Re: "Correct" term for a 1:1 relationship between a "database" and an "instance" where > 1 such things are on the same physical server?
From: Mark D Powell <Mark.Powell_at_eds.com>
Date: Tue, 28 Jul 2009 06:55:51 -0700 (PDT)
Message-ID: <c75ea68a-1865-4418-aac2-eb1184d22430_at_i6g2000yqj.googlegroups.com>
On Jul 24, 7:11 pm, joel garry <joel-ga..._at_home.com> wrote:
> On Jul 24, 12:58 pm, Mark D Powell <Mark.Pow..._at_eds.com> wrote:
>
>
>
>
>
> > On Jul 22, 10:29 am, ddf <orat..._at_msn.com> wrote:
>
> > > On Jul 22, 8:37 am, dana <dana_at_w..._at_yahoo.com> wrote:
>
> > > > Here's another question for you:
>
> > > > 1) Would there ever be any practical reason for two instances (procs +
> > > > SGA) to access the same database (collection of data treated as a
> > > > unit) on the same, unpartitioned physical server?
>
> > > > Dana
>
> > > > Dana
>
> > > That, I believe, is the basic definition of RAC -- two or more
> > > instances accessing a single database.
>
> > > David Fitzjarrell
>
> > But with RAC you would expect each instance to be running concurrently
> > on different servers. While some Oracle 'experts' have managed to
> > create a RAC setup with multiple instances on a single server the set
> > up is non-standard, unsupported, and for demonstartion purposes only.
>
> > On a UNIX platform it used to be fairly easy to change the instance
> > idenifier, SID, that was used to identify a running Oracle instance
> > used to access a database. That is you could shut the instance down,
> > make a few quick changes, and start a differently named instance then
> > access the same database you were just working with from a different
> > instance name. Only one instance can access a non-RAC database at a
> > time. There was little practical application for this functionality.
>
> > HTH -- Mark D Powell --
>
> Wasn't there something about oltp tuning during the day, then batch/
> report at night? Something is poking at the back of my brain, maybe
> it wasn't oracle.
>
> jg
> --
> _at_home.com is bogushttp://www.reuters.com/article/rbssTechMediaTelecomNews/idUSN21425536...- Hide quoted text -
>
> - Show quoted text -
Date: Tue, 28 Jul 2009 06:55:51 -0700 (PDT)
Message-ID: <c75ea68a-1865-4418-aac2-eb1184d22430_at_i6g2000yqj.googlegroups.com>
On Jul 24, 7:11 pm, joel garry <joel-ga..._at_home.com> wrote:
> On Jul 24, 12:58 pm, Mark D Powell <Mark.Pow..._at_eds.com> wrote:
>
>
>
>
>
> > On Jul 22, 10:29 am, ddf <orat..._at_msn.com> wrote:
>
> > > On Jul 22, 8:37 am, dana <dana_at_w..._at_yahoo.com> wrote:
>
> > > > Here's another question for you:
>
> > > > 1) Would there ever be any practical reason for two instances (procs +
> > > > SGA) to access the same database (collection of data treated as a
> > > > unit) on the same, unpartitioned physical server?
>
> > > > Dana
>
> > > > Dana
>
> > > That, I believe, is the basic definition of RAC -- two or more
> > > instances accessing a single database.
>
> > > David Fitzjarrell
>
> > But with RAC you would expect each instance to be running concurrently
> > on different servers. While some Oracle 'experts' have managed to
> > create a RAC setup with multiple instances on a single server the set
> > up is non-standard, unsupported, and for demonstartion purposes only.
>
> > On a UNIX platform it used to be fairly easy to change the instance
> > idenifier, SID, that was used to identify a running Oracle instance
> > used to access a database. That is you could shut the instance down,
> > make a few quick changes, and start a differently named instance then
> > access the same database you were just working with from a different
> > instance name. Only one instance can access a non-RAC database at a
> > time. There was little practical application for this functionality.
>
> > HTH -- Mark D Powell --
>
> Wasn't there something about oltp tuning during the day, then batch/
> report at night? Something is poking at the back of my brain, maybe
> it wasn't oracle.
>
> jg
> --
> _at_home.com is bogushttp://www.reuters.com/article/rbssTechMediaTelecomNews/idUSN21425536...- Hide quoted text -
>
> - Show quoted text -
Yes. You would use a separate pfile then just referred to an init.ora so you would want to use a separate instance identifier, SID, at the OS level so that Oracle would select the right init.ora automatically. The alternate was to use the same SID but specify the init.ora via the startup command.
- Mark D Powell --