Re: Combining snapclones with DataGuard

On Oct 21, 11:14 pm, Michael Austin <maus..._at_firstdbasource.com> wrote:
> macdba321 wrote:
> > On Oct 21, 8:02 am, jj3run..._at_gmail.com wrote:
> >> On Oct 20, 10:05 pm, DA Morgan <damor..._at_psoug.org> wrote:
>
> >>> macdba321 wrote:
> >>>> I read quotes from Ari Kaplan explaining how enterprises if possible
> >>>> should (if possible and appropriate) combine the use of snapclones
> >>>> (for SAN-to-SAN replication) with DataGuard.
> >>>> I am confused how an organization could use both of these though. If I
> >>>> am using snapclones, how can I also use DataGuard? What use would DG
> >>>> provide? (Assuming that I am already snapclone'ing the database to the
> >>>> secondary site.)
> >>>> Can anyone shed some light for me?
> >>>> Thanks.
> >>> What also is missing, in addition to the items Sybrand listed, is the
> >>> context.
> >>> For a DR site I wouldn't touch snap clones as doing so one loses some
> >>> of the most critical components of Data Guard Oracle brings to the
> >>> table.
> >>> Snapclones have their place but it isn't in DR. A DR site must be not
> >>> just a 100% clean duplicate of production it must also be logically
> >>> consistent. Something Snapcloning can not guarantee as snapping will
> >>> efficiently replicate corrupt blocks.
> >>> --
> >>> Daniel A. Morgan
> >>> Oracle Ace Director & Instructor
> >>> University of Washington
> >>> damor..._at_x.washington.edu (replace x with u to respond)
> >>> Puget Sound Oracle Users Groupwww.psoug.org
> >> Group,
>
> >> As a long, long, long, time lurker, this is a horribly impolite
> >> thread. I wish the posters here would ease up and try to share
> >> knowledge instead of trying to be as mean as possible. Perhaps some of
> >> us are new to the Oracle world, and others have been around for a few
> >> years or decades, it should not matter. Try to help each other and
> >> leave your egos at the door.
>
> >> When I read the original posters question, I interpreted it as a
> >> generic discussion. He/She (yes, there are actually females on here!)
> >> did not provide versions or platforms. Nor did Ari Kaplan! It's a
> >> generic discussion, how is DataGuard used in conjunction with
> >> Snapcloning (in an effective manner)?
>
> >> Yes, we all know that using old versions of the database, old
> >> networking capabilities, old OS's, old keyboards, old mice, old .....
> >> (you get the idea) will cause problems. Just assume we're talking
> >> generic.
>
> >> To the originial poster: at least you apologized and attempted to
> >> explain yourself (albeit, somewhat sarcastically).
>
> >> To those with the egos: Try to not be so rude, you may be new to the
> >> Oracle or professional arena ... I have a feeling you'll go further if
> >> you relax and try to help. But most importantly, feel free to not
> >> respond if you have nothing constructive to add. Silence definitely
> >> can be golden!
>
> >> $Jen
>
> > Well, as the original OP, I want to apologize to all. I did not intend
> > to cause such a stir.
> > Let me provide some lost information: The OS is Windows 2003 64-Bit.
> > It is running 11.2.0.3.1 (64-bit). There are no SANs (yet). The
> > potential customer will buy 2 based on the pros/cons capabilities that
> > my team outlines. They are a massive company and are planning on true
> > enterprise class SANs (to be used to rearchitect the infrastructure.
> > It is theoretical now, so saying the "latest & greatest" is not a
> > stretch.
>
> > My goal is simply high-high-level research for disaster recovery
> > situational setups for the Oracle portion. There will be >1000miles
> > between sites.
>
> > I read Mr. Kaplan's presentation (and yes, I did e-mail him days
> > ago ... but no response. I do not feel comfortable calling him via
> > telephone). I saw his statement that storage methods of backup are
> > "sometimes undiscovered in the world of DBAs". Mr. Kaplan goes on to
> > say that array-based replication "Works best in conjunction with Data
> > Guard by jointly reducing replication traffic and reducing or
> > eliminating single points of failure" .... is he most likely referring
> > to non-Oracle database files for the array-based replication, and DG
> > for the database?
>
> > Also, Daniel A. Morgan states: "A DR site must be not just a 100%
> > clean duplicate of production it must also be logically consistent.
> > Something Snapcloning can not guarantee as snapping will efficiently
> > replicate corrupt blocks." I have read that DG can be used to detect &
> > repair corrupt blocks (http://www.oracle.com/technology/deploy/
> > availability/pdf/corruption_wp.pdf). I have read that snap cloning
> > software is able to detect a level of corruption (if it can not read
> > the block, it can not propogate it to the destination filer.) Are
> > these 2 levels of corruption? (Is DG able to detect a level of
> > corruption that snap cloning software can not?
>
> > Again, thank you for any information.
>
> Just an observation - and not intending to start an OS war, but, if you
> are going to use enterprise class storage, why in the world would you
> ever deploy this on Windows?  None of the Windows databases -including
> Oracle could ever hope to keep up with say, Solaris, AIX , HPUX or even
> Linux - and is far more manageable (IMO) than a Windows environment.
> Windows and even Oracle on Windows has it's place, I, however would not
> consider the data center one of them...

That's a great observation, unfortunately all I can do is educate and make recommendations. The customer already had a massive investment in windows (computers and people), they're over 500,000 employees + contractors/consultants, and have exclusivity licensing deals with Microsoft. I would much prefer to be on Solaris, but all I can do is educate and recommend.

This has been a great thread. We got a little off-topic sure, but I'm glad some new folks came out of the woodwork.

I just wish testing something like DG combined with snapclones was as easy as testing something like encrypting a table! Received on Wed Oct 22 2008 - 07:32:36 CDT