[prev in list] [next in list] [prev in thread] [next in thread]
List: veritas-bu
Subject: RE: [Veritas-bu] EMC BCV Question
From: "Donaldson, Mark" <Mark.Donaldson () experianems ! com>
Date: 2002-10-30 19:06:46
[Download RAW message or body]
I've append a previous posting I did earlier on this subject.
Yes, the BCV image can used for recovery to a later point in time, you just
have to have the archived redo & redo logs handy. I backup mine hourly to
tape anticipating a big disaster.
I don't agree with the idea that you shouldn't put the Oracle DB in
hot-backup mode. Being in hot backup mode guarantees consistency across
data files. It just seems safe to do so. You should check with a DBA on
this. The good news is that the DB only has to be in Hot backup mode for
minutes, just long enough to snap the BCV's. The redo activity for such a
short period might be negligible.
Anyway, here's my previous posting on this subject. EMC has a white paper
on this, too, you should ask your rep to dig it up for you.
-Mark
-----------------
I backup EMC BCV's now.
Currently, I have a ~300M database that is on an EMC 3830. Each standard
mirror for the data disks has an associated BCV. These BCV's are not
visible (via port mapping) to the same host that can "see" the Standards.
In other words, when my BCV's are split, those luns are not visible to my
database box. They are visible, however, to my backup server. Once I split
the BCV's, I then import the disk group to my backup box, start the volumes,
fsck, & mount the filesystems. I then just do an ordinary file backup to
tape.
The fine detail order looks like this:
To start the backup process (part of a bpstart_notify process):
1. Backup Oracle Controlfile to trace
2. Place Oracle tablespaces in Hotbackup mode
3. Split BCV's for data volumes only.
4. End Hotbackup mode for tablespaces
5. Split BCV's associated with Oracle archived redo
6. vxdctl enable on backup server (unnecessary on some
servers - I think it's scsi chipset on controller)
7. import (with "Clear") the disk groups for data and
archived redo (vxdg -C import <dgname>) on backup
server. The volumes, plexes, & subdisk config
comes along for the ride on the import.
8. recover the volumes (vxrecover -g <dgname> -sv)
9. start the volumes (vxvol -g <dgname> startall)
10. fsck the volumes
11. mount them locally on the backup box for simple file
backup to tape.
To reverse this after backup (bpend_notify):
1. unmount the filesystems
2. stop the volumes ( vxvol -g <dgname> stopall )
3. deport the disk groups ( vxdg deport <dgname> )
4. Establish the BCV's
Let me know if you want more
-----Original Message-----
From: Lee, Kenneth [mailto:Kenneth.Lee@sbs.siemens.com]
Sent: Wednesday, October 30, 2002 10:01 AM
To: 'veritas-bu@mailman.eng.auburn.edu'
Subject: [Veritas-bu] EMC BCV Question
First let me apologize for asking this question as it is only somewhat
related to NB. We are in the process of implementing EMC BCV to backing up
an Oracle database (500GB). I was told that you could create a BCV while
Oracle is live (no shutdown or hot backup mode required) and I agree with
this statement. What I am not sure of is can you roll-forward with this
BCV. For example, at 1:00, you snap your BCV and the production Oracle
continues to run (no stoppage). At 4:00, something bad happens to
production Oracle. You restore your BCV on top of production, can you
roll-forward the production logs up to 4:00? The reason I am puzzle is
because when you snap the BCV at 1:00, the Oracle database (BCV) is
startable but will go into recovery mode. It will rollback in-flight
transactions. Can you apply the production logs to this database to bring
it to 4:00?
Thanks,
Ken
_______________________________________________
Veritas-bu maillist - Veritas-bu@mailman.eng.auburn.edu
http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu
[Attachment #3 (text/html)]
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
<HTML>
<HEAD>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=iso-8859-1">
<META NAME="Generator" CONTENT="MS Exchange Server version 5.5.2653.12">
<TITLE>RE: [Veritas-bu] EMC BCV Question</TITLE>
</HEAD>
<BODY>
<P><FONT SIZE=2>I've append a previous posting I did earlier on this subject. \
</FONT> </P>
<P><FONT SIZE=2>Yes, the BCV image can used for recovery to a later point in time, \
you just have to have the archived redo & redo logs handy. I backup mine \
hourly to tape anticipating a big disaster. </FONT></P>
<P><FONT SIZE=2>I don't agree with the idea that you shouldn't put the Oracle DB in \
hot-backup mode. Being in hot backup mode guarantees consistency across data \
files. It just seems safe to do so. You should check with a DBA on \
this. The good news is that the DB only has to be in Hot backup mode for \
minutes, just long enough to snap the BCV's. The redo activity for such a short \
