RAID1+0 - i.e., a stripe of two normal RAID1 pairs.
> 
> A single md RAID1 disk can be accessed like a normal disk, /if/ it uses 
> metadata format 0.90 which is put at the end of the drive.  If you have 
> later metadata formats that are at the beginning, then that will cause 
> trouble if you try to view the disk without using md.  A single disk 
> from a RAID10 is, as you say, useless without md.
> 
> However, assuming your recovery PC supports md raid, then you can 
> assembly your single RAID10 disk as a degraded RAID10 array.  After all, 
> it wouldn't be very redundant if you only had access to your data when 
> all the disks were working!

I agree with what David Brown says here.

Just some further remarks:

You can have Linux MD raid10 arrays with only 2 disks. 
This is different from RAID1+0 that requires at least 4 disks.

You can also with Linux MD have an odd number of disk, which is
impossible with RAID1+0.

Linux MD raid10 is a kernel driver that handles everything in
one module, while RAID1+0 is handled in two drivers - raid0 over raid1.
The latter is called nested RAID, and the Linux MD raid10 is not a
nested RAID, while RAID1+0 is.

With Linux MD raid10,near and superblock 0.90 you can also 
boot it from old grub, lilo etc, as each partition can be seen as a normal
file system.

As David pointed out, "a single disk out of a RAID10 array is next to
useless" is not true. Linux MD RAID10 is designed to always be able to 
hold all data intact if one disk is out. You just need Linux MD raid10
aware software. 

I apologise for being abrupt in my first answer. But the thing is that
I am getting tired of explaining and correcting all these misconceptions
again and again. And my proposal is an attempt to stop at least some
of the need for explaining, while also bringing the best practise forward to 
the users of our technology, for the benefit of the users.

best regards
keld
and then 

Re: wish for Linux MD mirrored raid types

[Keld Jørn Simonsen]

On Fri, May 06, 2011 at 03:26:52PM +0200, Mikael Abrahamsson wrote:
> On Fri, 6 May 2011, Keld Jørn Simonsen wrote:
> 
> >I am proposing that we call our raid10 layout types for RAID1 - in 
> >accordance with SNIA standards, and in accordance with the common 
> >understanding as demonstrated by you and others, even here on the 
> >linux-raid kernel mailing list.
> 
> Just so we're clear, if I create a raid1 with 4 drives I want 4 identical 
> copies of the same information, not RAID1+0.
> 
> How should this be handled with your suggestion?

With parameters like the ones that we use today for raid10. Something
like

mdadm --create /dev/md0 --level=1 -n 4 --layout=c4 /dev/sd[abcd]1 

assuming a new layout type "classic" abbreviated as "c". "n4" could also
do for the layout value.

best regards
keld
--
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: wish for Linux MD mirrored raid types

[Miles Fidelman]

Keld Jørn Simonsen wrote:
>> As you say, RAID10,near on four disks is pretty much identical to
>> RAID1+0 - i.e., a stripe of two normal RAID1 pairs.
>>      

I don't that's exactly right.  At least as I understand it:

- RAID1+0 (and RAID0+1) nests things - you start with two sets of RAID1 
mirrors, then stripe across them (or vice versa) - it's a nested set of 
steps

- md RAID10 provides both mirroring and striping, but it's a more 
integrated function - (from the man page) "RAID10 provides a combination 
of RAID1 and RAID0, and sometimes known as RAID1+0. Every datablock is 
duplicated some number of times, and the resulting collection of 
datablocks are distributed over multiple drives." - but there isn't an 
inherent nesting in the process (i.e., no two disks are copies of each 
other, and md RAID10 will work over odd numbers of drives)

Miles Fidelman




-- 
In theory, there is no difference between theory and practice.
In<fnord>  practice, there is.   .... Yogi Berra


--
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: wish for Linux MD mirrored raid types

[Keld Jørn Simonsen]

On Fri, May 06, 2011 at 10:01:48AM -0400, Miles Fidelman wrote:
> Keld Jørn Simonsen wrote:
> >>As you say, RAID10,near on four disks is pretty much identical to
> >>RAID1+0 - i.e., a stripe of two normal RAID1 pairs.
> >>     
> 
> I don't that's exactly right.  At least as I understand it:
> 
> - RAID1+0 (and RAID0+1) nests things - you start with two sets of RAID1 
> mirrors, then stripe across them (or vice versa) - it's a nested set of 
> steps
> 
> - md RAID10 provides both mirroring and striping, but it's a more 
> integrated function - (from the man page) "RAID10 provides a combination 
> of RAID1 and RAID0, and sometimes known as RAID1+0. Every datablock is 
> duplicated some number of times, and the resulting collection of 
> datablocks are distributed over multiple drives." - but there isn't an 
> inherent nesting in the process (i.e., no two disks are copies of each 
> other, and md RAID10 will work over odd numbers of drives)

Yes, you are right, RAID1+0 is nested, while Linux MD raid10 is not.
But the data layout of Linux MD RAID1+0 and Linux MD
RAID10,near is almost identical.

keld
--
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: wish for Linux MD mirrored raid types

[Roberto Spadim]

hum, i asked this question one time, the point is:
raid1 code is very easy
raid10 code is more complex

easy = faster, less memory, less cpu
complex = faster?, more memory? more cpu?

check others raid system (freebsd, netbsd) and check how they do...

