SQL CLuster and Instances

  • Hi,

    Anyone knows for fact that in a SQL clustered environment you can only have one instalnce installed? 

    I have two large servers that are clustered (MS clustering win2k3) and need to create 6 instances of SQL2000 (3 on each node).  Is that possible?  If yes, is there any good docs on the how tos?

     

    Thanks

    A.

     

  • Multiple instance clusters are supported. See Brian Knight's articles on clustering (most of it is applicable to Win2k3).

    Also, you might check out the high availability guides from MS:

    Microsoft SQL Server 2000 High Availability Series

    You're biggest problem is going to be resource contention, especially memory. Keep in mind you need to consider the case where all 6 instances are running on a single node and have the memory and processors on a single box to handle that.

    K. Brian Kelley
    @kbriankelley

  • Also, make sure you have the appropriate licenses. You are setting up an Active/Active cluster (3 instances on each node). That means you have to have a license for each node. (Rule of thumb-any ACTIVE node must have it's own licenses).

    -SQLBill

  • No problem. Just make sure you install only named instances and do not begin with a defautl instance. Only named instances is the way to go.

    Here I support 2 clusters of 2 machines with PROD having 4 instances in active/active - 2/2 and a cluster with ACC/SYS/DEV  with on one node 4 ACC instances and on the other node the SYS  and DEV instances.

  • Do I need to have a sperate drive associated with each instance, or can all instances be on the same drive, same IP? (I have different drives associated with Log, Data and backup files).

     

    Thanks everyone

     

  • You don't need separate drives, but for each group of instances that share a drive, they must belong to the same cluster group and have the disk resources as dependencies for the instance resources.

    i.e. you have instances INST1 - INST6 - and INST1-INST3 use D:, E: and F: and INST4-INST6 use G:, H:, and I:

    cluster group 1 would have the resources for INST1-INST3 and DISK_D, DISK_E and DISK_F (each instance dependent on the drives)

    cluster group 2 would have the resources for INST4-INST6 and DISK_G, DISK_H and DISK_I (each instance dependent on the drives)

    This would also allow you to split the groups across nodes.

    The problem comes in when you mix/match because the disk resource is only available on a single node at a time...

  • I would recommend to do so. You do not want an instance to be offline because another instance needs to be offline for maintenance. Why would you consider not to?

Viewing 7 posts - 1 through 6 (of 6 total)

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