KB440866: What are the prerequisites for clustering Intelligence Servers?

 
Strategy prerequisites
 
• You must have purchased an Intelligence Server license that allows clustering. To determine the license information, use the License Manager tool and verify that the Clustering feature is available for Intelligence Server.
 
• The computers to be clustered must all have the same version of Intelligence Server installed.
 
• All Strategy projects on the clustered machines must be based on the same metadata.
• At least one project must be defined in the metadata.
 
• No more than one Intelligence Server can be configured for a single machine. Multiple instances of Intelligence Server should not run on the same machine for clustering purposes.
 
• The user account under which the Intelligence Server service is running must have full control of cache and History List folders on all nodes. Otherwise, Intelligence Server will not be able to create and access cache and History List files.
 
• Server definitions store Intelligence Server configuration information. Strategy strongly recommends that all servers in the cluster use the same server definition.This ensures that all nodes have the same governing settings.
• Developer must be installed on a Windows machine to administer the cluster. This version of Developer must be the same as the version of Intelligence Servers. For example, if the Intelligence Servers are running Strategy Analytics Enterprise, Developer must also be Analytics Enterprise.
 
• You must have access to the Cluster view of the System Administration monitor in Developer. Therefore, you must have the Administration privilege to create a cluster.
 
• The computers that will be clustered must have the same intra-cluster
communication settings.
 
• The same caching method (localized or centralized caching) should be used for both result caches and file-based History Lists.
 
 
Server prerequisites
 
• The machines to be clustered must be running the same version of the same operating system. For example, you cannot cluster two machines when one is running on Windows 2008 and one is running on Windows 2003.
 
• Load balancing and system configuration are simpler if identical hardware is used for each of the clustered nodes.
 
• If you are using time-based schedules in a clustered environment, all the nodes in the cluster must have their clocks synchronized.
 
• The RDBMS containing the metadata and warehouse instances must already be set up on machines separate from the Intelligence Server nodes.
 
• Information on the clustered configuration is stored in the metadata, so the machines to be clustered must use the same metadata repository. The metadata may be created from any of the nodes, and it needs to be set up only once. When you create or modify the server definition in the Strategy Configuration Wizard, you can specify either a new or an existing metadata repository for Intelligence Server to use.
 
• The required data source names (DSNs) must be created and configured for Intelligence Server on each machine. Strategy strongly recommends that you configure both servers to use the same metadata database, warehouse, port number, and server definition.
 
• All nodes must join the cluster before you make any changes to any governing settings, such as in the Intelligence Server Configuration Editor.
 
Prerequisites for Windows clustering
 
• On all machines to be clustered, each network card must be configured to enable Netbios over TCP/IP. Otherwise, cache sharing is not possible using Netbios names (ClusterCaches, ClusterCube, and ClusterInbox).
 
• When Intelligence Server is installed, the last step is to choose a user identity under which the service will run. To run a clustered configuration, the user must be a domain account that has a trust relationship with each of the computers in the cluster. This allows resources to be shared across the network.
 
• The service user’s Regional Options settings must be the same as the clustered system’s Regional Options settings.
 
Prerequisites for UNIX/Linux clustering
 

• Strategy strongly recommends that all servers in a cluster use the same server definition. Therefore, in some cases you cannot specify the cache location with an absolute path such as /<machine_name>. This occurs because the location would have to be different for each server machine. To solve this problem, use relative paths and soft links. A soft link is a special type of UNIX file that refers to another file by its path name. A soft link is created with the ln (link) command:

 
ln -s OLDNAME NEWNAME
 
where
 
OLDNAME is the target of the link, usually a path name.
NEWNAME is the path name of the link itself.
 
Most operations (open, read, write) on the soft link automatically de-reference it and operate on its target (OLDNAME). Some operations (for example, removing) work on the link itself (NEWNAME).
 
• Confirm that each server machine works properly, and then shut each down.
  KB440866