Consistent Backup

Dear Friends,
I am learning backup&recovery. Sometimes I feel confused during reading the guide. Here are 2 question:
1. When I take a tablespace offline normally, oracle takes a checkpoint to the tablespace's datafiles. So, All the committed data should be written to these files from online redo log, right? And these files may contain some uncommitted data, right?
2. If I shutdown the database with immediate option. Does the data files in this database contain uncommitted data? If yes, when I want to do a whole consistent backup, that's why I also need to backup the undo tablespace? is my understanding right?
Thanks,
Ricky

from the documentation regarding question A
When a tablespace goes offline, Oracle does not permit any subsequent SQL statements to reference objects contained in that tablespace. Active transactions with completed statements that refer to data in that tablespace are not affected at the transaction level. Oracle saves rollback data corresponding to those completed statements in a deferred rollback segment in the SYSTEM tablespace. When the tablespace is brought back online, Oracle applies the rollback data to the tablespace, if needed.

Similar Messages

Consistent Backup Of Online Database

Hi
How can I take consistent backup of online database.
If I open database in read only mode and take RMAN backup + Full Cold Backup from operating system commands.
Did the above backups will be consistent?
If yes please advise how to alter the status of system to read only.
Wishes
Jawad

Cold backups are known to be a consistent backups , while Hot Backups are not considerd as conssistent backups, you required archivelogs to make them consistent . I am bit strange that you are using two backup methods . I would suggest you to use Hot Backups , if you have production database , otherwise cold backups will do the trick for you .
Hare Krishna
Alok

Operating System Consistent Backup (Archive Mode)

Hi
In archive log mode, after putting a tablespace in backup mode one by one as follows:
Alter tablespace tablespace_Name begin backup;
If I create backup using operating system commands one by one for all tablespaces. Will this backup consider as consistent? Will it require archive logs to keep it consistent?
Wishes
Jawad

hi,
Your hot backups consider to be consistent only,when you have all the archivelogs sice you made the hot backup.when you put tablespace in backup mode, then the dbfile header freezes at the point you fire 'alter tablespace <name> begin backup' ;
on the other hand the rest of the files have diff scn's.this is the reason it does not consider to be a consistent backups.that dosent mean you should not take hot backups,insted oracle recommend to operate your db in archivelog mode specially if it is prod db.
hare krishna
Alok

RMAN and consistent backup

Hi,
in10g, I stop my DB and make a cold backup of my database (it's then a consistent backup).
Can RMAN use this backup ? How ?
Thank you.

Yes, you can use a Cold Backup that was not taken by RMAN -- provided that you have the controlfile included in the backup. In RMAN you'd have to CATALOG each of the files in the backup as DATAFILE COPYies before RMAN can be "aware" of them.
If the backup was taken with an OS command, you simple use the revers e to restore it. If your backup did not include the Online Redo Logs, you'd need to use
STARTUP MOUNT
ALTER DATABASE RECOVER
CANCEL
ALTER DATABASE OPEN RESETLOGSin SQLPlus.
If you want to Apply ArchiveLogs generated after the backup, then don't open resetlogs !
continue with
RECOVER DATABASE USING BACKUP CONTROLFILE UNTIL CANCELand CANCEL when you have applied the last available archivelog, after which you will
ALTER DATABASE OPEN RESETLOGSHemant K Chitale
http://hemantoracledba.blogspot.com
Edited by: Hemant K Chitale on Aug 30, 2009 11:31 PM

Restore RMAN consistent backup

Hi Team,
Please let me know how to restore only my consistent backup in RMAN?
Example: Sunday consistent incremental level 0 backup by keeping the database in mount state and Monday-Saturday incremental level 1 backup.
If my database crash on Tuesday, I want to restore only the sunday consistent backup and open the database and I don't want to run the recover database command. (restore database -> alter database open). Please let me know how to do this in RMAN.
Thanks,

Thanks Hemant for the Solution.
Action Taken:_
RMAN> startup nomount
RMAN> RUN
ALLOCATE CHANNEL c1 DEVICE TYPE sbt
restore controlfile;
ALTER DATABASE MOUNT;
restore database;
recover database noredo;
alter database open RESETLOGS;
Reference:
How to Restore a Cold Backup and Open Resetlogs via RMAN Without Applying Redo [ID 257326.1]
Thanks,
Mukarram

