Hi,
Trying to get the whole picture with DAG, native backup and disaster recovery.
Lets say we have two HUB/CAS servers CAS01 and CAS02.
Then we have 3 MBX servers. MBX01 and MBX02 have active copies splitted among them and MBX03 is having a passive lagged copy of all the databases.
MBX01 are in Datacenter 1, MBX02 are in datacenter 2 and MBX03 are in a third datacenter only for disaster recovery purposes.
So, if datacenter 1 would go down all active databases would fail over to MBX02 in datacenter 2.
And if both datacenters would go down the databases would mount on the "lagged one" (MBX03).
Since the databases are not up to date (lets say with 14 days of lagged logfiles) what would be the best/recommended way here?
1. I read/heard that if using a lagged copy, that one should only be for restore and not let users on to it? (this means that we would not let users onto a database that is not up to date? or shall one start committing the logs?)
2. Restore DC 1 and 2 and get the "primary" MBX servers up and reseed from the lagged one? (if reseeding I guess one must fist commit the logfiles into the database before seeding the "most-up-to-date" data back?
3. What would be the best way for disaster recovery if DC 1 and 2 is lost...
4. If DAC mode is enabled I wont ahve to wory about two databases will mount at the same time since the DACP bit is set to 0? (have been trying to read technet! :) )
5. Since I have en odd number of MBX servers I wont need a witness right? But guess one will be created by default? So this is for issues with split brain. Is it still recommended to have an alternate witness even with an odd number of MBX servers?