Planning Ahead - Single Node Clusters

  • Comments posted to this topic are about the item Planning Ahead - Single Node Clusters

    --
    Andrew Hatfield

  • What a great idea. It just so happens that I have arrived at work this morning to find our main file server is not available because our IT guys decided to swap the hardware last night and it didn't go quite according to plan!

    I think I'll forward them a copy of your article in the hope they take the hint. The irony is that we have invested in blade servers and shared storage (SAN). We have virtualised everything but not created the servers as single node clusters. After reading your article it seems we might have missed a trick.

  • Great to see this and to know that some one els is also thinking like we do.

    We already using single node clusters in our SQL environment now for more than 3 years.

    Especially to make the sql installation server independed.

    If the server needs to be replaced we add the new one to the cluster, do the switch over and than remove the old one if everything is running ok.

    So it's not only good for scaling out but also for server replacement without a lot of down time and migration issues.

  • Thanks for that.

    Don't know why I never thought of this before, coming from a VMS background, where clustering systems is an (almost) trivial task that we took for granted because, well, it was just what you did.

    In particular, your thoughts about upgrading - this was exactly what we always did when upgrading systems, add the new hardware to the cluster, move the apps to the new system and move the old one to the 'development cluster'.

    Moving in the small business circles that I have over the last 10 years, I'd all but forgotten about clustering, but you've given me a reason to start building up my skills in Windows clusters.

    /Ryan

  • Thanks for sharing this tip. I can confirm that this works having created a single node cluster 3 years ago on SQL 2005. However, I only had a single node cluster for 4 months. When the new budget year was approved, I was able to add a second node.

  • Good article, and very good idea about planning ahead. This methodology takes more work to set up, but can save you hours or even days of work in a replacement or hardware failure.

    Tim Mitchell, Microsoft Data Platform MVP
    Data Warehouse and ETL Consultant
    TimMitchell.net | @Tim_Mitchell | Tyleris.com
    ETL Best Practices

  • Good idea. I have noticed clustered servers are more sensitive than non-cluster servers. They automatically restart services by failing over whenever something seems to be wrong. I think of it as an automated reboot of the server.

    How does a single node handle a failover issue? Does it just restart itself or waits for someone to handle it?

    David Bird

  • Thanks for good insight of planning ahead. Appreciate your help to the community. Venkat.

  • Can anyone add additional insight on steps for moving a single SQL Server instance into a clustered environment. Thanks.

    Regards,

    Reggie

  • It try to restart the resource x times, but if it keeps failing it becomes failed.

  • Andrew,

    Great job and I'm glad you wrote the article. When I saw you mention it, I was surprised as it seems many other people were today. It's a great idea.

  • Andrew,

    Well done!

    How hard is it to add a new node to SQL Cluster? Does SQL setup happen auto-magically (like when you install SQL on a 2 node cluster) when you add the node or are additional steps required?

    Mark

  • The article seemed to lack software requirements for a MSSQL cluster. We operate different versions of MSSQL, from what I've read only SQL Enterprise supports clustering. With that said, there's a high cost involved with the software itself. Anyone with ideas on clustering with a lower end version of MSSQL like Standard, or Web? We're considering "log shipping" and backing up to a central location so we can restore from a second active or standby SQL server.

  • Hello Admin,

    SQLServer 2005 standard version supports only 2 node clustering, where as Enterprise version supports multiple nodes.

    http://www.microsoft.com/Sqlserver/2005/en/us/compare-features.aspx

  • By the way many Hardware restrictions have been eliminated in SQL 2008


    * Noel

Viewing 15 posts - 1 through 15 (of 19 total)

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