The following list introduces IBM terminology used for clustering.
A cluster is a collection of interconnected complete computers that work together as a single unified computing resource. The cluster is made up of one or more cluster nodes and is identified by a name comprised of 10 or fewer characters.
A cluster node is any system or logical partition (LPAR) that is a member of a cluster. IBM i clustering supports up to 128 nodes in a cluster, but each node can be defined to only one cluster at a time. The set of nodes that are defined to a cluster is referred to as the cluster membership list.
Cluster communication uses TCP/IP protocol to provide communication paths between each node in the cluster. A node must be connected to the cluster using an IP network in order to communicate with other nodes in the cluster. IBM recommends that you establish a dedicated path that is not shared by users or other network traffic. For simplicity, the name of a node is often the same name as the host or system name. This name is then mapped to an 8-character cluster node identifier that is associated with one or more Internet Protocol (IP) addresses that represent a system.
Cluster resources are the resources that are required to be highly available by your business and are available to the nodes within a cluster. Cluster resources can be either moved or replicated to one or more nodes within a cluster. Examples include applications, data libraries, devices, and disk units. Resources are identified in cluster resource groups and controlled through cluster resource group exit programs.
A cluster resource group (CRG) is an IBM i system object that identifies a collection of cluster resources to be monitored and managed as a single unit. Each CRG defines the relationship between the nodes associated with those resources in a recovery domain that determines the role of each node in the CRG as well as the degree to which each can participate in events such as synchronizing or performing a recovery action. Several types of CRGs are available. Each of the following CRG types is designed for a specific type of cluster resource: application, data, device, and peer.
Each CRG has a CRG exit program that is called on each active node in the CRG’s recovery domain in response to a cluster event. The exit program manages cluster events for the environment established by the CRG. All possible cluster events have a pre-determined response in the exit program code.