harshljica

 Profile Answers by harshljica 

• Aug 19th, 2009

 

During execution of begin backup mode all the datafiles header are stuck up.
During Begin backup mode the scn no, ckpt no and log sequence no are not write in header of datafile so that more redo entries are generated.

oracleofdelphi

 Profile Answers by oracleofdelphi 

• Sep 9th, 2009

 

During begin backup mode datafile headers get freezed and as a result row information cannot be retrieved as a result the entire block is copied to redo logs as a result more redo generated and more log switch and in turn more archive logs.

atifkhan8686

 Profile Answers by atifkhan8686 

• Sep 13th, 2009

 

As soon as you issue 

SQL> Alter database begin backup;

This command freezes the datafile headers as well as controlfile headers, but the headers of redolog files are free.

Hence if any of the transaction is performed after the online backup command. The SCN's are recorded in the header of redolog files but in the header of datafile and controlfiles header hence extra redo's are generated.

murali_sklr

 Profile Answers by murali_sklr 

• Dec 17th, 2009

 

During begin backup mode particular datafile headers get freezed  after this any modification will be happened in particular datafile entire block is copied to redo logs as a result more redo generated and more log switchs will be happend so more archive generated.

dhandapanik

 Profile Answers by dhandapanik 

• Dec 30th, 2009

 

During begin backup mode, the data file headers get freezed and it doesn’t get updated until the end backup. So it causes a checkpoint to the datafile.  This causes all dirty blocks to be written to the datafiles and the datafile header is stamped with SCN consistent with those data blocks.  All other changes are occurring in the datafile from DML such as insert,  delete and update statements.  Due to which, transactions happened in particular datafile of the entire blocks to be copied to redologs as a result more log switches will be happened hence more redos will be generated.

Regards,
Dhandapani.k

ddj1966

 Profile Answers by ddj1966 

• May 28th, 2010

 

When a user managed Hot backup is executed, the tablespace or database has to be placed in backup mode. This informs the Oracle kernel that SCNs should not be updated on the datafile. i.e. the header is frozen until the end backup statement is executed. During this period the datafiles will be updated but not the SCNs in the header. To aid recoverabilty the entire Oracle block is written to the redo logs, consequenlty the redo logs increase in size. This is done to prevent split-block corrupting the backup. o/s blocks for e.g are lets say 512k, lets say Oracle block defined as 8k, and therefore a multiple. So if you were in the process of backing up a datafile at O/S and lets say 3 of the O/S blocks had been copied you still technically have more O/S blocks to copy and in the meantime someone could have changed a row that resides on the O/S blocks that have been copied or yet to be copied. By having the entire blocks in redo, if there issues and the backup is required you have all the information to recover any changes.