raidz-like operation?

raidz-like operation?

[Roy Sigurd Karlsbakk]

Hi

I know it's not in yet, but is there a plan to implement raid-z(2)-=20
like operation? I have heard of plans to integrate with the existing =20
md-subsystem in the kernel, but I'm not sure if this will be =20
appropriate comparing this to how things are done in zfs/raid-z.

roy
--
Roy Sigurd Karlsbakk
(+47) 97542685
roy@karlsbakk.net
http://blogg.karlsbakk.net/
--
I all pedagogikk er det essensielt at pensum presenteres =20
intelligibelt. Det er et element=E6rt imperativ for alle pedagoger =E5 =
=20
unng=E5 eksessiv anvendelse av idiomer med fremmed opprinnelse. I de =20
fleste tilfeller eksisterer adekvate og relevante synonymer p=E5 norsk.

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

Re: raidz-like operation?

[David Woodhouse]

On Fri, 2009-07-10 at 10:58 +0200, Roy Sigurd Karlsbakk wrote:
> I know it's not in yet, but is there a plan to implement raid-z(2)- 
> like operation? I have heard of plans to integrate with the existing  
> md-subsystem in the kernel, but I'm not sure if this will be  
> appropriate comparing this to how things are done in zfs/raid-z.

I'm currently working on the basic implementation of RAID5/RAID6 (qv).
I don't think there'll be much integration with the md subsystem --
we'll probably just shift the cunning GF arithmetic out into lib/raid6
or something, then let them both use it from there.

We do want to do full stripe writes to the RAID[56] parts, so we're
going to want to do something like falling back to RAID1 (or to a RAID5
chunk with smaller stripe length) for smaller allocations. That's
basically what you're talking about, isn't it?

-- 
David Woodhouse                            Open Source Technology Centre
David.Woodhouse@intel.com                              Intel Corporation

Re: raidz-like operation?

[Roy Sigurd Karlsbakk]

On 11. juli. 2009, at 16.59, David Woodhouse wrote:

> On Fri, 2009-07-10 at 10:58 +0200, Roy Sigurd Karlsbakk wrote:
>> I know it's not in yet, but is there a plan to implement raid-z(2)-
>> like operation? I have heard of plans to integrate with the existing
>> md-subsystem in the kernel, but I'm not sure if this will be
>> appropriate comparing this to how things are done in zfs/raid-z.
>
> I'm currently working on the basic implementation of RAID5/RAID6 (qv)=
=2E
> I don't think there'll be much integration with the md subsystem --
> we'll probably just shift the cunning GF arithmetic out into lib/raid=
6
> or something, then let them both use it from there.
>
> We do want to do full stripe writes to the RAID[56] parts, so we're
> going to want to do something like falling back to RAID1 (or to a =20
> RAID5
> chunk with smaller stripe length) for smaller allocations. That's
> basically what you're talking about, isn't it?


In ZFS/RAIDZ(2) they are said to use RAID3-like partial writes as with =
=20
with a file-level redundancy, meaning if a file won't cover a whole =20
stripe, it'll use a partial one. I'm not sure if this is the case, but =
=20
it's what I've been told by the people from Sun. This may open for =20
fewer IOPs per write. Is something like this in the thoughts of the =20
new RAID[56] in BTRFS?

roy
--
Roy Sigurd Karlsbakk
(+47) 97542685
roy@karlsbakk.net
http://blogg.karlsbakk.net/
--
I all pedagogikk er det essensielt at pensum presenteres =20
intelligibelt. Det er et element=E6rt imperativ for alle pedagoger =E5 =
=20
unng=E5 eksessiv anvendelse av idiomer med fremmed opprinnelse. I de =20
fleste tilfeller eksisterer adekvate og relevante synonymer p=E5 norsk.

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