Example: Use node groups with clusters
Use node groups to define groups of nodes that are capable of hosting members of the same cluster. An application that is deployed to a cluster must be capable of running on any of the cluster members. The node that hosts each of the cluster members must be configured with software and settings necessary to support the application.
By organizing nodes that satisfy the application requirements into a node group, you establish an administrative policy that governs which nodes can be used together to form a cluster. The people who define the cell configuration and the people who create server clusters can operate with greater independence from one another, if they are different people.
Example 1:
Assume the following information:
- A cell is comprised of nodes one to eight.
- Each node is a managed node, which means that each node is configured with an Application Server.
- Nodes six, seven, and eight are additionally configured as WebSphere Business Integration Server Foundation nodes.
- All nodes are either z/OS system nodes from the same sysplex, or some combination of distributed platform nodes and IBM i platform nodes.
- By default, all the nodes are in the default DefaultNodeGroup node group.
Applications that exploit WebSphere Business Integration Server Foundation functions can run successfully only on nodes six, seven, and eight. Therefore, clusters that host these applications can be formed only on nodes six, seven, and eight. To define a clustering policy that guides users of the WebSphere cell into building clusters that can span only predetermined nodes, create an additional node group called WBINodeGroup, for example. Add to the node group nodes six, seven, and eight. If we create a cluster on a node from the WBINodeGroup node group, the system allows only nodes from the WBINodeGroup node group to be members of the cluster.
Example 2:
Assume the following information:
- A cell is comprised of nodes one to six.
- Each node is a managed node, which means that each node is configured with an Application Server.
- Nodes one to four are some combination of distributed platform nodes and IBM i platform nodes.
- Nodes five and six are nodes on the z/OS operating system and are in the PLEX1 sysplex.
- The deployment manager is on a distributed platform node.
- Nodes one to four are members of the DefaultNodeGroup node group by default.
- We created empty PLEX1NodeGroup node group to group the z/OS operating system nodes on the PLEX1 sysplex.
- You joined the nodes on the z/OS operating system to the PLEX1NodeGroup node group when we added them to the cell. Nodes on the z/OS operating system cannot be in the same node group with the distributed platform nodes.
Applications that exploit z/OS functions in the PLEX1 sysplex can run successfully on nodes five and six only. Therefore, clusters that host these applications can be formed only on nodes five and six. The required separation of distributed platform nodes and IBM i platform nodes from z/OS system nodes establishes a natural clustering policy that guides users of the Application Server cell into building clusters that can span only predetermined nodes. If we create a cluster on a node from the PLEX1NodeGroup node group, the system allows only nodes from the PLEX1NodeGroup node group to be members of the cluster.
Related tasks
Viewing, adding, and deleting node group members Viewing, configuring, creating, and deleting node groups