Oracle FAQ | Your Portal to the Oracle Knowledge Grid |
Home -> Community -> Usenet -> c.d.o.server -> Re: HP MC / Service Guard / Oracle 8i
> > That is, will my secondary Oracle database (running on my secondary
> > server) be automatically switched over to in the event that my primary
> > server goes down?
>
> Only with RAC/OPS.
Incorrect.
In case of primary server failure you can start up *the same* database on another instance. No RAC needed, you can do it even with Standard Edition.
In case you want to have (almost) synchronized copy of your DB on second server, you can go with Standby Database.
>
> >
> > Is there no disruption to my connected users to the database? This is,
> > is there a transparation failover for the clients?
>
> Only with RAC/OPS.
Incorrect. No disruption is a myth. There IS disruption. Even it TAF is used, some transactions get cancelled, not to talk about package stated etc.
If you want *absolutely no* disruption, that means true redundancy, hire an
additional user for each existing user and let them work with 2 "identical"
systems parallelly- that means duplicate the transactions into different
systems :)
Im almost sure, that noone in world is doing such thing, so the truth might
be somewhere between, i.e. transaction server distributes transactions to
two different databases, or even GUI client will use two or more identical
transaction servers...
I normally prefer not to buy and use all kinds of expencive GUI software for
doing a simple thing.
If you have a RAID array which is connected to both servers, then in case of
first server failure you can simply mount the disks to second server and
start up your database. No Veritas or additional expensive software needed.
Tanel.
>
> >
> > Do I need a Oracle OPS (=Oracle Parallel Server) or Oracle RAC (=Real
> > Application Cluster) or something else in order to make all this
> > possible? What about the handling of control-files?
>
> To do what you want will require RAC and OPS and Veritas Cluster volume
manager.
> The software will cost you about 30-40K per node. Its not cheap but it is
> effective.
> >
> > Thanks for help in advance,
> >
> > Christian Hartmann
>
>
> Jim Krol
> james.krol_at_boeing.com
Received on Thu Feb 27 2003 - 11:12:02 CST