Received: (qmail 8479 invoked from network); 7 Jul 2010 04:17:49 -0500 Received: from freelists-180.iquest.net (HELO turing.freelists.org) (206.53.239.180) by static-ip-85-25-126-90.inaddr.intergenia.de with SMTP; 7 Jul 2010 04:17:47 -0500 Received: from localhost (localhost [127.0.0.1]) by turing.freelists.org (Avenir Technologies Mail Multiplex) with ESMTP id 68ABFD19FFA; Wed, 7 Jul 2010 05:17:03 -0400 (EDT) X-Virus-Scanned: Debian amavisd-new at localhost.localdomain Received: from turing.freelists.org ([127.0.0.1]) by localhost (turing.freelists.org [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id Ry7rQDnHI0tg; Wed, 7 Jul 2010 05:17:03 -0400 (EDT) Received: from turing.freelists.org (localhost [127.0.0.1]) by turing.freelists.org (Avenir Technologies Mail Multiplex) with ESMTP id 49A86D1A1D9; Wed, 7 Jul 2010 05:16:20 -0400 (EDT) Received: with ECARTIS (v1.0.0; list oracle-l); Wed, 07 Jul 2010 05:15:38 -0400 (EDT) Received: from localhost (localhost [127.0.0.1]) by turing.freelists.org (Avenir Technologies Mail Multiplex) with ESMTP id 6061DD19FFA for ; Wed, 7 Jul 2010 05:15:38 -0400 (EDT) Received: from turing.freelists.org ([127.0.0.1]) by localhost (turing.freelists.org [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id hnu5B9fuzRh7 for ; Wed, 7 Jul 2010 05:15:38 -0400 (EDT) Received: from mail-qy0-f179.google.com (mail-qy0-f179.google.com [209.85.216.179]) by turing.freelists.org (Avenir Technologies Mail Multiplex) with ESMTP id DEDA1D19F02 for ; Wed, 7 Jul 2010 05:15:37 -0400 (EDT) Received: by qyk2 with SMTP id 2so2373305qyk.10 for ; Wed, 07 Jul 2010 02:15:37 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=domainkey-signature:mime-version:received:received:in-reply-to :references:date:message-id:subject:from:to:cc:content-type; bh=fWt4OdVEL57DXv4sORIDEC3CSXoQrKAZZL3BZ4NSn9Q=; b=waKW7a1S8YnlNc9s1Ahuy+mkO0YJf+HF60rDuKPSbcz3nal7lEI6TX6W79I49dscFZ tu2lLiSS/hIwr2CBZpra+vTET3W+bfiTUwzPv8WvgXZ5VMpW/HOG6mUzo+Uxa/thXFru Wdl8KIHColJauJJNzbVT/8EC2xnFqwR8XyPZk= DomainKey-Signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type; b=w9RpSastVP5M5VUQMq6gmDA0dNzEEoaJnCCfx/1QgANREPkFzzRYtX1O1ROOpZDdW+ TuZ11UxUNLq2qzmz/Uj55k9hbRvI6LdKxY0bangyCl9t1veKwfIlvC2Zd9bEcn3CCovr fE3ED8/00L/2WsjSxaF83vdV+0jD8rpi5hpEc= MIME-Version: 1.0 Received: by 10.224.96.160 with SMTP id h32mr3324959qan.269.1278494136749; Wed, 07 Jul 2010 02:15:36 -0700 (PDT) Received: by 10.229.213.70 with HTTP; Wed, 7 Jul 2010 02:15:36 -0700 (PDT) In-Reply-To: <6AFC12B9BFCDEA45B7274C534738067F55A23BF9@AAPQMAILBX02V.proque.st> References: <6AFC12B9BFCDEA45B7274C534738067F565F77DC@AAPQMAILBX02V.proque.st> <6AFC12B9BFCDEA45B7274C534738067F55A23BF9@AAPQMAILBX02V.proque.st> Date: Wed, 7 Jul 2010 10:15:36 +0100 Message-ID: Subject: Re: VMWARE Snapshots to backup Oracle From: Niall Litchfield To: Mark.Bobak@proquest.com Cc: "big.dave.roberts@googlemail.com" , oracle-l Content-Type: multipart/alternative; boundary=00c09f89957ede50e5048ac89cf1 X-archive-position: 29542 X-ecartis-version: Ecartis v1.0.0 Sender: oracle-l-bounce@freelists.org Errors-to: oracle-l-bounce@freelists.org X-original-sender: niall.litchfield@gmail.com Precedence: normal Reply-To: niall.litchfield@gmail.com List-help: List-unsubscribe: List-software: Ecartis version 1.0.0 List-Id: oracle-l