MSSQL 2008 R2 Failover Clustering Design
-
Thursday, April 05, 2012 5:13 AM
Hi guys,
I am currently putting all my focus in lab to develop a SQL 2008 R2 Failover Cluster between 2 virtual machines. My main objective is to have a failover between SQL Services so that the user can continue to resume connection even though 1 of the nodes is down. Please refer to the this image for my design: http://social.technet.microsoft.com/Forums/getfile/88987
Here are a few explanation of my design:
1. Hyper-V 1 and Hyper-V 2 are joined to create 2-node Failover Cluster.
2. Starwind Native SAN are installed on both physical servers.
3. Virtual Machine 1 and Virtual Machine 2 are created with the 2 physical servers as their hosts. And then they are joined to create another 2-node Failover Cluster.
4. I installed SQL 2008 R2 Failover Cluster on Virtual Machine 1, and add Virtual Machine to the SQL Failover Cluster later.
5. The SQL Failover Cluster name is 'SQLCluster'.
So.. What is your opinion on this design? Please let me know if you require further explanation and information. I really appreciate your time to give your opinion on this. Thanks!
-ridhuan-
- Edited by Vincent HuModerator Thursday, April 05, 2012 5:58 AM resize the pic and insert it
All Replies
-
Thursday, April 05, 2012 6:03 AMModeratorHi,If you want to build a Failover Cluster between several virtual machines on Hyper-V, the only shared storage supported by guest clusters is iSCSI storage.For more information, you can refer to:Hyper-V Guest Clustering Step-by-Step Guide
http://blogs.technet.com/b/mghazai/archive/2009/12/12/hyper-v-guest-clustering-step-by-step-guide.aspxHyper-V Guest Clustering
http://blogs.technet.com/b/sainfrateam/archive/2011/05/09/hyper-v-guest-clustering.aspxBy the way, if you want to build a SQL Server Faiover Cluster between Hyper-V virtual machiens, you can check the following blog.Vincent Hu
TechNet Community Support
-
Thursday, April 05, 2012 6:58 AM
Hi Vincent,
Thanks for the reply. I will check the links, thanks. Meanwhile, the Starwind Native SAN for Hyper-V is iSCSI solution, am I right?
-
Thursday, April 05, 2012 7:17 AMModeratorHi,No idea about Starwind Native SAN for Hyper-V.By the way, Microsoft iSCSI Target software is a free now, you can use it in your environment.Microsoft iSCSI Software Target 3.3
http://www.microsoft.com/download/en/details.aspx?id=19867Vincent Hu
TechNet Community Support
-
Thursday, April 05, 2012 3:12 PMHi,No idea about Starwind Native SAN for Hyper-V.By the way, Microsoft iSCSI Target software is a free now, you can use it in your environment.Microsoft iSCSI Software Target 3.3
http://www.microsoft.com/download/en/details.aspx?id=19867
Vincent Hu
TechNet Community Support
In his environment MS target requires one dedicated physical machine ($$$) and Windows 2008 R2 license ($$$). So he'll end with THREE machines and storage being a single point of failure. With StarWind he'll have TWO machines and complete redundancy. So your solution has hidden costs and it's not usable for true production (unless RTO is hours).
-nismo
- Edited by VR38DETTMicrosoft Community Contributor Thursday, April 05, 2012 3:17 PM
-
Thursday, April 05, 2012 3:16 PM
Hi Vincent,
Thanks for the reply. I will check the links, thanks. Meanwhile, the Starwind Native SAN for Hyper-V is iSCSI solution, am I right?
YES, absolutely. Your design is correct and should work.
We'll find out why your scenario is not working and will fix everything.
-nismo
-
Tuesday, April 10, 2012 6:06 AM
YES, absolutely. Your design is correct and should work.
We'll find out why your scenario is not working and will fix everything.
-nismo
Noted that. Thank you so much. Meanwhile, I have installed Starwind iSCSI SAN for my iSCSI target, and will wait for the solution from Starwind tech for the Starwind Native SAN for Hyper-V.
I have successfully configured the SQL Failover Cluster, and now testing it. This will be my test environment and what I expect;
1. SQL Cluster name : SQLBullrun
2. 1 software called GFI EventsManager use SQLBullrun as the database backend.
3. When node 1 in SQLBullrun failed (active node), the SQL service will failover to node 2, and there connection from GFI EventsManager to the SQLBullrun will lost for a short period.
4. Then after the SQLBullrun cluster name is up again, the connection will reestablish.
Can I have your precious opinion this setting? I am expecting the correct thing, right? I have searched for the simulation testing around the web, but none seems to provide a good example. Thank you for your assistance!
-ridhuan-