2011/5/6 Keld Jørn Simonsen <keld@keldix.com>:
> On Fri, May 06, 2011 at 10:01:48AM -0400, Miles Fidelman wrote:
>> Keld Jørn Simonsen wrote:
>> >>As you say, RAID10,near on four disks is pretty much identical to
>> >>RAID1+0 - i.e., a stripe of two normal RAID1 pairs.
>> >>
>>
>> I don't that's exactly right.  At least as I understand it:
>>
>> - RAID1+0 (and RAID0+1) nests things - you start with two sets of RAID1
>> mirrors, then stripe across them (or vice versa) - it's a nested set of
>> steps
>>
>> - md RAID10 provides both mirroring and striping, but it's a more
>> integrated function - (from the man page) "RAID10 provides a combination
>> of RAID1 and RAID0, and sometimes known as RAID1+0. Every datablock is
>> duplicated some number of times, and the resulting collection of
>> datablocks are distributed over multiple drives." - but there isn't an
>> inherent nesting in the process (i.e., no two disks are copies of each
>> other, and md RAID10 will work over odd numbers of drives)
>
> Yes, you are right, RAID1+0 is nested, while Linux MD raid10 is not.
> But the data layout of Linux MD RAID1+0 and Linux MD
> RAID10,near is almost identical.
>
> keld
> --
> 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
>



-- 
Roberto Spadim
Spadim Technology / SPAEmpresarial
--
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: wish for Linux MD mirrored raid types

[Miles Fidelman]

Roberto Spadim wrote:
> hum, i asked this question one time, the point is:
> raid1 code is very easy
> raid10 code is more complex
>
> easy = faster, less memory, less cpu
> complex = faster?, more memory? more cpu?
>
> check others raid system (freebsd, netbsd) and check how they do...
>
> 2011/5/6 Keld Jørn Simonsen<keld@keldix.com>:
>    
>>> Keld Jørn Simonsen wrote:
>>>        
>>>>> As you say, RAID10,near on four disks is pretty much identical to
>>>>> RAID1+0 - i.e., a stripe of two normal RAID1 pairs.
>>>>>
>>>>>            

of course RAID10 offers the far option, and the option to run on an odd 
number of drives (I currently use it on a couple of 4-drive servers, but 
as I  look at some replacement hardware I've been thinking that RAID10 
on a 5-drive configuration offers a nice mix of reliability and performance)

re. easy/complex - I'm not sure I really believe that - when I've 
reviewed options, I come to the conclusion that:

RAID1 - wastes a lot of drive space, particularly if you want to 
maintain reliability after a single-drive failure (requires a minimum of 
3 mirrored drives)

RAID5 and RAID6 are better when it comes to mixing efficiency with 
multi-drive failures, but have a couple of odd failure modes and are a 
real pain to rebuild after a failure

RAID1+0 and RAID0+1 are interesting combination - but my sense is that 
every time you nest a layer, you're adding configuration complexity, 
processing delays (low level cpu cycles and i/o transactions), and just 
that much more complexity if you have to reconfigure or rebuild and 
array (particularly if you're running LVM and DRBD on top of the basic 
disk arrays, as I am)

md RAID10 chops out a layer of nesting and makes better use of disk 
space - by combining block replication and striping into a single layer 
- providing what, to me, is a good balance of disk use, multiple copies 
of data, performance, and managability - so far, its worked well for me



-- 
In theory, there is no difference between theory and practice.
In<fnord>  practice, there is.   .... Yogi Berra


--
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: wish for Linux MD mirrored raid types

[Keld Jørn Simonsen]

On Fri, May 06, 2011 at 12:34:36PM -0300, Roberto Spadim wrote:
> hum, i asked this question one time, the point is:
> raid1 code is very easy
> raid10 code is more complex
> 
> easy = faster, less memory, less cpu
> complex = faster?, more memory? more cpu?


I think more complex -> more intelligent, more features.

I think it is actually amazing what raid10 can do.
Is raid 1 in other systems really limited to say 2 disks
- mirrored? Then linux raid10 is much more intelligent. 
> 
> check others raid system (freebsd, netbsd) and check how they do...

Yes, and also HW raid. If Linux made raid10,far the default for RAID1,
then I think Linux would compare very well with other operating systems
and HW raid.

Best regards
keld

----

> 2011/5/6 Keld Jørn Simonsen <keld@keldix.com>:
> > On Fri, May 06, 2011 at 10:01:48AM -0400, Miles Fidelman wrote:
> >> Keld Jørn Simonsen wrote:
> >> >>As you say, RAID10,near on four disks is pretty much identical to
> >> >>RAID1+0 - i.e., a stripe of two normal RAID1 pairs.
> >> >>
> >>
> >> I don't that's exactly right.  At least as I understand it:
> >>
> >> - RAID1+0 (and RAID0+1) nests things - you start with two sets of RAID1
> >> mirrors, then stripe across them (or vice versa) - it's a nested set of
> >> steps
> >>
> >> - md RAID10 provides both mirroring and striping, but it's a more
> >> integrated function - (from the man page) "RAID10 provides a combination
> >> of RAID1 and RAID0, and sometimes known as RAID1+0. Every datablock is
> >> duplicated some number of times, and the resulting collection of
> >> datablocks are distributed over multiple drives." - but there isn't an
> >> inherent nesting in the process (i.e., no two disks are copies of each
> >> other, and md RAID10 will work over odd numbers of drives)
> >
> > Yes, you are right, RAID1+0 is nested, while Linux MD raid10 is not.
> > But the data layout of Linux MD RAID1+0 and Linux MD
> > RAID10,near is almost identical.
> >
> > keld
> > --
> > 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
> >
> 
> 
> 
> -- 
> Roberto Spadim
> Spadim Technology / SPAEmpresarial
--
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: wish for Linux MD mirrored raid types