Consistent backup RMAN

hi ,
10gR2 on win 2003
i want to take the consistent backup of the database using RMAN using controlfile i fired these queries
datbase is open & database is in noarchivelog mode
my question why i everytime fail to mount the database from RMAN? i tried it with archive log option also but the result is same.
or AM i missing a step if yes suggest me steps required.
C:/rman target /@orcl
rman> shutdown immediate;
using target database control file instead of recovery catalog
database closed
database dismounted
Oracle instance shut down
RMAN> startup force dba;
RMAN-00571: ===========================================================
RMAN-00569: =============== ERROR MESSAGE STACK FOLLOWS ===============
RMAN-00571: ===========================================================
RMAN-03002: failure of startup command at 07/20/2009 11:44:57
ORA-12514: TNS:listener does not currently know of service requested in connect
rman> startup mount
RMAN> startup mount
RMAN-00571: ====================================================
RMAN-00569: =============== ERROR MESSAGE STACK FOLLOWS ========
RMAN-00571: ====================================================
RMAN-03002: failure of startup command at 07/20/2009 11:47:47
RMAN-06171: not connected to target database
Thanks

Hi,
Please follow the procedure...
1. First please check if your listener is listnening for your database service by entering the Listener prompt
Lsnrctl
Then execute the command "Services"
C:\Documents and Settings\admin>lsnrctl
LSNRCTL for 32-bit Windows: Version 10.2.0.1.0 - Production on 20-JUL-2009 13:07:13
Copyright (c) 1991, 2005, Oracle. All rights reserved.
Welcome to LSNRCTL, type "help" for information.
LSNRCTL> services
Connecting to (DESCRIPTION=(ADDRESS=(PROTOCOL=TCP)(HOST=abc)(PORT=1522)))
Services Summary...
Service "PLSExtProc" has 1 instance(s).
Instance "PLSExtProc", status UNKNOWN, has 1 handler(s) for this service...
    Handler(s):
      "DEDICATED" established:0 refused:0
         LOCAL SERVER
Service "orcl" has 1 instance(s).
Instance "orcl", status UNKNOWN, has 1 handler(s) for this service...
    Handler(s):
      "DEDICATED" established:0 refused:0
         LOCAL SERVER
The command completed successfully2. If it is listening to your database service....go the machine from where you want to connect to the database and check your tnsnames.ora file for the entry for your database listener. In your case i suppose its already there as you have already pasted your file and it contains entry for the ORCL.
3. Go to the command prompt and try TNSPING ORCL...See if its working.
4. Go to to the RMAN prompt...
C:\Documents and Settings\admin>rman
Recovery Manager: Release 10.2.0.1.0 - Production on Mon Jul 20 13:10:19 2009
Copyright (c) 1982, 2005, Oracle. All rights reserved.
RMAN> connect target sys/oracle@abc
connected to target database: ABC (DBID=3898979772)
RMAN>Please let us know if you still face any issues connecting to the Database through RMAN
Regards

Consistent backup of online system (NOARCIVE mode)

Hi
Is there anyway I can make consistent backup of online system running in NOARCIVE mode?
Wishes
Jawad

There's only one kind of backup possible, if a database is running in noarchivelog mode, it's a cold backup, i.e. the database is down. In the past some people said an export with option 'consistent=y' is a consistent backup, but that's no longer the state of the art.

OAS 10.1.2.0.2 BI + Forms + Reports + Infrastructure DB - consist backup

Hello!
i Have created two scripts :
Backup_middle.sh
rm -rf /oracle/backup/middle/*
su - oracle -c "export ORACLE_HOME=/oracle/middle && /oracle/middle/backup_restore/bkp_restore.sh -vm configure > /skrypt/backup-middle.log && /oracle/middle/backup_restore/bkp_restore.sh -vm node_backup -o prepare >> /skrypt/backup-middle.log"
export ORACLE_HOME=/oracle/middle
/oracle/middle/backup_restore/bkp_restore.sh -vm node_backup -o image_backup -P /oracle/backup/middle >> /skrypt/backup-middle.log
chown -R oracle:oinstall /oracle/backup/middle >> /skrypt/backup-middle.log
Backup_infra.sh
rm -rf /oracle/backup/infra/*
su - oracle -c "export ORACLE_HOME=/oracle/infra && /oracle/infra/backup_restore/bkp_restore.sh -vm configure > /skrypt/backup-infra.log && /oracle/infra/backup_restore/bkp_restore.sh -vm node_backup -o prepare >> /skrypt/backup-infra.log"
export ORACLE_HOME=/oracle/infra
export ORACLE_SID=orcl
/oracle/infra/backup_restore/bkp_restore.sh -vm node_backup -o image_backup -P /oracle/backup/infra >> /skrypt/backup-infra.log
chown -R oracle:oinstall /oracle/backup/infra >> /skrypt/backup-infra.log
If i run frist script backup_middle.sh and wait for finished then i run sceond script backup_infra.sh wait for finished is my cold backup will be consist?
Regards - Marcin

Hi,
Why you do not use cp -pR command to take an cold back-up of your midtier & infra?.
Regards,
Hamdy

Howto take a CONSISTENT backup of AS-Java WITHOUT shutting down?

