Oracle database clustering
If you use an Oracle database system to host the metadata repository database or the IBM® InfoSphere® Information Analyzer analysis database, you can increase the availability of the metadata repository tier by implementing Oracle Real Application Clusters (RAC).
An Oracle RAC configuration consists of multiple computers, each of which runs an Oracle instance, which is made up of processes that provide access to stored data. Each Oracle instance has access to a common database file system across a high-speed network. If a computer or Oracle instance process fails, other Oracle instances continue to provide service. Oracle database cluster management software (Oracle Clusterware) monitors and manages the Oracle RAC configuration and tries to restart failed Oracle instances.
Client processes access the Oracle RAC configuration in the same way that they access a single Oracle database instance. With Oracle RAC 11g, a virtual IP address is assigned to each server in the cluster. If a node in the cluster fails, the virtual IP address fails over to another node and client processes can automatically reconnect.
Nodes in an Oracle RAC cluster require a separate, private high-speed network for messaging between nodes. This network is called the interconnect.
The following figure shows an IBM InfoSphere Information Server topology that includes Oracle RAC. The two Oracle RAC node computers share access to a database farm.
If a node in the RAC cluster fails, other nodes continue to provide service.
For greater high availability, you can implement an Oracle RAC on Extended Distance Clusters configuration. In this configuration, nodes in the cluster are located in physically separated sites.
To install InfoSphere Information Server in an Oracle RAC configuration, install to a single node in the configuration. Then perform additional steps to cause InfoSphere Information Server to operate with the complete Oracle RAC configuration.