Re: Datagaurd snapshots in lieu of backups?

From: Kenny Payton <k3nnyp_at_gmail.com>
Date: Sat, 31 May 2014 05:53:21 -0400
Message-ID: <CAEidWqOZCiO3wp4_g8EaPGSTiSEPrMP=-j5gmzOL4fs4UkdKpQ_at_mail.gmail.com>


  1. block recover needs a full and archived logs, it does not use incrementals. Prior to the change the block we be restored from our weekly full and recovery would require all of the archives logs since the full to be restored and rolled through. The snaps are taken every 2 hours, which serve as the full, and the archived logs are available in their normal format via nfs on the netapp.
  2. We pause the apply process to quiesce the environment.

Kenny
On May 31, 2014 4:35 AM, "Maris Elsins" <elmaris_at_gmail.com> wrote:

> Hi,
>
> I'd like to hear a little more details on
>
> > block level recovery have been dramatically improved.
> 1) How and why did it improve?
>
> 2) How do you take the snapshots on the standby so that you wouldn't be
> affected by Fractured Block issues (as Laimutis pointed out).
>
>
> ---
> Maris Elsins
> _at_MarisElsins <https://twitter.com/MarisElsins>
> www.facebook.com/maris.elsins
>
>
>
>
> On Sat, May 31, 2014 at 3:32 AM, Kenny Payton <k3nnyp_at_gmail.com> wrote:
>
>> I thought I would provide an update on this, since I find it pretty cool
>> and very useful for those using local physical standby databases.
>>
>> I will not go into a ton of detail but If you're interested in more let
>> me know.
>> We have officially migrated four 10+T databases to this model and no
>> longer do RMAN backups on the primaries, except for autobackup of spfiles
>> and control files. We have improved our RPO and RTO in addition to
>> eliminating the backup overhead. Things like block level recovery have
>> been dramatically improved.
>>
>> To re-iterate, our production is not on NetApp storage, only our standby
>> has snapshot capabilities. Standby database snapshots are serving as the
>> source of the primary database backups.
>>
>> Here are some key benefits.
>>
>> No RMAN backups
>>
>> Reduced recovery times since we take regular snapshots that can be used
>> as fulls ( every 2 hours as opposed to weekly).
>>
>> Snapshots can be used to easily spin up thin test copies of the standby
>> without interrupting apply processing.
>> These test instances can be used for data recovery or point in time
>> comparisons.
>>
>> And so you know, I am not at all affiliated with NetApp and they are
>> certainly not the only vendor providing snapshot solutions. I'm just happy
>> to say I no longer run backups on production and wanted to share.
>>
>> Kenny
>>
>>
>>
>>
>> On Mar 25, 2014 9:00 AM, "Kenny Payton" <k3nnyp_at_gmail.com> wrote:
>> >
>> > We've tested and are starting to move some of our Data Guard physical
>> standby's to storage with snapshot capabilities, specifically Netapp. Our
>> plan is to use snapshots and transported log files in Lieu of production
>> backups. Anyone else doing this? I'm not talking about backups of the
>> standby, just using the storage snapshots similar to the way you would use
>> image file copies. Archived logs will be used to roll forward any snapshot
>> and we're planning on taking daily, potentially hourly, snapshots to meet
>> our recovery window requirements.
>> >
>> > I've successfully tested database, datafile and block level recovery
>> operations. The approach provides a number of advantages including
>> maximizing productivity of capital costs of the standby environments,
>> reducing primary database resource consumption as well as providing a
>> shorter RTO. An additional benefit is the ability to bring up a copy of
>> the standby, via Flex Clones, in read/write mode for testing without
>> interrupting the continuity of our DG apply processes.
>> >
>> > We do not currently ship anything offsite, we have remote DR DG copies
>> for that purpose.
>> >
>> > Thanks,
>> > Kenny
>>
>
>

--
http://www.freelists.org/webpage/oracle-l
Received on Sat May 31 2014 - 11:53:21 CEST

Original text of this message