Since engineering needs often lead customers 'live data to the company's development environment, as customers' live data volume is increasing, the original
The exported introduction method takes more and more time, so testing the efficiency of RMAN, verifying if it can be recovered RMAN
The method of the part replaces the previous EXP / IMP method. I originally wanted to use the transmission table space, but because we usually create the database
When you create a large data file to the table space used in your business, this will cost it on the step of the COPY data file of the table space.
For a long time, so I decided to test RMAN.
Test purposes: Implement data transfer for JHJKREC users.
Test on 10.101.48.71 machine. Test environment: P42.8G single CPU, 1G memory, 80G single hard drive, Windows 2000 Server, Oracle817
The total size of data files related to JHJK users is: 21.4g where JhjkRec users have a total size of: 2.1g
1. Rman makes a Full backup, used for 20 minutes, generate file size 5.47GEXP as a jhjkrec user data export, 2 minutes 20 seconds, generate file size 546m
2. Rman is recovered, 23 minutes (including RESTORE data file, Recover database) IMP is imported for JHJKREC user data, 55 minutes
3. RMAN Backup In Non-Archive Mode, you need to close the database and start the database to the mount status EXP does not need to turn off the database in non-archive mode, but the data table itself is inconsistent.
Conclusion: Time to use the RMAN backup is significantly longer than EXP, the backup file is also much larger than the EXP file but the time required to recover less than IMP.
With RMAN recovery, you also need to back up control files, and because it is a non-archive mode, you need to check the maximum SCN that is currently recovered, then in SQLUS, Recover Database Until Change SCN, then Alter Database Open ResetLogs, the entire operation ratio exp / IMP Cumbersome, may be more difficult for on-site business people, so use RMAN or use export import import as data transfer, or see you.
The following is a command to be used in the test, which is basically not optimized.
RMAN backup script: @@ connect.rcv # Backup up database at full level # shutdown immediate; startup mount; run {allocate channel d1 type disk; set limit channel d1 kbytes 2097150; backup full database format 'E: / rman_bak / dbfull_ % D_% S_% p_% T 'tag =' dbfull '; Release Channel D1;} ALTER DATABASE OPEN; EXIT;
RMAN recovery script: @@ connect.rcvshutdown immediate; startup mount; run {sql "alter session set nls_date_language = AMERICAN"; sql "alter session set nls_date_format = '' yyyy-mm-dd hh24: mi: ss ''"; SET Unsteil Time '2004-9-7 15:30:00'; allocate channel ch1 type disk; restore database; release channel ch1;} exit; eXP scripts: userid = jhjkrec / jhjkrecfile = E: /RMan_bak/jhjkrec.dmpowner = jhjkRecdirect = ncompress = nfeedback = 1000
IMP script: userid = jhjkrec / jhjkrecfile = E: /RMAN_BAK/JHJKREC.DMPIGNORE =yfromuser=jhjkRectouser=jhjkRecfeedback=1000