[Leslie Rhorer]

> -----Original Message-----
> From: linux-raid-owner@vger.kernel.org [mailto:linux-raid-
> owner@vger.kernel.org] On Behalf Of Keld Jørn Simonsen
> Sent: Friday, May 06, 2011 8:28 AM
> To: David Brown
> Cc: linux-raid@vger.kernel.org
> Subject: Re: wish for Linux MD mirrored raid types
> 
> On Fri, May 06, 2011 at 12:54:28PM +0200, David Brown wrote:
> > On 06/05/2011 12:05, Jonathan Tripathy wrote:
> > >
> > >>>>RAID1 is traditionally a mirror only setup (ok, some RAID
> > >>>>implementations may do some load-balancing of some sort). So a RAID1
> > >>>>with 4 disks is one data set copied onto 4 disks. Bandwidth is
> roughly
> > >>>>the same as a single disk (ignoring any load balancing).
> > >>>>RAID10 is mirror and stripe. A RAID10 with 4 disks is similar to a 2
> > >>>>disk RAID0 (double bandwidth with data split in half across both
> > >>>>disks),
> > >>>>but with each disk having a mirror (which brings the total up to 4
> > >>>>drives).
> > >>>>
> > >>>>Additionally, a RAID1 disk (at least using MD) can be accessed just
> > >>>>like
> > >>>>a normal disk (good for recovery etc.) however a single disk out of
> a
> > >>>>RAID10 array is next to useless.
> > >Just so I can sleep at night, is my understanding of RAID10 and RAID1
> > >above correct?
> > >
> >
> > It's mostly right (assuming, of course, that /I/ am correct here...).
> >
> > RAID1 is traditionally a two-way mirror on two disks (or possibly on two
> > other raid sets, as in raid0+1).  So if you talk about RAID1 with 4
> > disks, you should probably qualify it more precisely - otherwise people
> > will wonder what you mean, or think it is impossible (many other RAID1
> > solutions, hardware or software, don't support more than two-way
> > mirrors).  /I/ would certainly say that a 4-disk RAID1 is a four-way
> > mirror as you described - but some people might think of a standard
> > layout RAID1+0.
> >
> > As you say, RAID10,near on four disks is pretty much identical to
> > RAID1+0 - i.e., a stripe of two normal RAID1 pairs.
> >
> > A single md RAID1 disk can be accessed like a normal disk, /if/ it uses
> > metadata format 0.90 which is put at the end of the drive.  If you have
> > later metadata formats that are at the beginning, then that will cause
> > trouble if you try to view the disk without using md.  A single disk
> > from a RAID10 is, as you say, useless without md.
> >
> > However, assuming your recovery PC supports md raid, then you can
> > assembly your single RAID10 disk as a degraded RAID10 array.  After all,
> > it wouldn't be very redundant if you only had access to your data when
> > all the disks were working!
> 
> I agree with what David Brown says here.
> 
> Just some further remarks:
> 
> You can have Linux MD raid10 arrays with only 2 disks.
> This is different from RAID1+0 that requires at least 4 disks.
> 
> You can also with Linux MD have an odd number of disk, which is
> impossible with RAID1+0.
> 
> Linux MD raid10 is a kernel driver that handles everything in
> one module, while RAID1+0 is handled in two drivers - raid0 over raid1.
> The latter is called nested RAID, and the Linux MD raid10 is not a
> nested RAID, while RAID1+0 is.
> 
> With Linux MD raid10,near and superblock 0.90 you can also
> boot it from old grub, lilo etc, as each partition can be seen as a normal
> file system.
> 
> As David pointed out, "a single disk out of a RAID10 array is next to
> useless" is not true. Linux MD RAID10 is designed to always be able to
> hold all data intact if one disk is out. You just need Linux MD raid10
> aware software.
> 
> I apologise for being abrupt in my first answer. But the thing is that
> I am getting tired of explaining and correcting all these misconceptions
> again and again. And my proposal is an attempt to stop at least some
> of the need for explaining, while also bringing the best practise forward
> to
> the users of our technology, for the benefit of the users.

	I think there are several more issues at hand, here, than your
suggestion takes into consideration.  Special Relativity, as the name
suggests, is a special case of General Relativity, but no one, and I mean
*NO ONE* ever approaches an SR problem by applying the mechanics of GR and
letting the equations collapse along the way.  Indeed, Einstein himself
developed SR several years before tackling the much more complex problem of
GR, the fact all the same physical processes are involved and the fact the
postulates are identical notwithstanding.

	It doesn't really much matter whether the developers decide to
implement RAID1 as a part of the same engine that delivers RAID10, or not.
It also does not matter whether the admin can produce the exact same results
at the code execution level by implementing a specific topology of RAID10 or
not.  OTOH, however, if there are any functional or operational differences
between the implementation of RAID1 and an implementation of RAID10 that
produces the same layout, then the two are simply not the same, at all.
Period.

	The bottom line, however, is that even if RAID1 is at the code level
just a specific implementation of RAID10, it still is a sufficiently
distinct operational mode to warrant its own name, setting it aside from the
more general RAID10 implementation.  Aside from that, when I say, "My
servers boot from 2 disk RAID1 arrays", everyone knows exactly what I mean
without my having to go into more detail about the layout.  For the most
part, it doesn't really matter if I happened to create the array using a
specific layout of RAID10.  This has its greatest importance for someone
inexperienced with RAID systems, who perhaps is setting up his very first
mirrored array with a pair of disks.  Having to try to learn all the ins and
outs of RAID10 layouts just to be able to create a pair (or triplet) of
mirrored disks is not appropriate.