According to the official documentation one needs to shutdown the Java instance (EP, KM, BI-Java) and backup filesystem and database at the same time to ensure consistency, so far I understood.
Since there is no easy way of finding out, who's logged in at a specific moment, we have no chance to inform logged in/working users about an imminent downtime/shutdown.
Things become even worse, if backups are taken at night when different timezones are using the system productively.
What is a recommended "best practice" for those kind of setups? Is there anywhere an information on which of the running applications writes data into the filesystem so we could probably keep it in sync with the database "manually" in case we need to recover a backup, that was taken online?
Any input greatly appreciated.
Markus

Hi Markus,
my strategy is to take an online (cons) backup every night and an offline backup every sunday at 00:00.
There is a way to take an online cons backup, and now is supported in 2004s (maybe too in NW04 but NOT in was 6.20).
Here are my scripts for online/offline backup.
Online :
xcopy E:\usr\sap E:\BO\j2e /Q /E /C /H /R /O /Y
xcopy E:\FSDB E:\BO\fsdb\ /Q /E /C /H /R /O /Y
brbackup -p initp1pOnline.sap -d disk -t online_cons -m all -c -u system/password -k no -e 0 -l E -a -p initp1pOnline.sap -save_delete -d disk -n 10000 -c -u system/password -k no -l E
Offline :
stopsap name=p1p nr=00 sapdiahost=sapep1p
stopsap name=p1p nr=01 sapdiahost=sapep1p
net stop SAPP1P_00
net stop SAPP1P_01
xcopy E:\usr\sap E:\bof\j2e /Q /E /C /H /R /O /Y
xcopy E:\FSDB E:\bof\fsdb /Q /E /C /H /R /O /Y
brbackup -c -u system/password
net start SAPP1P_01
net start SAPP1P_00
startsap name=p1p nr=01 sapdiahost=sapep1p
startsap name=p1p nr=00 sapdiahost=sapep1p
I tested both backups and was able to restore from scratch without problem.
Brad

Sybase ASE - consistent Backups from active snapshots coming?

Looks like it.
http://www.sybase.com/detail?id=1099394&EMC=notificationmysybase&attr=1099394
“Adds the PREPARE DATABASE command. This command is very similar to the QUIESCE DATABASE command, with the exception that it does not block write activity during the HOLD phase. The HOLD/RELEASE options can be used to execute a mirror split or snapshot on the database using a certified agent, to create a copy of a database on which transaction logs can be loaded”.
What this is suggesting is you can ‘snapshot’ a running ASE instance – and create a consistent copy from that snaphot.
This is SAP playing catchup with other vendors – and about time. It opens up a number of possibilities that should simplify managing ASE – integrating backup management into the VMware layer.
However – this release is not out until October.

hi Andy
ok i've just learned how to operate the console...
and the noise just happened at 8.52 though there is no record at 8.52 !
17/11/2013 18:51:34.945 coreservicesd[67]: SFLEntryBase::ListHasChanged mach_msg returned 10000004d
17/11/2013 18:51:34.946 coreservicesd[67]: SFLEntryBase::ListHasChanged mach_msg returned 10000004d
17/11/2013 18:51:34.946 coreservicesd[67]: SFLEntryBase::ListHasChanged mach_msg returned 10000004d
17/11/2013 18:54:20.530 Console[1851]: setPresentationOptions called with NSApplicationPresentationFullScreen when there is no visible fullscreen window; this call will be ignored.
17/11/2013 18:55:56.802 xpcd[177]: restored permissions (100600 -> 100700) on /Users/davidjones/Library/Containers/com.apple.Notes/Container.plist
17/11/2013 18:55:58.545 WindowServer[1674]: disable_update_timeout: UI updates were forcibly disabled by application "Notes" for over 1.00 seconds. Server has re-enabled them.
though i did scroll back earlier and though i can't be certain it was happening at the this time i saw this suspicious entry repeated hundreds of times. i'm a complete novice on console - this is just on the start-up screen - not tried any of the side menus.
will try in safe mode next unless you think anything below is the culprit!
17/11/2013 17:07:43.009 coreaudiod[169]: kAudioUnitErr_TooManyFramesToProcess : inFramesToProcess=1115, mMaxFramesPerSlice=512
17/11/2013 17:07:44.007 coreaudiod[169]: kAudioUnitErr_TooManyFramesToProcess : inFramesToProcess=1115, mMaxFramesPerSlice=512
17/11/2013 17:07:45.006 coreaudiod[169]: kAudioUnitErr_TooManyFramesToProcess : inFramesToProcess=1115, mMaxFramesPerSlice=512
17/11/2013 17:07:46.004 coreaudiod[169]: kAudioUnitErr_TooManyFramesToProcess : inFramesToProcess=1115, mMaxFramesPerSlice=512

