August 5, 2013 at 9:48 am
Hello All,
I have a question on SQL Server 2008 R2 Std Edition clustering. I am trying to setup clustering for our SQL Server staging environment. We have a lot of applications and all of them are using the name of our database server which is CPNDB01 in their app config connection strings. But if I setup clustering and bring up a new node CPNDB02 and perform a failover do you think I need to change the name of the database server in the connection string to CPNDB02 for all of our applications?
I know we need to create a virtual ip address for SQLServer as well and we can use that in the connection string to prevent the name change every time a failover or a failback is performed. But what if I have to use only Server name but no IP in my app connection string?
Thanks for your inputs
“If your actions inspire others to dream more, learn more, do more and become more, you are a leader.” -- John Quincy Adams
August 5, 2013 at 10:07 am
No, when creating a cluster you will assign a virtual machine name (not the same as a VM OS) this is then the name you will use in your connection string. Obviuosly including any instance name if its a named instance.
For Example if you have two hosts participating in a cluster called DBSERVERA ahd DBSERVERB, you virtual name could be DBSERVER. Note that the Virtual name (DBSERVER) will requires its own unique IP address which will differ to those used for DBSERVERA AND DBSERVERB.
MCITP SQL 2005, MCSA SQL 2012
August 5, 2013 at 10:15 am
OK. But the requirement is strange. This is a company with all new IT and development team. None of us are completely aware of where the actual 40+(none of us know the exact number still) app configs are sitting and the management wants us to build a cluster for our SQL Server which is a single point of failure without breaking any apps. So this is not going to be something from scratch. But I will need to bring up a new node and use the drives on LUN as a shared storage. We only have a small downtime so will need to set this up during this time. So I would still want to use the name of the current database server at all times(during a failback or failover). So is it possible to use the node1 name as the virtual name?
I know there is a third party tool double take that can help with current situation with their HA solution. But I was looking for something native to SQLServer.
“If your actions inspire others to dream more, learn more, do more and become more, you are a leader.” -- John Quincy Adams
August 5, 2013 at 10:29 am
If I understand your auestion correctly, then no you cannot use the node name I.e DBSERVERA as the server names on a windows domain must be unique and any cluster regardless of the number of nodes will require a virtual name independent of the host name.
You can however create your cluster as a single node and additional Nodes and or LUNs at anytime without downtime.
You will however have an interruption to the applications while you change there connection strings to the new virtual name of the cluster.
To minimise down time you could look at log shipping or mirroring to transfer your data to the new cluster prior to the switch over.
MCITP SQL 2005, MCSA SQL 2012
August 5, 2013 at 12:14 pm
Then I think double take's HA solution to protect SQL Server is something that I shd be looking at, since we are not interested in implementing something from scratch.
“If your actions inspire others to dream more, learn more, do more and become more, you are a leader.” -- John Quincy Adams
August 6, 2013 at 10:27 am
Sapen (8/5/2013)
Hello All,I have a question on SQL Server 2008 R2 Std Edition clustering. I am trying to setup clustering for our SQL Server staging environment. We have a lot of applications and all of them are using the name of our database server which is CPNDB01 in their app config connection strings. But if I setup clustering and bring up a new node CPNDB02 and perform a failover do you think I need to change the name of the database server in the connection string to CPNDB02 for all of our applications?
I know we need to create a virtual ip address for SQLServer as well and we can use that in the connection string to prevent the name change every time a failover or a failback is performed. But what if I have to use only Server name but no IP in my app connection string?
Thanks for your inputs
If you're smart and have a big enough maintenance window there should no reason why you cant achieve what you aim to do.
You'll need to pre configure the cluster and have it ready to go with disks attached and formatted.
You'll need to script any objects to move across, immediately you'll need logins and user databases, the rest could be done afterwards.
At the appropriate time shutdown the old servers, making sure you have your backups of your user databases ready
Install the new instance providing the existing name CPNDB01 and IP, restore logins and dbs and also any jobs, etc.
I did something similar before in a weekend for 3 existing systems into 3 new clusters.
-----------------------------------------------------------------------------------------------------------
"Ya can't make an omelette without breaking just a few eggs" 😉
Viewing 6 posts - 1 through 5 (of 5 total)
You must be logged in to reply to this topic. Login to reply