Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: clean up ocr?

Re: clean up ocr?

From: herta <herta.vandeneynde_at_gmail.com>
Date: Sun, 01 Jul 2007 00:33:17 -0000
Message-ID: <1183249997.831981.194750@k29g2000hsd.googlegroups.com>


Miaden,

That is probably true for many situations. Not for ours.

Re. "Machines are expensive."
As already described, the servers that will be moved to the DR site no longer have hardware support (they're old production recups), so a bit of redundancy is welcome. Performance wise, they more or less equal half of our production environment. We estimate that, in case of disaster, they'll be able to run the most critical databases until extra systems can be brought in. In the meantime, we want to spread the standby databases over the two cluster members, and run our backups on them as well.
And true, expenses are a tad higher because of that choice.

Re. "Likelihood is that you're using some kind of SAN and RAID 1+0" The DR site uses NAS. The idea is to move the data to SAN if we do have a disaster. Since we use ASM, we should be able to do that online.

But that's neither here not there, as it has running Data Gaurd on a single system or in a RAC environment has no impact on the storage solution we chose.

Re. "why would you want to have 2 machines doing nothing together (RAC), instead of having one machine and then adding a spare" Yes, you can start of with a single, non-RAC system and add a spare when you have a disaster. Downside is that the procedure to do that not only is slightly more complex, but would also require another downtime.
We start our BCP (Business Continuity Plan) scenario from the perspective that the DBAs may not be around to do the recovery, so we want to keep the procedure as simple as possible. And even if the DBAs are around, let's assume we restart production in the DR site. Critical applications will be restarted first. I wouldn't want to take them down again to add them to a cluster. Several applications aren't RAC aware, and restarting them unfortunately can take hours. (Don't ask. I'm not involved in that part.)

Re. "The other machine can, in the meantime, be used for development, file server or as a doom3/battlezone server." The company I currently work for operates in a niche market, selling information. Development/Test/Acceptance is a critical part of the business. We have state of the art equipment for that. We don't need file servers as we already have NAS. doom3/battlezone? Hmm. Perhaps I should have a word with our manager.

Kind regards,

Herta

On Jun 27, 6:19 pm, Mladen Gogala <mgogala.SPAM..._at_not-at-verizon.net> wrote:
> On Wed, 27 Jun 2007 07:39:30 +0000,hertawrote:
> > On Jun 27, 5:30 am, DA Morgan <damor..._at_psoug.org> wrote:
> >> Mladen Gogala wrote:
> >> > On Tue, 26 Jun 2007 14:49:33 +0000,hertawrote:
>
> >> That's not an explanation. I understand why you would want a DR site. I
> >> understand why you might want a RAC cluster or Data Guard at a DR site.
> > (...)
>
> >> > The manual says you can, but it doesn't explain why would you do
> >> > something like that.
>
> >> Which is the point of my questions. The fact that I can hit my head on
> >> a rock doesn't make me want to do it: At least not today.
> > (...)
>
> > Curious, you can understand people wanting a DR site, you can understand
> > them wanting a RAC cluster "or" Data Guard at the DR site. But you
> > cannot understand them wanting both a RAC cluster "and" Data Guard at
> > the DR site? Sorry, but I don't think I can explain better than I did.
>
> > The point of my question is not to discuss business cases which are
> > perfectly clear to me.
> > My reason for posting was to see if there is a way to clear out a single
> > entry from an OCR.
>
> > Kind regards,
>
> >Herta
>
> Herta, the problem is that with a physical standby, machine(s) doing
> recovery are inactive, standing by. Thus the name. Machines are expensive.
> You have to pay licence fees, electricity, allocate space (in my case
> space is on the Times Square, New York City, which is not known for being
> cheap) and spare parts. Likelihood is that you're using some kind of SAN
> and RAID 1+0, so you must have a spare disk or two. My question is, simply
> put, why would you want to have 2 machines doing nothing together (RAC),
> instead of having one machine and then adding a spare, should the need
> arise. The other machine can, in the meantime, be used for development,
> file server or as a doom3/battlezone server.
>
> --http://www.mladen-gogala.com
Received on Sat Jun 30 2007 - 19:33:17 CDT

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US