Consistant MaxDB and J2EE backup on Windows

Hello!
I need a Windows script for a consistent MaxDB+J2EE(and SDM) full offline backup to local disk.
Does anyone have something to share...
I usually do this in via cron in unix, but now I have to set up a similar backup-routine in Windows.
//Fredrik

Hi Fredrik,
I assume you know how to create MaxDB backups (in ONLINE and in ADMIN mode).
So the major question would be to backup the J2EE data stored outside the MaxDB, For that one - I guess - this is the wrong forum.
A backup script for such a consistent backup should anyhow include the steps:
1. Shutdown J2EE,
2. Bring DB to ADMIN mode,
3. perform a Backup of the DB,
4. Backup the J2EE data,
5. restart both DB and J2EE.
Since this obviously involves a downtime I'm wondering if there isn't any mechanism of the J2EE server that would recreate lost or outdated data outside the database after a crash or a recovery.
==> This is really something for the J2EE forum ...!
KR Lars

Test RMAN consistent "as copy" backup outside of ASM (RAC)

Hi!
We have create RMAN consistent backup in our RAC (ASM with FRA) environement.
Backup steps looks like:
alter system set cluster_database=false scope=spfile sid='HAC4';
srvctl stop database -d HACThen we start RMAN backup script like (database is shut downed before):
run {
startup mount;
allocate channel t1 type disk format '/u01/rman_backup/HAC/cold_backup/HAC_%U';
allocate channel t2 type disk format '/u01/rman_backup/HAC/cold_backup/HAC_%U';
SET CONTROLFILE AUTOBACKUP FORMAT FOR DEVICE TYPE DISK TO '/u01/rman_backup/HAC/cold_backup/HAC_%F_ctl';
backup as copy database include current controlfile;
}After backup we perform:
alter system set cluster_database=true scope=spfile sid='HAC4';
shutdown immediate;and later
srvctl start database -d HACBackup files from this backup are placed on local disk area accessed by node4, where HAC4 SID is on.
We transfer those backup files to tape on some schedule basis...
Now we want to test this backup.
Scenario:
Because we have consistent backup we think that we can perform restore from it (from that disk files) and from it ONLY!
Our problem is that all the time restore is using ASM files, so we are not sure is our backup valid.
So, how to prevent RMAN restore/recover process from using any ASM file - just to use our consistent backup data?
Please some script example for that....
Our config:
Oracle 10gR2.0.4 PS3 (CRS, ASM, database)
Linux Rhel 5.3 IA64 on "raw devices" (not raw disks)
THX
Damir

We'll...look this restore for our "as copy" consitent backup. Assume that ${BKUP_DEST} is env variable in bash script which points to same place where backup was done. Here is restore part:
run {
startup force nomount;
RESTORE CONTROLFILE from '${BKUP_DEST}/HAC_20090916T152408_u5kpd8le_1_1_post_ctl';
alter database mount;
sql "alter session set NLS_DATE_LANGUAGE=American";
allocate channel t1 type disk format '${BKUP_DEST}/HAC_%U';
allocate channel t2 type disk format '${BKUP_DEST}/HAC_%U';
restore database FROM TAG='TAG20090916T152532' ;
release channel t1;
release channel t2;
}where
TAG='TAG20090916T152532'is tag that was done in backup. This is done from "ac copy" backup. But when we run:
run {
allocate channel t1 type disk format '${BKUP_DEST}/HAC_%U';
allocate channel t2 type disk format '${BKUP_DEST}/HAC_%U';
recover database FROM TAG='TAG20090916T152532' ;
release channel t1;
release channel t2;
}RMAN is using archive logs from ASM....and we want to use only files from "as copy" backup-remember we test "as copy" consistent backup.
If we do not run "recover" part then we cannot open database!?
(i.e. alter database open resetlogs).
Hope now is all more clear..
Regards,
Damir
P.S.
Either our backup script is not OK or restore/recover script is bad....
Unfortunatelly we do not know what?

Is Retrospect Necessary if have Backup 3?

I am finally getting serious about backing up and forming a consistent backup plan for my computer. I think I have Backup 3 setup right for my home folder to be backed up on iDisk but I haven't tested it to be sure yet.
My next plan is to back up my data to at least one external hard drive and or DVD's at least once a month. I also have a ton of media files that I will need to store seperetatly and backup as my powerbook does not have the room for all that media. Is backup 3 good enough to organize a plan for this or should I take a look at retrospect? thanks for your help.