--
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: wish for Linux MD mirrored raid types

[Miles Fidelman]

Leslie Rhorer wrote:
> 	The bottom line, however, is that even if RAID1 is at the code level
> just a specific implementation of RAID10, it still is a sufficiently
> distinct operational mode to warrant its own name, setting it aside from the
> more general RAID10 implementation.  Aside from that, when I say, "My
> servers boot from 2 disk RAID1 arrays", everyone knows exactly what I mean
> without my having to go into more detail about the layout.  For the most
> part, it doesn't really matter if I happened to create the array using a
> specific layout of RAID10.  This has its greatest importance for someone
> inexperienced with RAID systems, who perhaps is setting up his very first
> mirrored array with a pair of disks.  Having to try to learn all the ins and
> outs of RAID10 layouts just to be able to create a pair (or triplet) of
> mirrored disks is not appropriate.
>    

Good points all.

It is worth noting, that in this context, it's pretty silly to use a 
fancy setup for /boot and swap space - simple RAID1 mirrored partitions 
work just fine.  Where RAID10 is excellent is as a starting point for 
LVM, and volumes that support virtual machines.

-- 
In theory, there is no difference between theory and practice.
In<fnord>  practice, there is.   .... Yogi Berra

Re: wish for Linux MD mirrored raid types

[Luca Berra]

On Fri, May 06, 2011 at 09:17:52AM +0200, Keld Jørn Simonsen wrote:
>I would like linux MD raid10 functionality to be part of the Linux MD
>RAID1 module, and be called raid1. This is in accordance with the
then we need rename the current raid1 functionality to 'mirror'. in
order to avoid further confusion.
besides, current raid10 does not support resizing, so the feature
should be added before ditching 'mirror'

L.

-- 
Luca Berra -- bluca@comedia.it
--
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: wish for Linux MD mirrored raid types

[Miles Fidelman]

Luca Berra wrote:
> On Fri, May 06, 2011 at 09:17:52AM +0200, Keld Jørn Simonsen wrote:
>> I would like linux MD raid10 functionality to be part of the Linux MD
>> RAID1 module, and be called raid1. This is in accordance with the
> then we need rename the current raid1 functionality to 'mirror'. in
> order to avoid further confusion.
> besides, current raid10 does not support resizing, so the feature
> should be added before ditching 'mirror'
>

The current md RAID1 does exactly what RAID1 is supposed to do.

The md RAID10 is a very specific, and unique approach that has 
similarities to, but is distinct from, RAID1+0, RAID0+1, RAID5, and RAID6.

What say we leave the names alone.  Just beause one person is confused is no reason to further confuse things.

Just one man's opinion.

Miles Fidelman



-- 
In theory, there is no difference between theory and practice.
In<fnord>  practice, there is.   .... Yogi Berra


--
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: wish for Linux MD mirrored raid types

[Keld Jørn Simonsen]

On Sun, May 08, 2011 at 05:25:24PM -0400, Miles Fidelman wrote:
> Luca Berra wrote:
> >On Fri, May 06, 2011 at 09:17:52AM +0200, Keld Jørn Simonsen wrote:
> >>I would like linux MD raid10 functionality to be part of the Linux MD
> >>RAID1 module, and be called raid1. This is in accordance with the
> >then we need rename the current raid1 functionality to 'mirror'. in
> >order to avoid further confusion.
> >besides, current raid10 does not support resizing, so the feature
> >should be added before ditching 'mirror'
> >
> 
> The current md RAID1 does exactly what RAID1 is supposed to do.

I changed my mind a little.

I think we should follow SNIA wrt. RAID1 - all that SNIA would say is
RAID1 we also should be able to do as RAID1 - that would include
raid10-offset which directly was implemented in the Linux kernel
because of the SNIA RAID1 specification. It should also include raid10-far
in so far it is a raid1 type - say a raid10,f2 with only 2 disks. 

Then we should keep the raid10 stuff.

> The md RAID10 is a very specific, and unique approach that has 
> similarities to, but is distinct from, RAID1+0, RAID0+1, RAID5, and RAID6.

Yes, Linux MD raid10 is a very distinct type. We should talk with SNIA to get it
recognized.

> What say we leave the names alone.  Just beause one person is confused is 
> no reason to further confuse things.

The confusion is not just one person. The confusion is unbelievable common, 
and has proven to be very hard to eliminate. If we align with the SNIA
standard, and further get the standard to align with us, then we should
have a chance in say 5 years to have reduced the confusion considerably.

Best regards
Keld
--
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: wish for Linux MD mirrored raid types

[NeilBrown]

On Mon, 9 May 2011 05:40:36 +0200 Keld Jørn Simonsen <keld@keldix.com> wrote:

> On Sun, May 08, 2011 at 05:25:24PM -0400, Miles Fidelman wrote:
> > Luca Berra wrote:
> > >On Fri, May 06, 2011 at 09:17:52AM +0200, Keld Jørn Simonsen wrote:
> > >>I would like linux MD raid10 functionality to be part of the Linux MD
> > >>RAID1 module, and be called raid1. This is in accordance with the
> > >then we need rename the current raid1 functionality to 'mirror'. in
> > >order to avoid further confusion.
> > >besides, current raid10 does not support resizing, so the feature
> > >should be added before ditching 'mirror'
> > >
> > 
> > The current md RAID1 does exactly what RAID1 is supposed to do.
> 
> I changed my mind a little.
> 
> I think we should follow SNIA wrt. RAID1 - all that SNIA would say is
> RAID1 we also should be able to do as RAID1 - that would include
> raid10-offset which directly was implemented in the Linux kernel
> because of the SNIA RAID1 specification. It should also include raid10-far
> in so far it is a raid1 type - say a raid10,f2 with only 2 disks. 

