Sunday, October 26, 2008

Clustering in MicroStrategy

Clustering is a popular strategy in which two or more servers/machines are configured to provide common services in a unified manner such that they behave like a single server. Each machine in the cluster is called a node. Even though it may sound very hi-fi, Clustering is one of the simplest thing in MicroStrategy. It aims for Load balancing as well as fail-over. One can cluster Intelligence server running on differ OS. eg, AIX, UX, Solaris, RH Ent Linux provided following condition are met.
  1. All MicroStrategy Intelligence Servers MUST point to the same metadata
  2. All MicroStrategy Intelligence Servers MUST have the projects in the same state. (i.e., if Node A has Project A in a 'Loaded' state, then Node B must also have Project A in a 'Loaded' state).
  3. From each machine where I-server is installed, one should be able to browse the folders for Cache and History List. (This is aachieved by Samba on Unix Environment)
Posted by Ashish Tiwari @ GMT 5:30 PM
Labels: ,

3 comments:

  1. AnonymousDec 8, 2008, 3:14:00 PM

    I am sorry but i think you have it wrong as clustering can also be asymmetric ie project state on the clustered server can be different.(Loaded on one unloaded on another)

    ReplyDelete
  2. Ashish TiwariDec 11, 2008, 7:55:00 AM

    Asymmetric clustering is used by very less number of clients. And even in that method remains same (almost).

    ReplyDelete
  3. AnonymousDec 31, 2008, 5:25:00 PM

    Just making a correction to the following statement.
    "i.e., if Node A has Project A in a 'Loaded' state, then Node B must also have Project A in a 'Loaded' state"

    ReplyDelete

Note: Only a member of this blog may post a comment.

Subscribe to: Post Comments (Atom)