Database Reference
In-Depth Information
Inter-node connectivity: RAC employs a shared-everything approach. A cluster database
has one global SGA that spans all the nodes on which the database is up and running. For
various reasons such as the transfer of database blocks over the network or the exchange of
cluster-wide enqueue information database instances must communicate with one another.
Resource management: Everything Clusterware manages is referred to as a resource. The ASM
instance, network, database, services—literally everything in a RAC environment is a resource.
Each resource has an associated profile, stating on which node to execute the resource, how
to monitor the resource's health, and what to do if a resource has stopped working. Local
resources by definition exist on one node only and cannot be failed over to another cluster
node. Cluster resources however can be moved within the cluster.
Figure 2-2 shows a typical configuration for a Real Application Cluster consisting of two nodes. Note that the
figure does not take any of the new 12c features into account. The management interface used for out-of-band
management is not shown in order to simplify the diagram.
Figure 2-2. Conceptual design of a RAC environment
 
Search WWH ::




Custom Search