I think you misread SNIA-DDF.

In DDF, arrays with a PRL (Primary RAID Level) of 1 can have an RLQ (RAID
level qualifier) of 0 or 1.
RLQ = 0  ->  RAID1 with 2 devices
RLQ = 1  ->  RAID1 with 3 devices.

DDF also devices a PRL of '11' which it calls "RAID-1E" (though this term
only appears once in the DDFv1.2 spec)

For PRL = 11  there are two options

RLQ = 0 -> Integrated Adjacent Stripe Mirroring
RLQ = 1 -> Integrated Offset Stripe Mirroring.

These correspond to md/raid10 "near2" and "offset2".

So DDF: RAID-1E  corresponds to md: RAID-10

So an 'E' rather than a '0'.

I would not be against allowing mdadm to accept "raid1e" as a synonym for
'raid10', and mentioning the alternate name in the documentation would be
entirely appropriate.

But RAID-1E is not RAID-1.  Nor is RAID-10.


> 
> Then we should keep the raid10 stuff.
> 
> > The md RAID10 is a very specific, and unique approach that has 
> > similarities to, but is distinct from, RAID1+0, RAID0+1, RAID5, and RAID6.
> 
> Yes, Linux MD raid10 is a very distinct type. We should talk with SNIA to get it
> recognized.
> 
> > What say we leave the names alone.  Just beause one person is confused is 
> > no reason to further confuse things.
> 
> The confusion is not just one person. The confusion is unbelievable common, 
> and has proven to be very hard to eliminate. If we align with the SNIA
> standard, and further get the standard to align with us, then we should
> have a chance in say 5 years to have reduced the confusion considerably.

The world is full of confusion that is hard to eliminate.

The problem here I think is simply people who do not educate themselves,
either because they cannot be bothered, or because they cannot easily find
the materials.

The first is not really a fixable problem.
The second we can address.  Improve the already-good wiki or add more text to
the man pages.  Have an aim that every general-information question can be
answered by simply posting a like or a passage from the man page.

That would be really worthwhile.

Changing names around is, I think, less valuable.

Thanks,
NeilBrown

--
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: wish for Linux MD mirrored raid types

[Emmanuel Noobadmin]

On 5/6/11, Keld Jørn Simonsen <keld@keldix.com> wrote:
> Hi List
>
> based on the recent discussion, that showed lacking knowledge
> on Linux MD RAID10 features, I have some thoughts:
>
> It is really hard to disseminate information on "new" features
> in MD RAID. RAID10 has been in the kernel since 2.6.9 - from 2004.
> I have tried to give info on RAID10 at a number of web pages,
> and still many people, even on our linux-raid list are not aware
> of it.
>
> Also many people are confused about Linux MD raid10 and RAID1+0.
>
> So I think we shopuld rather name things in another way.
>
> I would like linux MD raid10 functionality to be part of the Linux MD
> RAID1 module, and be called raid1. This is in accordance with the
> use of the RAID1 term as standadized by SNIA. In fact the RAID10-offset
> layout is an implementation of a SNIA RAID specification. The RAID10-near
> layout is an implementation of a simple RAID layout. And the RAID10-far
> layout is just another layout far a mirrored RAID.  So all these types
> could just be defined as different RAID1 layouts.

Giving my noob's 2 cents worth although I haven't followed the
original discussion. As a noob,  I think doing this will just confuse
us more.

There are plenty of existing materials around for those of us who try
to figure things out by googling. As it is, our (or maybe just me)
understanding is Linux RAID 1 is just like every other raid 1: simple
and straightforward, 2 drives mirroring each other.

This is is also usually the level that most of us start with. If the
instructions are short, easy to understand and simple to implement, we
usually gain confidence in using it and exploring mdraid further.

Most of us noobs are also aware that RAID 10 is more complicated and
there are two versions, i.e. 1+0 and 0+1. So psychologically, I had no
problems accepting that once I looked into it, there were much more
complex stuff and all these possible layouts: mdraid is cool!

Now, if RAID 10 was renamed to RAID 1, with the corresponding change
in documentation, what's going to happen for us noobs is this: "Omg,
why are there so many different versions and options just for raid 1?"
and importantly "Why is this manual/wiki different from the tons of
other pages about using mdraid 1?" For some, this would mean mdraid is
too difficult even for raid 1, mdraid is not cool! :)

So newbies will get more confused/frustrated as a result.

Personally, I had to spend some time figuring out (I'm noob and I'm
not very smart) the different layouts from the examples on wiki. This
is because there wasn't enough examples, at least to me, to clearly
show what's the difference if more/less disks were used. So for me,
and other noobs, it would probably help if the wiki had more examples
of each layout, maybe graphics to show the difference since it's
probably easier to see things if they were colour coded blocks rather
than stuff like A1 a1 A2 a2.

