Re: Latest recommendation for asmlib + number of LUNs
From: Frits Hoogland <frits.hoogland_at_gmail.com>
Date: Thu, 15 Aug 2013 14:45:38 +0200
Message-Id: <F5C54B18-1CEE-40AC-8649-958050635A6E_at_gmail.com>
Content-Transfer-Encoding: quoted-printable Content-Type: text/plain;
> From: Niall Litchfield [mailto:niall.litchfield_at_gmail.com]
> Sent: Thursday, August 15, 2013 5:15 AM
> To: Stephens, Chris
> Cc: oracle-l_at_freelists.org
> Subject: Re: Latest recommendation for asmlib + number of LUNs
>=20
> As you say you'll encounter at least 2 different opinions. I fall into =
the "don't think its worth it" camp, so bear that in mind in what = follows. We also use it exclusively here for our RAC systems - haven't = got that standard rewritten as yet :)
>=20
> Technically in 11.x asmlib only provides 2 features, both of which I =
would argue are available without it.
> 1) it provides an exclusive I/O interface to the device
> 2) it labels the disk so that there is persistent naming for the =
device.
>=20
> That's it. It's worth remembering that ASMLIB was born in 2004 when =
async i/o and persistent device naming on Linux platforms was, errm, = deficient. Those conditions just aren't true today. Disk labelling and = aliasing is built into both multipath and udev on Linux and works = perfectly well to ensure that Oracle always finds the same devices at = the same name. In addition ASMLIB consists both of a userland and kernel = module, consequently if you patch the O/S kernel - say for security = reasons - then you *will* have to update asmlib each and every time as = well. In many cases the *you* that has to do this will be your sys admin = team, in the case of base O/S packages this shouldn't be an issue. In = addition as of RHEL/OEL 6 asmlib is pre-compiled only for the = Unbreakable Enterprise Kernel supplied by Oracle, you should probably be = using this to be honest, but many people won't be. For these reasons I = see asmlib as an extra layer requiring maintenance and providing no = added value.
>=20
> Definitely read https://blogs.oracle.com/wim/entry/asmlib for a =
definitive background. Now Oracle could, and indeed to some extent (on = Oracle hardware) has enhanced asmlib to provide further features - see = http://www.scaleabilities.co.uk/2013/02/07/diagnosing-asmlib/ for an = example.
> On Wed, Aug 14, 2013 at 8:34 PM, Stephens, Chris =
<Chris.Stephens_at_adm.com<mailto:Chris.Stephens_at_adm.com>> wrote:
> I'm venturing into the land of Oracle RAC for the first time and will =
be installing an 11.2.0.3 2-Node cluster from scratch. The storage will = be allocated from a Dell Compellent (not sure if that's relevant to my = question or not).
> The official documentation recommends the use of ASMLib but I seem to =
remember a lot of folks on here thinking that was a bad idea or at least = not a no brainer.
> What's the current recommendation / bestest idea on ASMLib use?
>=20
> Also, any recommendation on the number of LUN's to start off with for =
each disk group? (+data, +fra) Oracle recommends a minimum of 4 but I'm = not sure how they arrived at that number or whether more would be = advantageous. That's probably an "it depends" question so what info do = I need to make that determination?
> Thanks,
> Chris
> CONFIDENTIALITY NOTICE:
> This message is intended for the use of the individual or entity to =
which it is addressed and may contain information that is privileged, = confidential and exempt from disclosure under applicable law. If the = reader of this message is not the intended recipient or the employee or = agent responsible for delivering this message to the intended recipient, = you are hereby notified that any dissemination, distribution or copying = of this communication is strictly prohibited. If you have received this = communication in error, please notify us immediately by email reply.
> --
> http://www.freelists.org/webpage/oracle-l
> --
> Niall Litchfield
> Oracle DBA
> http://www.orawin.info
> CONFIDENTIALITY NOTICE:
> This message is intended for the use of the individual or entity to =
which it is addressed and may contain information that is privileged, = confidential and exempt from disclosure under applicable law. If the = reader of this message is not the intended recipient or the employee or = agent responsible for delivering this message to the intended recipient, = you are hereby notified that any dissemination, distribution or copying = of this communication is strictly prohibited. If you have received this = communication in error, please notify us immediately by email reply.
> --
> http://www.freelists.org/webpage/oracle-l
>=20
>=20
Date: Thu, 15 Aug 2013 14:45:38 +0200
Message-Id: <F5C54B18-1CEE-40AC-8649-958050635A6E_at_gmail.com>
Content-Transfer-Encoding: quoted-printable Content-Type: text/plain;
charset=us-ascii
Please mind asmlib is not available from the public yum server, which =
means updating from there will not update asmlib, meaning asmlib is gone =
if there's a new kernel.