X-List-ID: oracle-l List-subscribe: List-owner: List-post: List-archive: X-list: oracle-l --00c09f89957ede50e5048ac89cf1 Content-Type: text/plain; charset=ISO-8859-1 First of all I'd like to say that I agree with Mark below for snapshots of databases. Or indeed hot backup full stop. I may be misunderstanding this, but I didn't think that VMWare snapshots were directly comparable to storage snapshots of filesystems, in particular since they also snapshot the memory state of the machine at the time - though this is optional. In this sense isn't a VMWare snapshot comparable to a noarchivelog mode backup (albeit probably of a running instance) - i.e you rewind the entire machine to 12:00 pm in Mark's scenario below and lose the work from 12 - 2. It seems to me that most of this discussion seems to be around storage snapshots of the database mount points. Or am I just lala again. Niall On Wed, Jul 7, 2010 at 3:28 AM, Bobak, Mark wrote: > Hi Dave, > > This is no different than any other hot backup. You put database into > backup mode, and copy off the datafiles, exit backup mode. Then, you can > use 'alter database backup controlfile....' to take a logical or physical > backup of the controlfile. But, do *not* backup the online redo logs. > > Suppose you take a hot backup (snapshot or otherwise) with tablespaces all > in backup mode, at 12:00pm. Suppose you *did* take a backup of the online > redo as part of that backup/snapshot. Now, at 2:00pm, your database > crashes. So, what happens? You restore all files from the snapshot, > *including* the online redo. Now, remember, that online redo is from > 12:00pm. You want to recover to 2:00pm, when the crash occurred. So, you > have your archivelog backups. So, you put database into recovery mode > (recover database), and start applying logs. Now, suppose the last > archivelog you apply is from 1:55pm. Where are the last remaining > transactions from 1:55pm till 2:00pm? They're in the online redo! But, > you've overwritten it, with the data from 12:00p, which is useless, and, by > the way, was already archived off, shortly after 12:00pm, and has been > applied to the database. So, now you're stuck. You lost the data from the > current online redo when you did the restore, and your > online redo is from 12:00pm (backup time), which is useless to you. So, > your only option is to do an incomplete recovery, open resetlogs, and > explain to your boss why you lost all the transactions from 1:55pm-2:00pm. > > Now, you could argue that, in the above circumstance, you simply don't > restore the online redo, and you'll be fine. That's true, (assuming that > the snapshot technology you're using will allow for you to specify which > files do and don't get restored), but, tell me, why backup the on-line redo > in the first place? Show me a valid recovery scenario, from a hot backup, > that requires a copy of the on-line redo to be in the backup set. I'm not > aware of any such scenario. Can you think of one? > > Cold backups are a bit different. I always argue against backing up online > redo with cold backup either. People always argue that the online redo is > required if the database was not consistent (i.e. cleanly shutdown) when > backup was done. And, I'd argue that if that's how you're doing