ideas - raid write

hi guys, i have some questions...
how write works in raid1?
md send write to all mirrors and wait write end to return 'ok' to filesystem?
could we send write to mirrors and return with one 'ok'? like a async write?

--
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: ideas - raid write

if others fail, fail the disks =3D), it=B4s something like nbd (mysql) =
cluster
two disks ok, return ok others works like async write
on error, resync nodes/mirrors
i don=B4t know the name but they call somethink like isosyncronous or
semisyncronous, i don=B4t remember now

2011/3/3 Mathias Bur=E9n :
> And if md sends ok when first write is complete, then other writes fa=
il?
>
> On 3 Mar 2011 22:45, "Roberto Spadim" wrote:
>> hi guys, i have some questions...
>> how write works in raid1?
>> md send write to all mirrors and wait write end to return 'ok' to
>> filesystem?
>> could we send write to mirrors and return with one 'ok'? like a asyn=
c
>> write?
>>
>> --
>> 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
>



--=20
Roberto Spadim
Spadim Technology / SPAEmpresarial
--
To unsubscribe from this list: send the line "unsubscribe linux-raid" i=
n
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html

Re: ideas - raid write

ops... ndb mysql cluster

it=B4s like to get more speed on small (cache time/size) writes
may be dirty, but if only X disks are in sync array could continue
running with async writes without problems


2011/3/3 Roberto Spadim :
> if others fail, fail the disks =3D), it=B4s something like nbd (mysql=
) cluster
> two disks ok, return ok others works like async write
> on error, resync nodes/mirrors
> i don=B4t know the name but they call somethink like isosyncronous or
> semisyncronous, i don=B4t remember now
>
> 2011/3/3 Mathias Bur=E9n :
>> And if md sends ok when first write is complete, then other writes f=
ail?
>>
>> On 3 Mar 2011 22:45, "Roberto Spadim" wrote:
>>> hi guys, i have some questions...
>>> how write works in raid1?
>>> md send write to all mirrors and wait write end to return 'ok' to
>>> filesystem?
>>> could we send write to mirrors and return with one 'ok'? like a asy=
nc
>>> write?
>>>
>>> --
>>> Roberto Spadim
>>> Spadim Technology / SPAEmpresarial
>>> --
>>> To unsubscribe from this list: send the line "unsubscribe linux-rai=
d" 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
>



--=20
Roberto Spadim
Spadim Technology / SPAEmpresarial
--
To unsubscribe from this list: send the line "unsubscribe linux-raid" i=
n
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html

Re: ideas - raid write

in other words...
the write speed is limited by the fastest disk (one or more disks how
many sync/async writes should we allow is another problem, not for
now), not the slowest
the cache is another limit (when cache get full, async writes should wa=
it)

2011/3/3 Roberto Spadim :
> ops... ndb mysql cluster
>
> it=B4s like to get more speed on small (cache time/size) writes
> may be dirty, but if only X disks are in sync array could continue
> running with async writes without problems
>
>
> 2011/3/3 Roberto Spadim :
>> if others fail, fail the disks =3D), it=B4s something like nbd (mysq=
l) cluster
>> two disks ok, return ok others works like async write
>> on error, resync nodes/mirrors
>> i don=B4t know the name but they call somethink like isosyncronous o=
r
>> semisyncronous, i don=B4t remember now
>>
>> 2011/3/3 Mathias Bur=E9n :
>>> And if md sends ok when first write is complete, then other writes =
fail?
>>>
>>> On 3 Mar 2011 22:45, "Roberto Spadim" wrote=
:
>>>> hi guys, i have some questions...
>>>> how write works in raid1?
>>>> md send write to all mirrors and wait write end to return 'ok' to
>>>> filesystem?
>>>> could we send write to mirrors and return with one 'ok'? like a as=
ync
>>>> write?
>>>>
>>>> --
>>>> Roberto Spadim
>>>> Spadim Technology / SPAEmpresarial
>>>> --
>>>> To unsubscribe from this list: send the line "unsubscribe linux-ra=
id" 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
>>
>
>
>
> --
> Roberto Spadim
> Spadim Technology / SPAEmpresarial
>



--=20
Roberto Spadim
Spadim Technology / SPAEmpresarial
--
To unsubscribe from this list: send the line "unsubscribe linux-raid" i=
n
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html

Re: ideas - raid write

just a idea, try new write performaces improvements, with 'partial
security', some mirrors stay 'in sync' (give some security) and others
are doing async writes
with raid1 (or raid10) with more than 2 mirrors

2011/3/3 chris :
> What is the point of these posts...
>
> On Mar 3, 2011 6:24 PM, "Roberto Spadim" wrot=
e:
>
> in other words...
> the write speed is limited by the fastest disk (one or more disks how
> many sync/async writes should we allow is another problem, not for
> now), not the slowest
> the cache is another limit (when cache get full, async writes should =
wait)
>
> 2011/3/3 Roberto Spadim :
>> ops... ndb mysql cluster
>>
>> it=B4s like to get mo...



--=20
Roberto Spadim
Spadim Technology / SPAEmpresarial
--
To unsubscribe from this list: send the line "unsubscribe linux-raid" i=
n
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html

Re: ideas - raid write

any idea if this could work?

2011/3/3 Roberto Spadim :
> just a idea, try new write performaces improvements, with 'partial
> security', some mirrors stay 'in sync' (give some security) and other=
s
> are doing async writes
> with raid1 (or raid10) with more than 2 mirrors
>
> 2011/3/3 chris :
>> What is the point of these posts...
>>
>> On Mar 3, 2011 6:24 PM, "Roberto Spadim" wro=
te:
>>
>> in other words...
>> the write speed is limited by the fastest disk (one or more disks ho=
w
>> many sync/async writes should we allow is another problem, not for
>> now), not the slowest
>> the cache is another limit (when cache get full, async writes should=
wait)
>>
>> 2011/3/3 Roberto Spadim :
>>> ops... ndb mysql cluster
>>>
>>> it=B4s like to get mo...
>
>
>
> --
> Roberto Spadim
> Spadim Technology / SPAEmpresarial
>



--=20
Roberto Spadim
Spadim Technology / SPAEmpresarial
--
To unsubscribe from this list: send the line "unsubscribe linux-raid" i=
n
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html