Re: AIX Mirror Write Consistency

JL >
>> At a later date, when the database has been running
>> for some time, the block is read from the other side
>> of the mirror ?
>>
>

DC >
>Is it really that random? There is the concept of a primary isn't there?

I hope not (at least for reading) - if so then the benefit of "mirrored disks
give you twice the read rate and half the contention" goes out of the window.

>Also,I would think that the rollback would have re-written this
information,
>so that the
>mirrors would be in sync again anyhow. I think it all depends on wether or
>not Oracle
>waits for writes to go thru to the mirror or not before considering it a
>committed transaction.

Don't forget that data (and rollback) blocks relating to committed transactions
may be written to disc long after the redo log that is the transaction goes to
disc. Oracle will not accept the transaction as committed until the redo write is complete - and the redo write won't be complete until the O/S knows that both redo block copies are on disc. But if DBWR is writing when the machine crashes and the mirrors have not been written then you have an opportunity for the recovery process to read the clean half of a pair and do nothing with it leaving the dirty half of the pair to cause a subsequent problem. Received on Fri Dec 24 1999 - 03:14:01 CST