I'm asking these questions together so they better describe what I am trying to understand overall.
We have two next major goals for our application and data center environment: 1) implement hot/warm (Active/Passive) failover for DR and 2) implement hot/hot (Active/Active) for HA after my service application is enhanced to sync over the network in order to scale out. I'm mostly interested in scenario #1 now where the failover data center is keeping a DR replica of our primary database that we want to use R/O for reporting, invoicing etc. in the meantime when not failed over to. This is a great AlwaysOn feature.
We are Microsoft centric (Server 2008 R2 and SQL 2012 Enterprise) so I hope to use SQL AlwaysOn but am confused on how a few networking mechanisms work in practice and want to better understand the SQL AlwaysOn big picture.
I'm a full time developer and also do most of the IT and understand a fair bit of networking (for a developer) but don't configure much since we are running at an ISP where our IAAS is hosted under VMware and they mostly manage the network. Our long term goal is a multi data center, multi vendor Active/Active geo diverse configuration. We will eventually be converting to Windows Server 2012 or can sooner if required or there is some useful benefit in conjunction with AlwaysOn.
Here are my questions which mostly are asking about how certain IP and Windows Failover Cluster mechanisms work in practice with AlwaysOn.
Do I use DNS to map a VNN to some IP(s) or VIP that all the machines in the Availability Group listen to? How are DNS, VNN, IP(s), VIP, etc. used to make this happen?
After a failover, how does a client find the new machine? Is the connection broken and the client application must have retry logic, or be restarted in some poor design cases, to reconnect? I think a VNN is used here and if so would like to understand how that works before, during and after the primary machine fails.
Can we setup a Windows Server Failover Cluster on Windows Workgroups or must we convert all the involved OS and SQL machines to a domain based system under Active Directory? Does the DC have to be its own machine? We are a very small operation with a high real time transaction rate so setting up AD would be a a big hit on my time as I've worked in that environs before and thus far we have run with workgroups just fine.
How is the R/O replica database accessed before and after failover? Can it be accessed after failover and did it just become no longer R/O to applications that expected it to be? I could see someone coding a R/W test to see if they had the correct database! Iow, can/how do I specifically access/address the replica database machine for guaranteed R/O reports rather than how a R/W client might typically connect to the database not caring if they got the primary or the failover database as long as it was R/W and live?
Thanks so much, Dave
Dave