Re: backing up a big DB

From: Seth Miller <sethmiller.sm_at_gmail.com>
Date: Thu, 31 Mar 2022 10:42:28 -0500
Message-ID: <CAEueRAVcvBL83faRw_iC2-rnsT00do3MW_tnZa+xCC_kx3_ykw_at_mail.gmail.com>



Orlando,

A healthy skepticism is always good when it comes to choosing how to protect Oracle databases but consider that ostensibly every database backup in OCI is an incrementally updated backup.

Just to clarify, RESTORE VALIDATE requires the *source* database to be mounted. It does not require you to mount a new instance or restore the backup to validate the backup files.

Seth Miller
Oracle Exadata Product Management

On Tue, Mar 29, 2022 at 1:10 PM Orlando L <oralrnr_at_gmail.com> wrote:

> So it looks like you need to restore the backup to validate the backup.
>
> I am reading the manual
> <https://docs.oracle.com/en/database/oracle/oracle-database/19/bradv/validating-database-files-backups.html#GUID-18A4B00E-0D9B-410D-8ABB-2AC78DB99AA4>:
> "You can run RESTORE ... VALIDATE to test whether RMAN can restore a
> specific file or set of files from a backup. RMAN chooses which backups to
> use. The database must be mounted or open for this command" So I guess I
> can restore the control file in the test server (test server can see the
> same NFS) and run a restore validate.
>
> This line confirms it. "When validating files on disk or tape, RMAN *reads
> all blocks in the backup piece or image copy*." After incrementally
> updated backups, we can run 'restore controlfile' and run RESTORE VALIDATE
> once a month in the test server to check the backups. I think I have my
> strategy.
>
> OL.
>
> On Mon, Mar 28, 2022 at 1:36 PM Ilmar Kerm <ilmar.kerm_at_gmail.com> wrote:
>
>> <Shameless plug of my old code, that I'm trying to not use anymore>
>>
>> We have done similar thing for many-many years now, update image copy on
>> a dedicated NFS server incrementally - while letting NAS system keep the
>> long term history in snapshots. But as "the state of any backup is unknown
>> until a restore is attempted" - automating the restores is the key here.
>> We restore test all databases every day, and once a month also run a full
>> validate database on the restored copy, to catch these potentially rotten
>> bits.
>> All code is here:
>> https://ilmarkerm.eu/blog/oracle-imagecopy-backup/
>>
>> </Shameless plug of my old code, that I'm trying to not use anymore>
>>
>> As for timings - it all depends on the storage. Our all flash NAS systems
>> and all pipes in between ar pretty fast, so for small databases like 23TB
>> do not really notice the validate database command at all, it probably runs
>> a few hours. For a 300TB+ database it is already noticeable and it usually
>> runs a few days.
>>
>> I'm trying to move away from this kind of backup system lately, since
>> taking a backup is quite expensive and block change tracking keeps having
>> nasty bugs occasionally (19c), up to freezing the entire database. Seems
>> to be easier just to create another dataguard for storing database on a
>> separate storage with long retention time (aka backup) and have NAS
>> snapshot it internally on a few hour schedule to keep long retention time.
>> Writing restore tests for it ATM.
>>
>>
>> On Mon, Mar 28, 2022 at 6:06 PM Orlando L <oralrnr_at_gmail.com> wrote:
>>
>>> Hi
>>>
>>> We have a 23TB Oracle database and the full backup times are a problem.
>>> We are currently backing it up to an NFS on weekends. I am trying to see
>>> options on cutting down the time. I am looking into incrementally updated
>>> backups, which I think may cut down the backup time drastically. I am
>>> concerned about the long run though. Since it copies over only the changed
>>> data, I am wondering what will happen if some not-frequently accessed block
>>> in the backup goes corrupt in the backup. I am thinking that it may be a
>>> problem when it is time to do a restore. Am I warranted in this kind of
>>> thinking? I am wondering about the VALIDATE command used on a backup of a
>>> big DB of this size. Anyone uses VALIDATE on such big backups? How long
>>> does it take. All ideas welcome. 19c.
>>>
>>> PS. No money for BCV or a parallel dataguard server to offload backups.
>>>
>>> Orlando.
>>>
>>>
>>
>> --
>> Ilmar Kerm
>>
>

--
http://www.freelists.org/webpage/oracle-l
Received on Thu Mar 31 2022 - 17:42:28 CEST

Original text of this message