Hi, 10smom
1. You wrote: "75 gb capacity on mac. 100 gb external hd. Plan on adding another external hardrive as 2nd backup. How large should i get if want to use part for media data storage as well? "That all depends on (a) just how much "media storage" capacity you want, and (b) what types of backups you'll be creating on the second hard drive.
If you read my previously-cited"Backup and Recovery" FAQ, there are some guidelines therein on sizing / setting up a FireWire drive for backup and additional uses.
You could also go on the assumption of 75 GB + whatever media storage space you think you will want. Drives in the 250 GB range are getting pretty cheap these days...I always recommend buying more than you think you need.
2. The other question you have to ask yourself is the following: what is your plan to backup the data on the space used as "media storage"? If all the "media" being stored is also on your Mac, then a full backup of your Mac's (presumably single) hard drive will back up the "media," by which I presume you mean things like photos, movies, and the like.
If you plan on archiving "media" from your Mac to the "media storage" space, i.e. moving — not copying — it from your Mac to the media storage space on the new FireWire drive, then you'll want to backup the media storage space or risk losing "the media" stored therein.
Good luck!
Dr. Smoke
Author: Troubleshooting Mac® OS X
Note: The information provided in the link(s) above is freely available. However, because I own The X Lab™, a commercial Web site to which some of these links point, the Apple Discussions Terms of Use require I include the following disclosure statement with this post:
I may receive some form of compensation, financial or otherwise, from my recommendation or link.

Time Capsule does not keep or recognize older backups

Hello,
I have a 1TB time capsule that I've had for a couple years. It used to back up to my old MacBook (blackbook) until I got a brand new MacBook Pro Retina a few months back. From the moment I tried to do the initial back up I noticed problems with it. It seems to work decently well but recently have had a number of issues that I would like to resolve. I just ran disk utility and it said the drive is working fine.
The machine always seems to have the "!" and the drop down menu says "latest backup delayed" and when I "backup now" it goes through its cycle and all is good. However the machine is about 90% full yet when I "enter time machine" there are only backups for about a week. Right now my oldest back up is March 11th on a drive that have over 910GB of used storage and a 750GB HD on my comp so clearly there should be more backups there.
I don't necessarly care about keeping what I have on the drive because if its important to me then its on my computer. How do I go about reformatting this drive and setting it up in the correct manner so I can go back having consistant backups that are saved for a long period of time just like my old computer was? I just want this to work how Apple products are meant to work, easy and painless!
Thanks,

robcan wrote:
Hi all,
I have a Time Capsule 2TB 4th Gen and have also had similar problems. My Mac ran out of space and I was finally forced to start a new iPhoto library and leave the old iPhoto library only on the TC. The thing is I now cannot access the old library.
How did you put the old iPhoto library on the TC? It sounds like you used Time Machine?
If so, no, you can't access TM backup files with the iPhoto app; you must restore them somewhere for that.
Has my TC just lost me over 100GB of priceless photos?
Maybe, maybe not. But you need to do something different with them. Get an external HD quickly; one plenty large enough to store the originals of all your photos. Actually, get two, one a portable model, so you can copy or back up your photos to it, and take it to a secure off-site location, such as your workplace, relative's house, or safe deposit box, so you're protected against fire, flood, theft, etc. Format them for a Mac (per #1 in Using Disk Utility).
Also when I click on the sparsebundle it wont open so cant access the location where my iPhoto library probably is.
What happens? Do you get a message? If so, what?
You may be able to repair the backups, per #A5 in Time Machine - Troubleshooting.
Once this is sorted out, you'll want to let Time Machine back up both your Mac's internal HD and the external with your photos. To do that, just remove the external from the exclusion list via TM Preferences, per Time Machine - Frequently Asked Question #10.