This and perhaps more elaboration on the difference between mdraid 10
and normal raid 10  would probably be better to clear up confusion
than renaming something we might have some familiarity with, into
something we also already have familiarity, resulting in something
that contradicts existing familiarity.
--
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: wish for Linux MD mirrored raid types

[Roberto Spadim]

:) yes, i didn't like the idea, and raid1 at linux can have many
mirrors not only 3...
the point here is:

raid1 isn't a industrial internacional name at linux eheh
raid1 is a module at linux kernel
raid10 is another module at linux kernel
raid10 and raid1 are part of md, so... don't check internacional
industry documentation to understand what raid1 and raid10 linux
modules are... check linux documentation

raid1 have write-behind and write-mostly, raid10 don't
there's different codes for raid10 and raid1, you must check
raid1 is an easy code, very simple
raid10 is more complex, with many layouts

i think the point here is a feature request to allow raid1 some
layouts, but... i think that's badblocks and others todo are more
important, users can use raid10 if they want different layouts, and...
they can code too... a patch is very welcome



2011/5/9 Emmanuel Noobadmin <centos.admin@gmail.com>:
> On 5/6/11, Keld Jørn Simonsen <keld@keldix.com> wrote:
>> Hi List
>>
>> based on the recent discussion, that showed lacking knowledge
>> on Linux MD RAID10 features, I have some thoughts:
>>
>> It is really hard to disseminate information on "new" features
>> in MD RAID. RAID10 has been in the kernel since 2.6.9 - from 2004.
>> I have tried to give info on RAID10 at a number of web pages,
>> and still many people, even on our linux-raid list are not aware
>> of it.
>>
>> Also many people are confused about Linux MD raid10 and RAID1+0.
>>
>> So I think we shopuld rather name things in another way.
>>
>> I would like linux MD raid10 functionality to be part of the Linux MD
>> RAID1 module, and be called raid1. This is in accordance with the
>> use of the RAID1 term as standadized by SNIA. In fact the RAID10-offset
>> layout is an implementation of a SNIA RAID specification. The RAID10-near
>> layout is an implementation of a simple RAID layout. And the RAID10-far
>> layout is just another layout far a mirrored RAID.  So all these types
>> could just be defined as different RAID1 layouts.
>
> Giving my noob's 2 cents worth although I haven't followed the
> original discussion. As a noob,  I think doing this will just confuse
> us more.
>
> There are plenty of existing materials around for those of us who try
> to figure things out by googling. As it is, our (or maybe just me)
> understanding is Linux RAID 1 is just like every other raid 1: simple
> and straightforward, 2 drives mirroring each other.
>
> This is is also usually the level that most of us start with. If the
> instructions are short, easy to understand and simple to implement, we
> usually gain confidence in using it and exploring mdraid further.
>
> Most of us noobs are also aware that RAID 10 is more complicated and
> there are two versions, i.e. 1+0 and 0+1. So psychologically, I had no
> problems accepting that once I looked into it, there were much more
> complex stuff and all these possible layouts: mdraid is cool!
>
> Now, if RAID 10 was renamed to RAID 1, with the corresponding change
> in documentation, what's going to happen for us noobs is this: "Omg,
> why are there so many different versions and options just for raid 1?"
> and importantly "Why is this manual/wiki different from the tons of
> other pages about using mdraid 1?" For some, this would mean mdraid is
> too difficult even for raid 1, mdraid is not cool! :)
>
> So newbies will get more confused/frustrated as a result.
>
> Personally, I had to spend some time figuring out (I'm noob and I'm
> not very smart) the different layouts from the examples on wiki. This
> is because there wasn't enough examples, at least to me, to clearly
> show what's the difference if more/less disks were used. So for me,
> and other noobs, it would probably help if the wiki had more examples
> of each layout, maybe graphics to show the difference since it's
> probably easier to see things if they were colour coded blocks rather
> than stuff like A1 a1 A2 a2.
>
> This and perhaps more elaboration on the difference between mdraid 10
> and normal raid 10  would probably be better to clear up confusion
> than renaming something we might have some familiarity with, into
> something we also already have familiarity, resulting in something
> that contradicts existing familiarity.
> --
> 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
>



-- 
Roberto Spadim
Spadim Technology / SPAEmpresarial
--
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: wish for Linux MD mirrored raid types

[Keld Jørn Simonsen]

On Mon, May 09, 2011 at 02:24:26PM +1000, NeilBrown wrote:
> On Mon, 9 May 2011 05:40:36 +0200 Keld Jørn Simonsen <keld@keldix.com> wrote:
> 
> > On Sun, May 08, 2011 at 05:25:24PM -0400, Miles Fidelman wrote:
> > > Luca Berra wrote:
> > > >On Fri, May 06, 2011 at 09:17:52AM +0200, Keld Jørn Simonsen wrote:
> > > >>I would like linux MD raid10 functionality to be part of the Linux MD
> > > >>RAID1 module, and be called raid1. This is in accordance with the
> > > >then we need rename the current raid1 functionality to 'mirror'. in
> > > >order to avoid further confusion.
> > > >besides, current raid10 does not support resizing, so the feature
> > > >should be added before ditching 'mirror'
> > > >
> > > 
> > > The current md RAID1 does exactly what RAID1 is supposed to do.
> > 
> > I changed my mind a little.
> > 
> > I think we should follow SNIA wrt. RAID1 - all that SNIA would say is
> > RAID1 we also should be able to do as RAID1 - that would include
> > raid10-offset which directly was implemented in the Linux kernel
> > because of the SNIA RAID1 specification. It should also include raid10-far
> > in so far it is a raid1 type - say a raid10,f2 with only 2 disks. 
> 
> I think you misread SNIA-DDF.
> 
> In DDF, arrays with a PRL (Primary RAID Level) of 1 can have an RLQ (RAID
> level qualifier) of 0 or 1.
> RLQ = 0  ->  RAID1 with 2 devices
> RLQ = 1  ->  RAID1 with 3 devices.
> 
> DDF also devices a PRL of '11' which it calls "RAID-1E" (though this term
> only appears once in the DDFv1.2 spec)
> 
> For PRL = 11  there are two options
> 
> RLQ = 0 -> Integrated Adjacent Stripe Mirroring
> RLQ = 1 -> Integrated Offset Stripe Mirroring.
> 
> These correspond to md/raid10 "near2" and "offset2".
> 
> So DDF: RAID-1E  corresponds to md: RAID-10
> 
> So an 'E' rather than a '0'.
> 
> I would not be against allowing mdadm to accept "raid1e" as a synonym for
> 'raid10', and mentioning the alternate name in the documentation would be
> entirely appropriate.