Frits Hoogland
http://fritshoogland.wordpress.com
frits.hoogland_at_gmail.com
+31 6 53569942
On Aug 15, 2013, at 2:29 PM, "Stephens, Chris" <Chris.Stephens_at_adm.com> = wrote:
> I forgot to mention that we will be using the UEK version of Oracle =
Linux so <I think> the patching is a non-issue.
> Thanks for the links and input from everyone.
>=20
> From: Niall Litchfield [mailto:niall.litchfield_at_gmail.com]
> Sent: Thursday, August 15, 2013 5:15 AM
> To: Stephens, Chris
> Cc: oracle-l_at_freelists.org
> Subject: Re: Latest recommendation for asmlib + number of LUNs
>=20
> As you say you'll encounter at least 2 different opinions. I fall into =
the "don't think its worth it" camp, so bear that in mind in what = follows. We also use it exclusively here for our RAC systems - haven't = got that standard rewritten as yet :)
>=20
> Technically in 11.x asmlib only provides 2 features, both of which I =
would argue are available without it.
>=20
> 1) it provides an exclusive I/O interface to the device
> 2) it labels the disk so that there is persistent naming for the =
device.
>=20
> That's it. It's worth remembering that ASMLIB was born in 2004 when =
async i/o and persistent device naming on Linux platforms was, errm, = deficient. Those conditions just aren't true today. Disk labelling and = aliasing is built into both multipath and udev on Linux and works = perfectly well to ensure that Oracle always finds the same devices at = the same name. In addition ASMLIB consists both of a userland and kernel = module, consequently if you patch the O/S kernel - say for security = reasons - then you *will* have to update asmlib each and every time as = well. In many cases the *you* that has to do this will be your sys admin = team, in the case of base O/S packages this shouldn't be an issue. In = addition as of RHEL/OEL 6 asmlib is pre-compiled only for the = Unbreakable Enterprise Kernel supplied by Oracle, you should probably be = using this to be honest, but many people won't be. For these reasons I = see asmlib as an extra layer requiring maintenance and providing no = added value.
>=20
> Definitely read https://blogs.oracle.com/wim/entry/asmlib for a =
definitive background. Now Oracle could, and indeed to some extent (on = Oracle hardware) has enhanced asmlib to provide further features - see = http://www.scaleabilities.co.uk/2013/02/07/diagnosing-asmlib/ for an = example.
>=20 >=20 >=20 >=20 >=20
> On Wed, Aug 14, 2013 at 8:34 PM, Stephens, Chris =
<Chris.Stephens_at_adm.com<mailto:Chris.Stephens_at_adm.com>> wrote:
> I'm venturing into the land of Oracle RAC for the first time and will =
be installing an 11.2.0.3 2-Node cluster from scratch. The storage will = be allocated from a Dell Compellent (not sure if that's relevant to my = question or not).
> The official documentation recommends the use of ASMLib but I seem to =
remember a lot of folks on here thinking that was a bad idea or at least = not a no brainer.
>=20
> What's the current recommendation / bestest idea on ASMLib use?
>=20
> Also, any recommendation on the number of LUN's to start off with for =
each disk group? (+data, +fra) Oracle recommends a minimum of 4 but I'm = not sure how they arrived at that number or whether more would be = advantageous. That's probably an "it depends" question so what info do = I need to make that determination?
>=20
> Thanks,
> Chris
>=20 >=20
> CONFIDENTIALITY NOTICE:
> This message is intended for the use of the individual or entity to =
which it is addressed and may contain information that is privileged, = confidential and exempt from disclosure under applicable law. If the = reader of this message is not the intended recipient or the employee or = agent responsible for delivering this message to the intended recipient, = you are hereby notified that any dissemination, distribution or copying = of this communication is strictly prohibited. If you have received this = communication in error, please notify us immediately by email reply.
>=20 >=20 >=20
> --
> http://www.freelists.org/webpage/oracle-l
>=20 >=20 >=20 >=20
> --
> Niall Litchfield
> Oracle DBA
> http://www.orawin.info
>=20
> CONFIDENTIALITY NOTICE:
> This message is intended for the use of the individual or entity to =
which it is addressed and may contain information that is privileged, = confidential and exempt from disclosure under applicable law. If the = reader of this message is not the intended recipient or the employee or = agent responsible for delivering this message to the intended recipient, = you are hereby notified that any dissemination, distribution or copying = of this communication is strictly prohibited. If you have received this = communication in error, please notify us immediately by email reply.
>=20 >=20 >=20
> --
> http://www.freelists.org/webpage/oracle-l
>=20
>=20
-- http://www.freelists.org/webpage/oracle-lReceived on Thu Aug 15 2013 - 14:45:38 CEST