ENTERPRIZE BACKUP UTILITY

제품 : ORACLE SERVER
작성날짜 : 1996-11-29
1. The Problem
과거 메인 프레임 환경에서 사용하던 고객들이 저 비용, C/S환경으로
Migration시에는 주로 Open system과 Oracle RDBMS를 채택하고 있다.
이런 관계로 해당 제품 시장은 고 성장을 구가하고 있기도 하다.
이러한 고 성장을 가능케 하는 요소는 물론 H/W,S/W에서의 혁신적인
진보이나, 메인 프레임급에 걸 맞는 대용량 DB를 관리하고 Backup하는
데 필수적인 System management tool은 과거와 크게 달라진 바가 없다.
tar, cpio, dd같은 UNIX backup tool들은 실제 backup media의 관리
측면에서 별 기여 를 하지 못하고 있다. 대용량 DB를 위한 Backup script는
source만 해도 수백line에 달하며 test와 troubleshoot에서도 많은 문제점을
갖고 있다. 뿐만 아니라 대용량 Database(이하 VLDB)의 Restore는
Backup보다도 훨씬 더 복잡한 문제를 갖고 있다.
최근의 Backup file이 어느 곳에 저장 되었는지를 알기 위해 Backup
tape을 분류하고, damaged file들을 restore 해야 하는 등의 일련의 작업이
필요하다.
2. The Solution
최근에 몇몇 media management vendor들은 media management기능,
scheduling기능, 강력한 security기능, UNIX system에서 제공하는 backup
tool이상의 신뢰성을 갖는 강력한backup product를 공급하기 시작했다.
이러한 product들은 system level에서의 관리를 용이하게 해주는 장점은
있으나 이미 위에서 지적된 VLDB의 backup시 제기되는 문제점들을
완전히 해결해 주지는 못한다.
Oracle7 Enterprise Backup Utility(이하 EBU)는 이런 문제점을 해결하는
solution이다. EBU는 backup/restore시 media management product와 강력한
인터페이스를 제공하며 다른 어느것과 비교 할 수 없는 강점을 갖고
있다.
3. Introducing Oracle7 Enterprise Backup Utility
EBU는 open system상에서 메인 프레임급 성능을 제공하여 한 기업
전반에 걸쳐 전략적으로 중요한 역할을 감당하게 된다. 이 utility는 DB의
신속한backup과 restore를 위해disk와 backup될 media device에서H/W
parallelism을 이용한다.open system의 full capacity를 이용하기 위해
device가 add될 때 performance가 증가된다.parallel H/W, backup
configuration check ,error detection and cleaning ,restore시의
database의block verification과 같은 특징들이EBU가 high reliability를 갖게
하는 특징들이다.
Oracle7 데이타베이스는 data files, control files, redo log files로
구성되는데 EBU는data file, control file을 backup받고 archived redo log
file을 받는다. (release2.0.8) Media failure의 경우 DB를 full restore/
partial restore가 가능하며 data file도 선택적으로 최근의 last backup까지,
어느 일정 시점 까지 만의 선택이 가능하다. 또한 recovery도 일정 시점으로
완전recovery가 가능하다.
4. Architecture
EBU는 크게 두 부분으로 구성되는데 backup하려 하는 DB에 관한 모든
current정보와history를 관리하는 catalog부분, 그리고 실행 파일 부분으로
구성된다. 실행 파일들과 catalog는 반드시 같은 host내에 존재할 필요는
없다. 실행 파일들은 다음으로 구성된다.
*obackup : 모든 다른 프로세스들을 monitor하며 backup하려하는 DB,
catalog와의 communication을 담당한다.또한 Instance Manager를
running시킨다.
*brio : obackup, brdk, brtp사이 조정 역할을 수행한다.parallel I/O
stream당 하나의 brio process가 발생한다.
*brdk : Disk file의 read/write를 조절한다. File당 하나의 brdk를 발생시키
므로 만일 여러 개의 file들을 하나의 single data stream으로 결합시
키는 경우 하나의 I/O stream내에 여러 개의 brdk process를
create한다.
*brtp : Tape file들의 read/write를 조절한다. I/O stream당 하나의 brtp
create.
*brd : Instance Manager로써 backup catalog, obackup을 monitor하는
daemon process이다. backup을 위한 cleanup과 비정상 종료된 작업을
복구 한다.
Backup동안 database file들은 disk process인 brdk에 의해 읽혀져 backup
buffer area로 넘겨지고 그 후 third vendor의 media management s/w를
경유하여 backup device에 brtp process에 의해 write된다. Restore시는
반대의 과정이 진행된다. 위의backup buffer는 hard disk와 tape device간에
발생하는 I/O상의 speed mismatch를 조절하는 역할을 한다.
5. EBU의 장점(Benefits)
Third vendor의 media management product와 함께 사용되는 EBU는
기존의 UNIX 에서 제공하는 backup tool보다 다음과 같은 강력한 장점을
지닌다.
(1) 강력한 Reliability.
*. Consistent backup procedures
기존의 hot backup script의 경우 만일 datafile이 추가될 경우 script에
새로 지정해 주어야 하며 VLDB의 경우 script만 해도 수백line에 달하여
관리상 번거로운 게 사실 이였다. 그러나 EBU의 경우는 어느 datafile이
