Oracle FAQ | Your Portal to the Oracle Knowledge Grid |
Home -> Community -> Usenet -> c.d.o.server -> Re: Are 9i incrementals realy incremental....
rich wrote:
> According to most docs: RMAN incremental backups only backup the data
> blocks that have changed since the last incremental.
>
> I don't believe it.
Why?
Or is this a game anyone can play? "I don't believe that Oranges are orange. They look blue to me. Please discuss."
> I do believe it probably backs up the data files
> that have changed since the last incremental,but NOT block changes.
Since when was Oracle a matter of "belief"?
Test it. It's not difficult.
> Anyone want to challenge this?
>
> Please send email to rheadrick_at_monster(REMOVE_ME).com
You don't get private email. If it's worth discussing (and this isn't) it's worth discussing here.
FYI: RMAN> list backup;
List of Backup Sets
BS Key Type LV Size Device Type Elapsed Time Completion Time ------- ---- -- ---------- ----------- ------------ ---------------
1 Incr 0 191M DISK 00:00:38 18/NOV/04 BP Key: 1 Status: AVAILABLE Tag: TAG20041118T210637 Piece Name: C:\ORACLE\ORA92\DATABASE\01G5E8QT_1_1SPFILE Included: Modification time: 18/NOV/04 List of Datafiles in backup set 1
BS Key Type LV Size Device Type Elapsed Time Completion Time ------- ---- -- ---------- ----------- ------------ ---------------
2 Incr 1 1M DISK 00:00:23 18/NOV/04 BP Key: 2 Status: AVAILABLE Tag: TAG20041118T210852 Piece Name: C:\ORACLE\ORA92\DATABASE\02G5E8V5_1_1SPFILE Included: Modification time: 18/NOV/04 List of Datafiles in backup set 2
You might notice that backup set 2 is just 1MB in size, compared to backup set 1, which is 191MB in size. The only command issued between the two backups was update emp set sal=900 where ename='MILLER'.
Now that should have modified just one block, and that's only 8K, so 1MB is an awful lot bigger than that. But then what do you expect when the database starts checkpointing, and the data file headers get updated, and EMP in any case has an index on it, and the update to the table generated some undo, and so did the update to the index... and so on.
And, by the way:
RMAN> report schema;
Report of database schema
File K-bytes Tablespace RB segs Datafile Name
---- ---------- -------------------- ------- ------------------- 1 256000 SYSTEM *** C:\ORACLE\ORADATA\TEST\SYSTEM01.DBF 2 204800 UNDOTBS1 *** C:\ORACLE\ORADATA\TEST\UNDOTBS01.DBF 3 25600 INDX *** C:\ORACLE\ORADATA\TEST\INDX01.DBF 4 10240 TOOLS *** C:\ORACLE\ORADATA\TEST\TOOLS01.DBF 5 25600 USERS ***C:\ORACLE\ORADATA\TEST\USERS01.DBF ...which indicates that file 5 (the one where EMP resides) is 25MB in size. If this is on a file-by-file basis, how can the backup be smaller than the file involved?? And just in case you don't trust RMAN's own reporting:
C:\oracle\ora92\database>dir
Volume in drive C is WINXP
Volume Serial Number is 84B2-63C0
Directory of C:\oracle\ora92\database
16/10/2004 11:15 PM <DIR> . 16/10/2004 11:15 PM <DIR> .. 09/09/1998 06:31 PM 31,744 oradba.exe 16/10/2004 11:15 PM <DIR> archive 18/11/2004 08:50 PM 762 OraDim.Log 18/11/2004 07:36 AM 11,312 sqlnet.log 18/11/2004 09:03 PM 1,536 PWDtest.ora 18/11/2004 09:06 PM 2,560 SPFILETEST.ORA 18/11/2004 09:08 PM 1,613,824 SNCFTEST.ORA 18/11/2004 09:07 PM 201,195,520 01G5E8QT_1_1 18/11/2004 09:09 PM 1,933,312 02G5E8V5_1_1
....where the O/S itself reports an approximate 100-times reduction in the backup size.
So would you now care to retract the suggestion that RMAN incrementals only increment on a file-by-file basis, and not a block by block one??
HJR Received on Thu Nov 18 2004 - 04:17:37 CST