Yes, I would like that we cocument the correspondance to the SNIA DDF
standard, and do other descriptions as you stipulate.

> But RAID-1E is not RAID-1.  Nor is RAID-10.

I see.

By "Nor is RAID-10." you mean RAID-1+0 or linux md raid10?

What is the difference between RAID-1E and Linux MD raid10?
Linux MD raid10 has "far" layout?
Linux MD raid10 can have more than 2 copies?
More stuff?



> > 
> > Then we should keep the raid10 stuff.
> > 
> > > The md RAID10 is a very specific, and unique approach that has 
> > > similarities to, but is distinct from, RAID1+0, RAID0+1, RAID5, and RAID6.
> > 
> > Yes, Linux MD raid10 is a very distinct type. We should talk with SNIA to get it
> > recognized.
> > 
> > > What say we leave the names alone.  Just beause one person is confused is 
> > > no reason to further confuse things.
> > 
> > The confusion is not just one person. The confusion is unbelievable common, 
> > and has proven to be very hard to eliminate. If we align with the SNIA
> > standard, and further get the standard to align with us, then we should
> > have a chance in say 5 years to have reduced the confusion considerably.
> 
> The world is full of confusion that is hard to eliminate.
> 
> The problem here I think is simply people who do not educate themselves,
> either because they cannot be bothered, or because they cannot easily find
> the materials.
> 
> The first is not really a fixable problem.
> The second we can address.  Improve the already-good wiki or add more text to
> the man pages.  Have an aim that every general-information question can be
> answered by simply posting a like or a passage from the man page.
> 
> That would be really worthwhile.
> 
> Changing names around is, I think, less valuable.

It is just that I have been around on other web pages for RAID, to
improve references to the linux-raid wiki and to correct errors in their
description of Linux RAID. Many - maybe most - people that write about
Linux RAID have a number of their details wrong.  That should be experts
conveying their expert wisdom to knowledge-hungry users. An example:
http://wiki.linuxquestions.org/wiki/RAID#RAID-10
They claim you need 4 disks for MD raid10. Another example is the German
wikipedia page on RAID - which was moderated and the moderator did not
accept my edits. This page does not describe Linux MD raid10. 

I think naming matters. If we could call Linux MD raid10 for raid1e
I think much would be achieved in terms of eliminating
misunderstandings.

Best regards
keld
--
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: wish for Linux MD mirrored raid types

[Keld Jørn Simonsen]

On Mon, May 09, 2011 at 01:22:42PM +0800, Emmanuel Noobadmin wrote:
> On 5/6/11, Keld Jørn Simonsen <keld@keldix.com> wrote:
> > Hi List
> >
> > based on the recent discussion, that showed lacking knowledge
> > on Linux MD RAID10 features, I have some thoughts:
> >
> > It is really hard to disseminate information on "new" features
> > in MD RAID. RAID10 has been in the kernel since 2.6.9 - from 2004.
> > I have tried to give info on RAID10 at a number of web pages,
> > and still many people, even on our linux-raid list are not aware
> > of it.
> >
> > Also many people are confused about Linux MD raid10 and RAID1+0.
> >
> > So I think we shopuld rather name things in another way.
> >
> > I would like linux MD raid10 functionality to be part of the Linux MD
> > RAID1 module, and be called raid1. This is in accordance with the
> > use of the RAID1 term as standadized by SNIA. In fact the RAID10-offset
> > layout is an implementation of a SNIA RAID specification. The RAID10-near
> > layout is an implementation of a simple RAID layout. And the RAID10-far
> > layout is just another layout far a mirrored RAID.  So all these types
> > could just be defined as different RAID1 layouts.
> 
> Giving my noob's 2 cents worth although I haven't followed the
> original discussion. As a noob,  I think doing this will just confuse
> us more.
> 
> There are plenty of existing materials around for those of us who try
> to figure things out by googling. As it is, our (or maybe just me)
> understanding is Linux RAID 1 is just like every other raid 1: simple
> and straightforward, 2 drives mirroring each other.
> 
> This is is also usually the level that most of us start with. If the
> instructions are short, easy to understand and simple to implement, we
> usually gain confidence in using it and exploring mdraid further.
> 
> Most of us noobs are also aware that RAID 10 is more complicated and
> there are two versions, i.e. 1+0 and 0+1. So psychologically, I had no
> problems accepting that once I looked into it, there were much more
> complex stuff and all these possible layouts: mdraid is cool!
> 
> Now, if RAID 10 was renamed to RAID 1, with the corresponding change
> in documentation, what's going to happen for us noobs is this: "Omg,
> why are there so many different versions and options just for raid 1?"
> and importantly "Why is this manual/wiki different from the tons of
> other pages about using mdraid 1?" For some, this would mean mdraid is
> too difficult even for raid 1, mdraid is not cool! :)
> 
> So newbies will get more confused/frustrated as a result.
> 
> Personally, I had to spend some time figuring out (I'm noob and I'm
> not very smart) the different layouts from the examples on wiki. This
> is because there wasn't enough examples, at least to me, to clearly
> show what's the difference if more/less disks were used. So for me,
> and other noobs, it would probably help if the wiki had more examples
> of each layout, maybe graphics to show the difference since it's
> probably easier to see things if they were colour coded blocks rather
> than stuff like A1 a1 A2 a2.
> 
> This and perhaps more elaboration on the difference between mdraid 10
> and normal raid 10  would probably be better to clear up confusion
> than renaming something we might have some familiarity with, into
> something we also already have familiarity, resulting in something
> that contradicts existing familiarity.

