Has anyone implemented SQL Server 2012 Always On for disaster recovery

  • Am helping a client upgrade his servers (currently SS 2005 with Windows 2003 cluster). Talked to Dell sales rep this afternoon and he had not heard of anyone implementing Always On for DR. He said all his sales for SS2012 have been for active/passive cluster.

    I've attended several SQL Saturday sessions and webinars on Always On, but now are worried that nobody has really implemented it in production. I don't want the client to be on the technology tip by himself, but would like to take advantage of Always On if it really works.

    I would appreciate it if anybody could tell me if they have implemented Always On and what pros and cons they have found.

    Thanks in advance,

    Mike Byrd

    Mike Byrd

  • Mike Byrd (4/12/2013)


    Am helping a client upgrade his servers (currently SS 2005 with Windows 2003 cluster). Talked to Dell sales rep this afternoon and he had not heard of anyone implementing Always On for DR. He said all his sales for SS2012 have been for active/passive cluster.

    I've attended several SQL Saturday sessions and webinars on Always On, but now are worried that nobody has really implemented it in production. I don't want the client to be on the technology tip by himself, but would like to take advantage of Always On if it really works.

    I would appreciate it if anybody could tell me if they have implemented Always On and what pros and cons they have found.

    Thanks in advance,

    Mike Byrd

    Your sales person is either misinformed, speaking only based on knowledge that if people are making SQL Server Cluster they state so to ensure they get "certified" hardware or is straight-up lying to you to get you to buy more expensive "cluster certified" hardware. I have 4 clients that have or are in the process of setting up some form of 2012 "enhanced mirroring" (e.g. Always On) for HA/DR/read-only scale-out (or any combination of those 3). There are some amazingly compelling reasons and benefits, and given that it's foundational underpinnings is database mirroring which has been around since SQL 2005 SP1, it is pretty solid given that technically Always On is a v1+SP1 feature! The one con that I know of is a limitation of single-threading the stuff across and hardening it on secondary - which I have heard means a real-world 50MB/second throughput give or take. I note that that is a LOT of data/transactions and that 99.x% of the installed base will never get even close to that. Well, there is also the cost of SQL 2012 licensing as well as the requirement for a Windows Failover Cluster of all machines involved (including any on a different subnet).

    If you would like more information feel free to Private Message me.

    Best,
    Kevin G. Boles
    SQL Server Consultant
    SQL MVP 2007-2012
    TheSQLGuru on googles mail service

  • We just rolled out AlwaysOn in our reporting environment last week to replicate 2 databases from our data warehouse to allow our data analysts query the report databases without effecting the ETL processes during the day on the primary.

    The only issue we found when rolling out one of the databases was that we had to turn off transaction log backups on the primary while the replica ws being made on the secondary. Its very early days but so far all is good - the setup was very straighforward and we are not having any issues with latency - we set it up as Asynchronous-commit mode as our primary is in Melbourne and Secondary in Darwin.

    As I said its early days but so far so good!

  • Did you need to upgrade all of your SQL instances to 2012 in order to enable AlwaysOn?

  • We have implemented AlwaysOn Availability Groups (AG) on a 3-node cluster, consisting of 2 subnets: primary and DR.

    2 nodes on the primary site host a failover-cluster instance (FCI) in an active-passive configuration.

    The 3rd node on the DR hosts a standalone SQL instance, serving as a secondary AG replica in async-commit mode.

    No issues so far with this configuration, although we have observed transaction-log latencies of up to 20 minutes between the primary and secondary replicas.

    __________________________________________________________________________________
    SQL Server 2016 Columnstore Index Enhancements - System Views for Disk-Based Tables[/url]
    Persisting SQL Server Index-Usage Statistics with MERGE[/url]
    Turbocharge Your Database Maintenance With Service Broker: Part 2[/url]

  • @SSCommitted Did all of your database servers have to be upgraded to 2012 in order for your configuration to work?

  • fb.cs99 (5/8/2013)


    @SSCommitted Did all of your database servers have to be upgraded to 2012 in order for your configuration to work?

    Yes.

    Also all database servers need to be in a Windows Server 2008 R2 SP1 (or later) Failover Cluster.

    __________________________________________________________________________________
    SQL Server 2016 Columnstore Index Enhancements - System Views for Disk-Based Tables[/url]
    Persisting SQL Server Index-Usage Statistics with MERGE[/url]
    Turbocharge Your Database Maintenance With Service Broker: Part 2[/url]

  • I have implemented Always ON, here is the configuration

    I have 3 node cluster , 1 with stand alone sql server and 2 nodes are HA active/passive sql server cluster. all three nodes are in same subnet

    now i want to add another node for DR which is at some other location with different subnet. (Stand alone sql only)

    i have the following question

    will adding new node to the existing WINDOWS cluster bring down services in the cluster?

  • I have recently did setup 2012 always on with 2 active/passive node with different NW subnet adding another instance in the AAG to be the mirroing. This is working fine.

    "More Green More Oxygen !! Plant a tree today"

Viewing 9 posts - 1 through 8 (of 8 total)

You must be logged in to reply to this topic. Login to reply