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: Full Mirroring and Oracle?

Re: Full Mirroring and Oracle?

From: Neil Greene <ngreene_at_laoc.SHL.com>
Date: 1997/06/14
Message-ID: <MPG.e0c80b3cbd5787898969c@news.internetMCI.com>#1/1

In article <01bc770a$1c721c70$0b320089_at_dbasrv>, mkoppel_at_barakitc.co.il says...
> Hello,
>
> Wh have two unix machines, one Primary, the other one Backup.
> Full RAID is implemented, so each bit written to the Primary is backed up
> to the Backup simultaneously.

I don't understand this. You have two machines. Are they sharing the same disk arrays, mounting the same volumes? Or, are you running OPS? Is this a clustered environment? Some other useful information would be your Unix version and what you are using for RAID management.

> Let alone, the Backup machine has an equal File System structure and naming
> conventions.

Sounds like you answer my question above by saying the disk arrays are not being shared and are actually different. Is this true?

> Will I be able to start the Oralce on the Backup Machine after a Primary
> crash?
>
> (I'm trying to avoid using a standby database).

I don't yet understand how your primary and secondary are sharing the same datafiles? You can do the following under Solaris:

	Place all of your production level file systems, data files and 
	required filesystems in a DISK GROUP known as foo.  Your
	production storage arrays need to be dual hosted to both your
	boxes.  So, on one primary machine you see X arrays, on the 
	secondary machine you see X arrays plus any Y arrays already 
	attached.

	Now, since your production arrays are hosted on your failover
	machines, assuming you are simply talking about a CPU failure
	in this case, you can use Volume Manager to deport/export
	disk groups and import them onto another host, in this case
	your secondary machine.  This can be accomplished while both 
	machines are available or for fail over cases where your primary
	is now dead.
	
	Now, your secondary or fail-over host has imported your
	production disk assignments and all you need is the mount
	point information for mounting the volumes.  You can write
	scripts to grab this information nightly to automate the
	mounting of the new volumes.  Or, I can send you the unsupported
	grabdg script from Sun if you are using Solaris/Volume Manager.

	Access to your custom apps, database and required files as normal.
	You will also have to create or pre-create any custom userids
	required or manage anything that is different about the two 
	machines.  Using ifconfig, you can set up virtual interfaces
	(le0:1) so that your secondary box answers to the ip assignment of
	your production box.  Start your listeners and walla, you are 
	done!

-- 
Neil Greene			Senior System Engineer / Oracle DBA	
MCI Systemhouse, Inc. 		<mailto:ngreene_at_laoc.SHL.com>

-----BEGIN PGP SIGNATURE-----
Version: 2.6.2

mQCNAzNWiBwAAAEEALmJ2Zho8BRcFc6vOHuUJp1TJ4+fZsmDgvi57DfDcnVCIcEv
e//qQ185dRN03821V7+MwfdKT51KOFcKRnHKFe8xGdDgMCB73ZFUn6X0acn3dVKn
K7kSTLpjcqlwGcEQb5MsH2oPz2ejUZ4+BghTN66nrZsEptkZOI+PVZH4HYSBAAUR
tCJOZWlsIEdyZWVuZSA8bmdyZWVuZUBsYW9jLlNITC5jb20+
=kEst
-----END PGP SIGNATURE---------------
Received on Sat Jun 14 1997 - 00:00:00 CDT

Original text of this message

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