I see your point. Given Neil's input I think we should  rateher call it
raid1e.

Best regards
keld
--
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: wish for Linux MD mirrored raid types

[Roberto Spadim]

hum, if you check docs, source, and test, you will check that md raid
can have >=2 mirrors without problems, you can have 3 disks and 3
mirrors, you can have 3 disks and 2 mirrors, you can have 1 disk with
10 partitions, and each partition be a mirror of each other, the point
here is, md software isn´t a hardware raid, it´s very near but have
more features, i don´t know if we should change md to allow it be
compatible with a 'industrial standard' since we have more feature
than standard, you must check what you want and use it
maybe some article at wiki,
standard X, layout Y, level W = md raid level A layout B devices C
mirrors D spares E .....
i think this is welcome to linux wiki page

2011/5/9 Keld Jørn Simonsen <keld@keldix.com>:
> On Mon, May 09, 2011 at 01:22:42PM +0800, Emmanuel Noobadmin wrote:
>> On 5/6/11, Keld Jørn Simonsen <keld@keldix.com> wrote:
>> > Hi List
>> >
>> > based on the recent discussion, that showed lacking knowledge
>> > on Linux MD RAID10 features, I have some thoughts:
>> >
>> > It is really hard to disseminate information on "new" features
>> > in MD RAID. RAID10 has been in the kernel since 2.6.9 - from 2004.
>> > I have tried to give info on RAID10 at a number of web pages,
>> > and still many people, even on our linux-raid list are not aware
>> > of it.
>> >
>> > Also many people are confused about Linux MD raid10 and RAID1+0.
>> >
>> > So I think we shopuld rather name things in another way.
>> >
>> > I would like linux MD raid10 functionality to be part of the Linux MD
>> > RAID1 module, and be called raid1. This is in accordance with the
>> > use of the RAID1 term as standadized by SNIA. In fact the RAID10-offset
>> > layout is an implementation of a SNIA RAID specification. The RAID10-near
>> > layout is an implementation of a simple RAID layout. And the RAID10-far
>> > layout is just another layout far a mirrored RAID.  So all these types
>> > could just be defined as different RAID1 layouts.
>>
>> Giving my noob's 2 cents worth although I haven't followed the
>> original discussion. As a noob,  I think doing this will just confuse
>> us more.
>>
>> There are plenty of existing materials around for those of us who try
>> to figure things out by googling. As it is, our (or maybe just me)
>> understanding is Linux RAID 1 is just like every other raid 1: simple
>> and straightforward, 2 drives mirroring each other.
>>
>> This is is also usually the level that most of us start with. If the
>> instructions are short, easy to understand and simple to implement, we
>> usually gain confidence in using it and exploring mdraid further.
>>
>> Most of us noobs are also aware that RAID 10 is more complicated and
>> there are two versions, i.e. 1+0 and 0+1. So psychologically, I had no
>> problems accepting that once I looked into it, there were much more
>> complex stuff and all these possible layouts: mdraid is cool!
>>
>> Now, if RAID 10 was renamed to RAID 1, with the corresponding change
>> in documentation, what's going to happen for us noobs is this: "Omg,
>> why are there so many different versions and options just for raid 1?"
>> and importantly "Why is this manual/wiki different from the tons of
>> other pages about using mdraid 1?" For some, this would mean mdraid is
>> too difficult even for raid 1, mdraid is not cool! :)
>>
>> So newbies will get more confused/frustrated as a result.
>>
>> Personally, I had to spend some time figuring out (I'm noob and I'm
>> not very smart) the different layouts from the examples on wiki. This
>> is because there wasn't enough examples, at least to me, to clearly
>> show what's the difference if more/less disks were used. So for me,
>> and other noobs, it would probably help if the wiki had more examples
>> of each layout, maybe graphics to show the difference since it's
>> probably easier to see things if they were colour coded blocks rather
>> than stuff like A1 a1 A2 a2.
>>
>> This and perhaps more elaboration on the difference between mdraid 10
>> and normal raid 10  would probably be better to clear up confusion
>> than renaming something we might have some familiarity with, into
>> something we also already have familiarity, resulting in something
>> that contradicts existing familiarity.
>
> I see your point. Given Neil's input I think we should  rateher call it
> raid1e.
>
> Best regards
> keld
> --
> 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
>



-- 
Roberto Spadim
Spadim Technology / SPAEmpresarial
--
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