November 14, 2013 at 12:50 pm
We are running a six instance, Active\Active 2008R2 cluster. We want to go to a 2012 failover cluster. I am familiar with the SQL Server 2008 R2 cluster and how it works, but not the 2012 with the Availablity Group feature. Do you create the Availabilty Group(s) in a Failover Cluster instance and then have the option to fail the entire instance or just move the Availabilty Group to the other node? From what I read... the Availabilty Group is more like a hot standby... allowing you to move the databases in the Availability Group to another node(s). I would think the Availabilty Group would reside separate from the Failover instances.... but would like to hear it from someone that has already set this up.
Thanks!
Charlie
November 14, 2013 at 1:05 pm
SQL Server Failover Cluster Instances are not mandatory for AGs. All you need is a Windows Cluster and standalone SQL Instances to make it work(All the Servers(nodes) participating in AlwaysON AG should be part of same Windows cluster). You can build AGs on top of SQL clustered Instances as well(same as we used to do with DB Mirroring), it really depends on your business RTOs/RPOs. Consider AG as "Super Mirroring". They both work pretty much on same principals, but AG has some Super powers like "Multiple databases Failover as a single unit", "readable Secondaries" etc...
Bottom Line: SQL Server failover Clustered Instances are not required for setting up AGs.
November 14, 2013 at 1:08 pm
Great! Thank you very much for the information.
Charlie
November 14, 2013 at 9:14 pm
Please don't just read a few blog posts, run a few forum questions and then set off doing AGs!!! NOTHING good will come of that! This is complex stuff with lots of caveats, provisos and gotchas!
Best,
Kevin G. Boles
SQL Server Consultant
SQL MVP 2007-2012
TheSQLGuru on googles mail service
November 15, 2013 at 11:22 am
TheSQLGuru (11/14/2013)
Please don't just read a few blog posts, run a few forum questions and then set off doing AGs!!! NOTHING good will come of that! This is complex stuff with lots of caveats, provisos and gotchas!
I get the feeling that its a bit of a rushed out product, I have been evaluating AG's for the last 2 months and there are a number of issues I have found so far. Specifically replication. Your distribution database cannot be part of an availability group meaning really you need to host it in a seperate SQL instance and possibly used clustering to provide any sort of HA. Although there is the option to utilities 2 or more of your servers used as part of the AG solution to host an additional clustered instance with shared storage.
Subscribers are not particularly well supported either and a real pain to deal with if part of a AG.
MCITP SQL 2005, MCSA SQL 2012
November 15, 2013 at 12:53 pm
If you do join a clustered instance to an AO group you will not be able to use automatic failover.
-----------------------------------------------------------------------------------------------------------
"Ya can't make an omelette without breaking just a few eggs" 😉
November 16, 2013 at 8:16 am
RTaylor2208 (11/15/2013)
TheSQLGuru (11/14/2013)
Please don't just read a few blog posts, run a few forum questions and then set off doing AGs!!! NOTHING good will come of that! This is complex stuff with lots of caveats, provisos and gotchas!I get the feeling that its a bit of a rushed out product, I have been evaluating AG's for the last 2 months and there are a number of issues I have found so far. Specifically replication. Your distribution database cannot be part of an availability group meaning really you need to host it in a seperate SQL instance and possibly used clustering to provide any sort of HA. Although there is the option to utilities 2 or more of your servers used as part of the AG solution to host an additional clustered instance with shared storage.
Subscribers are not particularly well supported either and a real pain to deal with if part of a AG.
Not sure about rushed - it's just a very difficult thing to do robustly and scalably. Leaving out replication isn't surprising. In my experience (dedicated SQL Server engine work, mostly consulting, since 6.5 first came out) very few installations use replication. Some of the issues have been addressed in patches. But again, HA/DR itself is VERY hard to do right and I have come across very few clients that had things done up correctly. And most were HORRIBLY messed up!
Best,
Kevin G. Boles
SQL Server Consultant
SQL MVP 2007-2012
TheSQLGuru on googles mail service
Viewing 7 posts - 1 through 6 (of 6 total)
You must be logged in to reply to this topic. Login to reply