SQL Server Performance

basic Quorum question....

Discussion in 'SQL Server Clustering' started by wlandymore, Nov 29, 2005.

  1. wlandymore New Member

    I've been looking into clustering 2 SQL 2000 servers together to achieve redundancy, but I have a question about the quorum.
    In all of the documents I've read there is always one of these, but if that's the case you have a single point of failure, where if the quorum goes down, everything goes down.

    How can you have a setup with 2 SQL 2000 servers where you eliminate that point of failure?
    Could you have a setup with two quorums that could somehow sync together? Or no quorum at all and then have both SQL servers sync with each other so that if one went down it just fails over to the other, but they're not BOTH dependant on one resource.
  2. Argyle New Member

    A SQL Server cluster is a failover cluster, not a data redundancy cluster. If you have two sql instances in a cluster they are totally independent of eachother. Only one quorum drive exist in a cluster, doesn't matter if you run SQL Server or not in the cluster. You put the quorum drive on a RAID disk in the shared storage to avoid the single point of failure of just using one disk.

    More info:

    SQL Server 2000 Failover Clustering
    http://www.microsoft.com/technet/prodtechnol/sql/2000/maintain/failclus.mspx

    How to Install SQL Server 2000 Clustering
    http://www.sql-server-performance.com/sqlserver2000_clustering_install_main.asp
  3. wlandymore New Member

    but then lets say that the motherboard fails on the computer with the RAID. The system goes down...

    Isn't there some way to set it up so that you don't have a single point of failure of that nature?

    Like if you had two machines with SQL and storage on them and then if anything about them failed, it would go to the second machine with SQL and storage on it to pick up the slack.
  4. bradmcgehee New Member

    In a cluster, the quorum, along with the data, are stored on a shared array. This array should be as redundant as possible (RAID) as if it fails, the the entire cluster fails. As Argyle notes, failover clustering does not protect data. A cluster is designed to protect from hardware failure, OS failure, or application failure, and if a node fails because any of these reasons, then a failover will occur and things are good. If the shared array (where the quorum is located) fails, the entire cluster fails. For very mission critical setups, you can use Database Mirroring in SQL Server 2005, or log shipping in 2000 or 2005, to help keep your data more highly available.

    -----------------------------
    Brad M. McGehee, MVP
    Webmaster
    SQL-Server-Performance.Com
  5. Twan New Member

    There is also the option with Windows 2003 to set up geographically seperated clusters, where there is no shared components at all, this relies on fast networks between the cluster nodes and will increase your storage requirements by a factor of n where n is the number of nodes in the cluster

    Cheers
    Twan
  6. wlandymore New Member

    That sounds good actually.

    This is only going to be a two node cluster and there really shouldn't be any outgrowth.

    Is there a place that there is a little documentation about setting up a cluster of this nature with SQL 2000?
  7. Twan New Member

    Hi ya,

    www.microsoft.com/windowsserver2003/techinfo/overview/clustergeo.mspx is the Microsoft area for this

    Cheers
    Twan
  8. wlandymore New Member

    Thanks for all the help

Share This Page