This content has been marked as final. Show 9 replies
That's what a failover is for.
Happy to share my test notes : ( Test was done in Max Availability - but does not show )
In addition you should make sure your Standby redo is setup, your Flashback and you could also consider Real time apply.
There are several Primary failures that can occur, in most cases you will not lose data if you setup correctly. If you cannot have data loss then you need to change your protection mode or consider a RAC solution.
Edited by: mseberg on May 16, 2012 11:17 AM
Ok, thank you! But what happens if your primary goes down - e.g. server crash. What happens to the logs that do not get shipped to standby? Do you lose that information?
We have experienced many "network glitches" whereby primary would failover to standby. So we have disabled failover for that reason.
In maximum availability architecture, Transactions do not commit until all redo data needed to recover those transactions has been written to the online redo logs & standby too..
So In this architecture of maximum performance, if the redo stream not in written in logs of primary & standby , it changes mode in maximum performance. So still there is no lost.
As an example if you take maximum performance real time, if primary server crashes, you will loose only uncommited transactions where log writer not occurred, if commit happens it immediately written into log files. So very minimal data lost.
If you have severe problem with network why can't you go for maximum performance? I can say this is suitable for such environments.
OK, I'll try again. I assume you mean "Auto failover" is disabled. If so you can still do the failover as I posted in my first reply. Can you do a switchover if the Primary goes away, total loss or otherwise. No.
If Failover is disabled and primary database goes down, I cannot run switchover command from standby, correct? Can I run failover from standby then? Thank you!
So your option would be a Manual Failover to the Standby. You seemed concerned about Data loss. Excellent, you should be. Make sure you have the correct number and size of Standby Redo logs on the Standby. Consider using "USING CURRENT LOGFILE" option when applying. The Standby Redo will buffer and the chances of data loss go down. Correct any network issues you have. If you need more please review these white papers :
Please consider closing some of your old questions too 40 (30 unresolved)
Yes, I meant auto for failover. Ok, thank you, it makes sense. If I am in Max Availability (aka commit doesn't happen until transactions are written to logs and are on standby), then number and size of logs on standby would bnot be an issue since they all should be there? So the only transactions we would be losing are those that were not commited, no?
I was not aware that I had to close questions, thank you, I will close them all now :)