period might be negligible. </FONT></P>
<P><FONT SIZE=2>Anyway, here's my previous posting on this subject. EMC has a \
white paper on this, too, you should ask your rep to dig it up for you.</FONT></P>
<P><FONT SIZE=2>-Mark</FONT>
<BR><FONT SIZE=2>-----------------</FONT>
<BR><FONT SIZE=2>I backup EMC BCV's now.</FONT>
</P>
<P><FONT SIZE=2>Currently, I have a ~300M database that is on an EMC 3830. Each \
standard mirror for the data disks has an associated BCV. These BCV's are not \
visible (via port mapping) to the same host that can "see" the \
Standards. In other words, when my BCV's are split, those luns are not visible \
to my database box. They are visible, however, to my backup server. Once \
I split the BCV's, I then import the disk group to my backup box, start the volumes, \
fsck, & mount the filesystems. I then just do an ordinary file backup to \
tape. </FONT></P>
<P><FONT SIZE=2>The fine detail order looks like this:</FONT>
</P>
<P><FONT SIZE=2>To start the backup process (part of a bpstart_notify \
process):</FONT> <BR><FONT SIZE=2>1. Backup Oracle Controlfile to trace</FONT>
<BR><FONT SIZE=2>2. Place Oracle tablespaces in Hotbackup mode</FONT>
<BR><FONT SIZE=2>3. Split BCV's for data volumes only.</FONT>
<BR><FONT SIZE=2>4. End Hotbackup mode for tablespaces</FONT>
<BR><FONT SIZE=2>5. Split BCV's associated with Oracle archived redo</FONT>
<BR><FONT SIZE=2>6. vxdctl enable on backup server (unnecessary on some</FONT>
<BR><FONT SIZE=2> servers - I think it's scsi chipset on \
controller)</FONT> <BR><FONT SIZE=2>7. import (with "Clear") the disk \
groups for data and</FONT> <BR><FONT SIZE=2> archived redo (vxdg -C \
import <dgname>) on backup</FONT> <BR><FONT SIZE=2> server. \
The volumes, plexes, & subdisk config</FONT> <BR><FONT SIZE=2> comes \
along for the ride on the import.</FONT> <BR><FONT SIZE=2>8. recover the volumes \
(vxrecover -g <dgname> -sv)</FONT> <BR><FONT SIZE=2>9. start the volumes (vxvol \
-g <dgname> startall)</FONT> <BR><FONT SIZE=2>10. fsck the volumes</FONT>
<BR><FONT SIZE=2>11. mount them locally on the backup box for simple file</FONT>
<BR><FONT SIZE=2> backup to tape.</FONT>
</P>
<P><FONT SIZE=2>To reverse this after backup (bpend_notify):</FONT>
<BR><FONT SIZE=2>1. unmount the filesystems</FONT>
<BR><FONT SIZE=2>2. stop the volumes ( vxvol -g <dgname> stopall )</FONT>
<BR><FONT SIZE=2>3. deport the disk groups ( vxdg deport <dgname> )</FONT>
<BR><FONT SIZE=2>4. Establish the BCV's</FONT>
</P>
<P><FONT SIZE=2>Let me know if you want more</FONT>
</P>
<P><FONT SIZE=2>-----Original Message-----</FONT>
<BR><FONT SIZE=2>From: Lee, Kenneth [<A \
HREF="mailto:Kenneth.Lee@sbs.siemens.com">mailto:Kenneth.Lee@sbs.siemens.com</A>]</FONT>
<BR><FONT SIZE=2>Sent: Wednesday, October 30, 2002 10:01 AM</FONT>
<BR><FONT SIZE=2>To: 'veritas-bu@mailman.eng.auburn.edu'</FONT>
<BR><FONT SIZE=2>Subject: [Veritas-bu] EMC BCV Question</FONT>
</P>
<BR>
<P><FONT SIZE=2>First let me apologize for asking this question as it is only \
somewhat</FONT> <BR><FONT SIZE=2>related to NB. We are in the process of \
implementing EMC BCV to backing up</FONT> <BR><FONT SIZE=2>an Oracle database \
(500GB). I was told that you could create a BCV while</FONT> <BR><FONT \
SIZE=2>Oracle is live (no shutdown or hot backup mode required) and I agree \
with</FONT> <BR><FONT SIZE=2>this statement. What I am not sure of is can you \
roll-forward with this</FONT> <BR><FONT SIZE=2>BCV. For example, at 1:00, you \
snap your BCV and the production Oracle</FONT> <BR><FONT SIZE=2>continues to run (no \
stoppage). At 4:00, something bad happens to</FONT> <BR><FONT SIZE=2>production \
Oracle. You restore your BCV on top of production, can you</FONT> <BR><FONT \
SIZE=2>roll-forward the production logs up to 4:00? The reason I am puzzle \
is</FONT> <BR><FONT SIZE=2>because when you snap the BCV at 1:00, the Oracle database \
(BCV) is</FONT> <BR><FONT SIZE=2>startable but will go into recovery mode. It \
will rollback in-flight</FONT> <BR><FONT SIZE=2>transactions. Can you apply the \
production logs to this database to bring</FONT> <BR><FONT SIZE=2>it to 4:00?</FONT>
</P>
<P><FONT SIZE=2>Thanks,</FONT>
<BR><FONT SIZE=2>Ken</FONT>
</P>
<P><FONT SIZE=2>_______________________________________________</FONT>
<BR><FONT SIZE=2>Veritas-bu maillist - \
Veritas-bu@mailman.eng.auburn.edu</FONT> <BR><FONT SIZE=2><A \
HREF="http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu" \
TARGET="_blank">http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu</A></FONT> \
</P>
</BODY>
</HTML>
_______________________________________________
Veritas-bu maillist - Veritas-bu@mailman.eng.auburn.edu
http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu
[prev in list] [next in list] [prev in thread] [next in thread]
Configure |
About |
News |
Add a list |
Sponsored by KoreLogic