어느directory에 위치하는지를 작업자가 알 필요가 없고 script는 짧고
단순하여 누구든 쉽게 알아볼 수 있다.
또한 DB 구성이 자동으로 확인되며 backup media를 자동으로 인식 및
관리한다.
예).full online backup script .partial online backup script
backup online database backup online
db_name = "PRODB" db_name = "PRODB"
oracle_sid = "PROD" oracle_sid = "PROD"
tablespace = "SALES"
.full restore script
restore database
db_name = "PRODB"
oracle_sid = "PROD"
이와 같이 간단한 script에 의해 DBA는 VLDB를 쉽게 backup과 restore를
할 수 있으며 작업에 필요한 세부정보는 backup catalog내에 있게 된다.
*. End-to-End checksum
Backup시 EBU는 각 backup data의 byte를 계산하고, restore시도 계산하여
정상 수행 여부를 판단 시 checksum 값을 상호 체크 한다.
*. Instance Manager
Instance Manager는 backup/restore시에 진행 상황을 monitor하고 작업
중에 에러가 발생시 복구 및 각종resource를 release시키는 기능을 담당하게
한다.
(2) High Performance
다음과 같은 EBU의 특징과 고속media device의 지원 아래 VLDB의
backup을 적은시간 내에 완전하게 수행 할 수 있게 되었다.
*. Parallel Hardware Capability
Backup/restore시 multiple device를 사용하여 동시에 여러 tablespace에
대한 작업을 수행 할 수 있다. VLDB와 같이 다수의 tablespace, datafile이
여러 disk에 분산되어 있는 경우 이와 같이 물리적으로 다수의 device를
parallel하게 사용함으로서 performance를 극대화 시킬 수 있다.
Backup시 동시 사용되는 device의 수와 performance는 정비례하게 나타난다.
*. Multiplexing
EBU는 backup device가 최고의 속도로 backup을 수행 할 수 있게 script에
지정 할 수가 있다.이것은 backup media access속도보다 disk access
속도가 훨씬 작은 경우에 여러 disk에 분산된 datafile들을 한device에
multiplexing되게 지정함으로 가능하다.
*. Null Block Compression
EBU는 backup시 null data block을 skip시키고 restore시 재구성함으로써
공간 을 절약하며 performance를 증가시킨다.
*. Buffered I/O
EBU는 disk I/O와 device I/O사이의 access speed의 mismatch를 줄이기
위하여 모든 I/O stream을 bufferring한다. 양쪽의 I/O는 shared memory
buffer를 거치게 함으로써 backup/restore수행 중 항시 일정 속도를 유지
하게 한다. shared memory는 user가 parameter로 구성할 수 있으며 기본
적으로는 parallel data stream의 개수와 buffer size에 dependent하다.
(3) Availability
한 기업에서 전략적으로 중요한 DB들은 대부분 24시간*7일 full로
가동되며, 사용 중에 backup을 수행할 수 밖에 없다. Offline full backup은
당연히 지원되며, EBU는 DB를 online상태로 full backup이 가능하며,
또한partial backup도 가능하다. 기존의 online backup이 backup으로 인해
performance를 저하시키는 단점이 있었으나 EBU는DB 운영에 최소의
영향만을 미친다.
Recovery시에도 full로 restore 시킬 수 있으며 나머지는 사용 중인
상태에서 필요 부분만 partial restore가 가능하다. 일단 EBU가 data를
restore시킨 후에는 기존의 Oracle7의 표준 recovery process가 적용되어
진다.
(4) Usability
*. Auto-configuration
Online backup을 수행 시 EBU는 current DB configuration과 backup catalog
내의 가장 최근의 configuration정보를 비교하여 backup catalog정보가 더
오래 전의 것이면 EBU는 DB의 backup직전으로 update시킨다.
Auto-configuration은 default이며 명시적으로 지정하여 disable시킬 수 있다.
Offline backup시는 backup script내에 backup command기술 전에 register
라는 명령어를 기술 해야만 configuration정보를 update시킬 수 있다.
*. Light-Out backups
Third party의 media management s/w를 이용하여 EBU를 operator의 개입 없
이 자동으로 수행이 가능하다. UNIX의 cron에서도 자동으로 수행되게 할 수
있다.
*. Raw device support
EBU는 raw device로 잡힌 DB도 operator의 특별한 조치 없이 일반 file
system의 backup과 똑같이 수행할 수 있게 한다.
*. Aggregated restore
기업의 mission-critical system에서의 backup 전략들은 흔히 partial
backup에 의존하게 된다. 이와 같은 환경에서는 backup data내의 모든file에
대한 current version을 검색하는 일이 중요한데 EBU는 이것을 자동으로
수행한다. partial backup data는 restore시 media failure이전에 취해진
가장 최근의 full backup data와 aggregated 되어져 이용된다.
*. Point-in-Time restore
EBU는 backup catalog내에 backup하려하는 DB에 대한 모든 configuration
history를 갖고 있어서, 만일 하나의 tablespace가 drop된 후 다시 필요하게
되면 비록 그 후 일정 시간이 경과되고 많은 변경이 일어났더라도 drop직전으로