cold > backups, you're doing it wrong. (Though, I think I'd first wonder why the > heck you're doing cold backups in the first place!) > > The only valid scenario that I can think of, where you *want* to include > the online redo, is if you want a *restartable* snapshot, rather than a > *recoverable* snapshot. Keep in mind, that restartable snapshot can only be > used to restore the database to the point in time when the snapshot was > taken. It's not recoverable, and cannot be rolled forward to a point in > time. > > Hope that helps, > > -Mark > ________________________________________ > From: oracle-l-bounce@freelists.org [oracle-l-bounce@freelists.org] On > Behalf Of David Roberts [big.dave.roberts@googlemail.com] > Sent: Tuesday, July 06, 2010 21:36 > To: oracle-l > Subject: Re: VMWARE Snapshots to backup Oracle > > "Danger, Will Robinson, Danger" > > Why would enabling backup mode, mean that backups of online redo logs > are not required? > > If you do a snapshot of the database while online, by implication the > backup will only be useful after recovery, and the redo logs would be > vital to guarantee the recovery? > > Perhaps, I've missed something, please correct me if I'm wrong. > > > Dave > > On Tue, Jul 6, 2010 at 5:24 PM, Bobak, Mark > wrote: > > Right. If you do begin backup/end backup, you should *not* snapshot > online > > redo, and you get a recoverable snapshot. If you do *not* do begin/end > > backup, you *should* snapshot the online redo logs, and you get > restartable > > snapshot. > > > > > > > > Hope that helps, > > > > > > > > -Mark > > > > > > > > From: oracle-l-bounce@freelists.org [mailto: > oracle-l-bounce@freelists.org] > > On Behalf Of Andrew Kerber > > Sent: Tuesday, July 06, 2010 12:19 PM > > To: post.ethan@gmail.com > > Cc: oracle-l > > Subject: Re: VMWARE Snapshots to backup Oracle > > > > > > > > I am not using it, but I understand that it works as long as you use the > > alter database begin backup command prior to taking the snapshot (and the > > alter database end backup command after taking it). > > > > On Tue, Jul 6, 2010 at 10:18 AM, Ethan Post > wrote: > > > > Anyone using snapshots to backup Oracle? Does it work? Anything I need to > > know? > > > > > > -- > > Andrew W. Kerber > > > > 'If at first you dont succeed, dont take up skydiving.' > -- > http://www.freelists.org/webpage/oracle-l > > > > > -- > http://www.freelists.org/webpage/oracle-l > > > -- Niall Litchfield Oracle DBA http://www.orawin.info --00c09f89957ede50e5048ac89cf1 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable First of all I'd like to say that I agree with Mark below for snapshots= of databases. Or indeed hot backup full stop. I may be misunderstanding th= is, but I didn't think that VMWare snapshots were directly comparable t= o storage snapshots of filesystems, in particular since they also snapshot = the memory state of the machine at the time - though this is optional. In t= his sense isn't a VMWare snapshot comparable to a noarchivelog mode bac= kup (albeit probably of a running instance) - i.e you rewind the entire mac= hine to 12:00 pm in Mark's scenario below and lose the work from 12 - 2= . It seems to me that most of this discussion seems to be around storage sn= apshots of the database mount points.

