Re: Few complicated Questions in Oracle

From: Matthias Gresz <GreMa_at_t-online.de>
Date: 1998/02/11
Message-ID: <6bs5hk$dhv$2@news02.btx.dtag.de>#1/1

On Tue, 10 Feb 1998 20:16:23 +0530, Ram Prasad <rprasad_at_baan.com> wrote:

(3)

backing up online redo locks is useless and dangerous, when using cold backups. Scenario:

state1:

	archived logs:		online (not archived):
		125			126
		124			127
		123
		122
		121...

This is the state you back up.

Your db crashes due to a disk-error that affects one or some of your tables spaces.

State2: Situation when db crashes.

	archived logs:		online (not archived):
		135			136
		134			137
		133
		132
		131...

So you'll restore from tape or s.e.. If you'd also restore the backed up online redo logs you'll get

State3:

	archived logs:		online (not archived):
		135			126
		134			127
		133
		132
		131...

Thus beeing not capable to recover your db near the point it crashed.

To get all the logged changes backed up in a manner that they can be reused, issue (number_of_online_logs) + 1 -times:

ALTER SYSTEM SWITCH LOGFILE; before starting cold backup.

This way all the changes made were archived. Thus a lost of archived redologs could be conquered, if you also will periodiquely copy newly created (not yet backed up) archive logs to a second location.

--
Regards
Matthias Gresz




>

>  (1)  Why creating an index is slower than creating a table ?

>  (2)   Is it posible for an instance on one server to mount the

>database on another server           in the absense of OPS architecture

>- Using NFS mount technology ?

>  (3)  Is it beneficial to backup an online redo log file , In other

>words an we recover from the loss of an online redo log file using

>previous backup ? If not why ?

>

>  Please mail your replies to tkrishna_at_baan.com

>

>  Thanks in advance.

>


--

Regards

Matthias Gresz    :-)

Received on Wed Feb 11 1998 - 00:00:00 CST