일부분만 restore하여 사용이 가능하게 할 수 있다. 이 같은 tablespace는
다른 machine에 restore 될 수도 있어 현재의 DB작동에 아무 영향을 미치지
않고 수행할 수가 있다.
*. Backup catalog
EBU에 의해 제공되는 backup catalog는 모든 backup대상 DB의 configuration
과 history의 저장소로써 제공된다. backup catalog에 저장된 정보는 backup된
file, tablespace 그리고 multiplexing정보, 마지막backup이나 restore의
시간기록,backup type(full or subset),backup들이 저장 되어 있는 file set
등을 포함하고 있다. 하나의 backup catalog는 기업 내 모든 분산된 dB들을 관리
할 수 있다.
*. Dry Runs
VLDB를 backup시 겪는 문제 중 하나는 과연 script가 정상적으로 수행이 완료
되는지를 검증하는 것이다. 특별한 방법이 없는 한 긴 시간 동안 수행시켜 끝난
후 확인 할 수 밖에 없는데 EBU는 I/O없이 간단히 test로 수행을 시킬 수가
있다. 이로 인해 신뢰성 있는 backup전략을 DBA가 수립 가능하게 한다.
*. Enterprise-wide backup
이것은 backup이 단순히 single database에 국한되지 않고 한 기업 전체에
걸쳐서 global하게 수행 가능함을 말한다. 예로 전국적으로 다수의 site에
database가 구축 되어진 상황에서 각database들이 two phase commit,
snapshot 같은 분산처리 작업을 통해 다른 database들을 synchronous하게
또는 asynchronous하게 update할 경우 기업 전체를 하나의 logical
database로 보고 이런 logical database를 중앙에서 EBU를 통해 backup시킬
수가 있다.EBU는 중앙의 catalog내에 모든database들에 대한 backup정보를
간직하고 있다.
이와 같은 중앙의 한 지점을 통하여 기업 전체 모든 database들에 대한
backup수행 기능은 EBU의 가장 큰 장점 중 하나이며 이런 경우 N/W상에 대량의
data가 전송되어 network의 bandwidth가 backup의 중요한 요소로 부각된다.
6. Flexibility
*. EBU API
EBU는 앞에서도 소개되었듯이 단독으로 사용되는 것이 아니고 third-part
에서 제공하는 media management product와 같이 사용된다. 이때 두 product
사이의 interface는 backup/restore Application Programming Interface
(API Glue)에 의해서 이루어 진다. 이것은 third part에 의해 제공되며
EBU install시 installer에 의해 자동적으로 access된다.
*. Media management products
Epoch*s EpochBackup, IBM Adstar*s Distributed Storage Manager,
Legato Systems*NetWorker, HP*s Omniback*, SprectraLogic*s
Alexandria,StorageTek*s REEL, etc.
*. Media device
4mm and 8mm DATS,3480s,3490s,WORM,Writable Optical Devices and
automated systems such as stackers and silos.
7. Pricing
EBU는 무상으로 공급되며 사용을 위해서는 third vendor의 media
management product를 구입하여 설치하여야 한다.

Sure, would you clarify which DB, which OS and which product?
Concept in general is quite simple:
- stop BOE (SIA on XI3, All services, finishing with CMS on XIr2, ideally putting a sleep between all servers shutdown and CMS shutdown). NET STOP on windows.
- Backup the CMS DB (imp/exp on Oracle, mysqldump for MySQL)
- Backup the FileStore (use robocopy in a batch for delta backup)
- Restart BOE
The "catches" are around Oracle doing export in US7ASCII by default under Unix, which is a destructive conversion from UTF8 (your CMS DB should be under this format as per supported platforms), so you need to make sure to set the NLS_LANG to UTF8 on Unix user doing the backup.
Other than that, batch as shell can do all of that.
Regarding the "general" best practices, I recommend the following doc:
http://www.sdn.sap.com/irj/scn/index?rid=/library/uuid/c0020482-ca8d-2c10-9bad-d1bd332bbb28
Note on integrated BO tools: Import Wizard as biarengine are good tools for promoting a set of data between wto BOE environment or if you need to backup a specific set, but they are definitly not recommended as backup solution. I would refrain from this solution IF the point of the backup is to be able to recover the system when it crashes.
Does that answer the question?

Consistent Backup

Similar Messages

Maybe you are looking for