none
Requirements for Always On Feature and Availability groups in SQL Server 2012

    Question

  • I am trying to test SQL Server Always On Feature, Availability Groups in SQL 2012. I have one basic question in terms of how it is different from 2008 Install. I have highlighted my questions in Italics, please try to answer each my questions

    In SQL 2008 install on a 2 cluster, I perform the Install on the first node and then perform 'Add Node' on the second node to complete the Install. If it were 2012, to make use of this feature, do I have to install 2 separate Instances of SQL on the 2 nodes?

    This url (http://i.brentozar.com/sql-server-2012-alwayson-availability-groups-setup-checklist.pdf) says 

    8. Choose standard drive letters because the SQL Server data/log file locations must 
    be exactly the same across all nodes. The C drive is obviously standard for the 
    system, but you won't be installing SQL there. Please. Pretty please.

    In point 8, it says the drive location has to be exactly the same across all nodes. In 2008, a drive letter is handled by only one node at a given time. How can I have exactly same location across all the nodes? Does it mean I have to install 2 separate Installs (and not 'Add Node' Install)?

    How many IP addresses do I need to perform the Insall? In 2008 I needed just one Virtual IP for the SQL Virtual Name and performed the Install. In 2012 install with AG groups, how many IP's do I need?

    Thank you

    • Moved by Tom Phillips Tuesday, June 25, 2013 3:15 PM HA question
    Tuesday, June 25, 2013 2:25 PM

Answers

  • Satya,

    AlwaysOn Availability Groups is not the same as SQL Server Failover Clustered Instance. Think of AlwaysOn Availability Groups as database mirroring configuration sitting on top of a Windows Server Failover Cluster. This means that you need to install separate SQL Server instances - whether they be clustered or standalone - and use those instances as secondaries in an Availability Group configuration. The configuration will be a bit complex if you are considering implementing Availability Groups between clustered instances (or between a clustered instance and a standalone instance) because you now have to deal with a single Windows Server Failover Cluster that spans all of the instances in the Availability Group. This is also the reason why you can have the same drive letter on your SQL Server instances because you do not share disks in an Availability Group configuration (similar to not having shared disks in a database mirroring configuration.)

    For IP addresses, it will depend on whether or not you are configuring AlwaysOn Availability Group in all standalone instances or a combination of clustered and/or standalone instances. For each Availability Group listener name, you would need a virtual IP address. So, if you decide to have 2 Availability Groups in your instance, you will need 2 virtual IP addresses for each listener name


    Edwin Sarmiento SQL Server MVP
    Blog | Twitter | LinkedIn

    • Marked as answer by Satya Krishna Tuesday, June 25, 2013 5:57 PM
    Tuesday, June 25, 2013 5:16 PM

All replies

  • Satya,

    AlwaysOn Availability Groups is not the same as SQL Server Failover Clustered Instance. Think of AlwaysOn Availability Groups as database mirroring configuration sitting on top of a Windows Server Failover Cluster. This means that you need to install separate SQL Server instances - whether they be clustered or standalone - and use those instances as secondaries in an Availability Group configuration. The configuration will be a bit complex if you are considering implementing Availability Groups between clustered instances (or between a clustered instance and a standalone instance) because you now have to deal with a single Windows Server Failover Cluster that spans all of the instances in the Availability Group. This is also the reason why you can have the same drive letter on your SQL Server instances because you do not share disks in an Availability Group configuration (similar to not having shared disks in a database mirroring configuration.)

    For IP addresses, it will depend on whether or not you are configuring AlwaysOn Availability Group in all standalone instances or a combination of clustered and/or standalone instances. For each Availability Group listener name, you would need a virtual IP address. So, if you decide to have 2 Availability Groups in your instance, you will need 2 virtual IP addresses for each listener name


    Edwin Sarmiento SQL Server MVP
    Blog | Twitter | LinkedIn

    • Marked as answer by Satya Krishna Tuesday, June 25, 2013 5:57 PM
    Tuesday, June 25, 2013 5:16 PM
  • Thank you very much for the detailed explanation, that clarified my questions
    Tuesday, June 25, 2013 5:57 PM
  • I have one more question. I need to test out a SQL 2012 Always On Feature.

    I have an existing Active-Passive cluster with a SQL 2008 R2 Instance installed on it already. Is this a good candidate to test the 2012 Install with AG feature. I understand that I need to present new drives with same names on both the nodes of the cluster, but wanted to know if the 2012 Install works on a cluster that already has 2008 R2 Instance in Active Passive mode.

    Tuesday, June 25, 2013 6:42 PM
  • You can upgrade a SQL Server 2008 R2 failover clustered instance to SQL Server 2012 so you can configure AlwaysOn Availability Group. This instance will become your primary replica. To add a secondary replica - whether clustered or standalone - you need to add the machine first as a node in this existing Windows Server Failover Cluster. With clustered instances, you do not have the option to do automatic failover. For more information, check out this MSDN article

    http://msdn.microsoft.com/en-us/library/ff929171.aspx


    Edwin Sarmiento SQL Server MVP
    Blog | Twitter | LinkedIn

    Thursday, June 27, 2013 9:37 PM
  • You can add stand alone sql2012 instances to both the nodes & do an AOAG setup.

    yup

    Tuesday, July 02, 2013 3:28 AM