Re: Oracle Backup and Recovery Handbook

Peggy,

It is possible depending upon the size of the redologs that one checkpoint may not completely be stored in one redo log, i.e suppose the sga is 200Mg, out of which the data area is 100M, and say 80% of this data needs to be written out, then one logswitch (either manualy or automaticlly) may not all be stored within one log. If you try to do something nasty to the instance ( I assume so, I have not read the book as yet) then, in case recovery is needed, having multible writes to the logs will ensure that all the required info is in the logs, from which recovery can be made.

I am assuming here that the book forces a crash, and then shows how to recover from a specific event. Although in real life you will have a crash without time to prepare for it.

Hope this helps.

--
Vikram Goel                                
Sr. Oracle DBA - Consultant
Aerotek Inc.                                My email:  vgoel_at_emi.net

Motorola Info:                              Motorola email: vgoel_at_pts.mot.com
Mail Stop 39, Room S1014
1500 Gateway Blvd,
Boynton Beach, FL 33426 

In article <4q9q29$htq_at_news.inforamp.net>, lokp_at_tdbank.ca (Peggy Lok) writes:


>

>I was going through those case studies listed in the last chapter of

>the 'Oracle Backup and Recovery Handbook' and noticed that the author

>likes to do several log switches in a row before simulate each failure

>case.   Does any one know why more than one log switch is required??

>

>Thanks...


>

>Peggy Lok

>

>