What happens during eva failover ?

[Bhushan Bumber]

Dear All,

I am asking this question as i didnot found the answer in any of search. My issue is that we have two sql servers and the luns are replicating on two EVA4400.At one time the SQL server is up & running at primary side.In case of any issue with the primary side , we can start the sql server at the DR side with the replicated luns in synch mode.My questions are

1)I am curious to know that if we have a unplanned outage , then we will have to trigger failover. Will I be able to mount database at DR side ? what will be the guarantee of the database mount ?

2) During failover what are the other process happens apart from change of source to destination and destination to source ? Is there any consistency checks done on san level ?

3) I read in CA admin book that in case of unplanned outage , i have to suspend the replication before failover? Is that mandatory ?

4) If the database gets mounted at DR after unplanned outage then why do i need to shutdown the server after sql database shutdown during failback if there replication is in synch mode ?

5) Are there are microsoft/sql patches which automatically clear the cache after database shutdown ?
I will appreciate the answers and points will be given to the convincing answer.

Regards

Bhushan

[Armin Kerl]

So, nobody answers to you, I will try:

1) If you have Syncronous Mode the Data flow is: Server > EVA1 > EVA2 > Respond to Server that Data is writen
So everytime the Server writes the Data to the EVA and he get a O.K. back, the Data is save on both EVA's.
I will say, there is a Guarantee on the EVA Side, but not on the Server Side, because Data in the Server Memory will be lost.

2) No, the Data on the EVA itself is consistent, there is no Check on the SAN.
But unwriten Data may be lost, this is a Server/OS/DB Problem, not SAN.

3) Not sure, but I don't do it and it works.

4) If you change your Source between EVA1 and EVA2 the LUN-Path changes, some OS can not handle this. You did not write what OS you have. Windows can not handle and need a Reboot (unsupported most Times a Rescan Disk works). If you run your SQL on Windows that is running on Top of ESX, you don't need a Reboot, because ESX can handle this.

5) Don't know. But SQL did very good "Self Recovery" on an unplanned Shutdown and I don't renember that any of my Customers needed a manual repair.