-
I agree with you, Modic.
But most of the production datawarehouse sysems do not operate in ARCHIVELOG mode. The reason is data can be rebuilt from the source system at any time.
I come across some big system where backup is also not done for many weeks. When I asked the DBA what he would do when block corruption occurs, he simply said that he could extract maximum number of rows from the table and reorganize it. The reason is over a period of time it is very difficult to maintain backup copy of the tapes for a large system. In fact at one site the total cost of tapes the organizaion maintains exceeds the total cost of the entire system.
-
Hi Friends,
In my production database we are running in noarchive mode.
But this has very valid reasons.
First, everyday's insertions are around 42 gigs. We have no deletes or updates but the data is deleted after 45 days. And whenever there is a problem and the inserts are not done, no sweat. We have the data being written in to a flat file, using which and the sql loader we do the inserts again.Noarchive log helps in speedier system.
Second, who will agree to bear the extra cost involved in maintaining backups when the data is really huge (the total size of db exceeds 3 tb)and there are options like I mentioned above in reloading the data?
Thanks
manjunath
-
jrpm wrote:
1. Use mirrored drives with control software that will let you split the mirrors 'interactively' ....
Which reminds me of the time a certain large aerospace company I worked for used this procedure to do their backups, but one night when they re-synched they got it backwards and overwrote the "current" half of the mirror with the "old" half. That was an exciting morning ;-)
Tim