Or am I just lala again.

Niall=A0

On Wed, Jul 7, 2010 at 3:28 AM, Bobak, Mark <Mark.Bobak@proquest.com> wrote:
Hi Dave,

This is no different than any other hot backup. =A0You put database into ba= ckup mode, and copy off the datafiles, exit backup mode. =A0Then, you can u= se 'alter database backup controlfile....' to take a logical or phy= sical backup of the controlfile. =A0But, do *not* backup the online redo lo= gs.

Suppose you take a hot backup (snapshot or otherwise) with tablespaces all = in backup mode, at 12:00pm. =A0Suppose you *did* take a backup of the onlin= e redo as part of that backup/snapshot. =A0Now, at 2:00pm, your database cr= ashes. =A0So, what happens? =A0You restore all files from the snapshot, *in= cluding* the online redo. =A0Now, remember, that online redo is from 12:00p= m. =A0You want to recover to 2:00pm, when the crash occurred. =A0So, you ha= ve your archivelog backups. =A0So, you put database into recovery mode (rec= over database), and start applying logs. =A0Now, suppose the last archivelo= g you apply is from 1:55pm. =A0Where are the last remaining transactions fr= om 1:55pm till 2:00pm? =A0They're in the online redo! =A0But, you'v= e overwritten it, with the data from 12:00p, which is useless, and, by the = way, was already archived off, shortly after 12:00pm, and has been applied = to the database. =A0So, now you're stuck. =A0You lost the data from the= current online redo when you did the restore, and your
=A0online redo is from 12:00pm (backup time), which is useless to you. =A0= So, your only option is to do an incomplete recovery, open resetlogs, and e= xplain to your boss why you lost all the transactions from 1:55pm-2:00pm.
Now, you could argue that, in the above circumstance, you simply don't = restore the online redo, and you'll be fine. =A0That's true, (assum= ing that the snapshot technology you're using will allow for you to spe= cify which files do and don't get restored), but, tell me, why backup t= he on-line redo in the first place? =A0Show me a valid recovery scenario, f= rom a hot backup, that requires a copy of the on-line redo to be in the bac= kup set. =A0I'm not aware of any such scenario. =A0Can you think of one= ?

Cold backups are a bit different. =A0I always argue against backing up onli= ne redo with cold backup either. =A0People always argue that the online red= o is required if the database was not consistent (i.e. cleanly shutdown) wh= en backup was done. =A0And, I'd argue that if that's how you're= doing cold backups, you're doing it wrong. =A0(Though, I think I'd= first wonder why the heck you're doing cold backups in the first place= !)

The only valid scenario that I can think of, where you *want* to include th= e online redo, is if you want a *restartable* snapshot, rather than a *reco= verable* snapshot. =A0Keep in mind, that restartable snapshot can only be u= sed to restore the database to the point in time when the snapshot was take= n. =A0It's not recoverable, and cannot be rolled forward to a point in = time.

Hope that helps,

-Mark
________________________________________
From: oracle-l-bounce@free= lists.org [oracle-l-bo= unce@freelists.org] On Behalf Of David Roberts [big.dave.roberts@googlemail.com]
Sent: Tuesday, July 06, 2010 21:36
To: oracle-l
Subject: Re: VMWARE Snapshots to backup O= racle

"Danger, Will Robinson, Danger"

Why would enabling backup mode, mean that backups of online redo logs
are not required?

If you do a snapshot of the database while online, by implication the
backup will only be useful after recovery, and the redo logs would be
vital to guarantee the recovery?

Perhaps, I've missed something, please correct me if I'm wrong.


Dave

On Tue, Jul 6, 2010 at 5:24 PM, Bobak, Mark <Mark.Bobak@proquest.com> wrote:
> Right. =A0If you do begin backup/end backup, you should *not* snapshot= online
> redo, and you get a recoverable snapshot. =A0If you do *not* do begin/= end
> backup, you *should* snapshot the online redo logs, and you get restar= table
> snapshot.
>
>
>
> Hope that helps,
>
>
>
> -Mark
>
>
>
> From: oracle-l-bounce= @freelists.org [mailto:oracle-l-bounce@freelists.org]
> On Behalf Of Andrew Kerber
> Sent: Tuesday, July 06, 2010 12:19 PM
> To: post.ethan@gmail.com > Cc: oracle-l
> Subject: Re: VMWARE Snapshots to backup Oracle
>
>
>
> I am not using it, but I understand that it works as long as you use t= he
> alter database begin backup command prior to taking the snapshot (and = the
> alter database end backup command after taking it).
>
> On Tue, Jul 6, 2010 at 10:18 AM, Ethan Post <post.ethan@gmail.com> wrote:
>
> Anyone using snapshots to backup Oracle? Does it work? Anything I need= to
> know?
>
>
> --
> Andrew W. Kerber
>
> 'If at first you dont succeed, dont take up skydiving.'
--
htt= p://www.freelists.org/webpage/oracle-l




--
htt= p://www.freelists.org/webpage/oracle-l





--
Niall Litch= field
Oracle DBA
http://www.orawin= .info
--00c09f89957ede50e5048ac89cf1-- -- http://www.freelists.org/webpage/oracle-l