Dynamic clusters

A dynamic cluster is a server cluster that uses weights and workload management to balance the workloads of its cluster members dynamically, based on performance information that is collected from the cluster members. Dynamic clusters enable application server virtualization.

A dynamic cluster is an application deployment target that can expand and contract depending on the workload in the environment. Dynamic clusters work with autonomic managers, including the application placement controller and the dynamic workload manager to maximize the use of the computing resources. Dynamic clusters are required for many of the Intelligent Management autonomic functions, including high availability and service policies.

For complete life-cycle management servers, the product controls the creation and deletion of server instances and can start and stop servers. For assisted life-cycle management servers, the product can control the state of servers by stopping and starting servers from a pool of predefined server instances.

Dynamic cluster membership

Two options exist for adding members to a dynamic cluster: automatically define cluster members with rules or manually define cluster members.

• Automatically define cluster members with rules

  By automatically defining cluster members with rules, we can create a subexpression that automatically selects nodes to host dynamic cluster members based on different node properties. This subexpression is called a membership policy. After creating the membership policy, we can preview the node membership before you finish creating the dynamic cluster.

  Automatically defining cluster members with rules is available only for servers that have complete life-cycle management. After creating the dynamic cluster with a membership policy, dynamic cluster instances can start on any of the selected nodes. If nodes become available that meet the criteria of the membership policy, dynamic cluster instances can also start on these nodes.

• Manually define cluster members

  When you manually define cluster members, you statically define which servers are cluster members by selecting servers to add to the cluster. You use this option instead of the membership policy for the following reasons:

  • You have an existing static cluster to convert to a dynamic cluster.

  • We are using assisted life-cycle management servers. Assisted life-cycle management servers cannot be created from the console. With this option, we create representations of the servers as cluster members. These members must be homogeneous, that is, be all of the same server type, for example, a group of BEA WebLogic servers. The same version of the middleware software must be installed on all the nodes in the dynamic cluster, and you must deploy the same applications to these servers before creating the dynamic cluster.

Server templates

A server template is a copy of a server configuration that can be used as a starting point when a server is added to the dynamic cluster. Predefined templates exist for different middleware server types. We can also define our own server templates.

Dynamic cluster server templates

After creating a dynamic cluster, the dynamic cluster server template defines the properties for all of the members in the dynamic cluster.

When we make a change to a dynamic cluster server template, this will overwrite any changes you made that are unique to an individual member of that cluster.gotcha

Cluster instances

We can control the creation and management of cluster instances for the dynamic cluster. These options include:

• Create a minimum and maximum number of cluster instances.

• Stopping cluster instances when other dynamic clusters need resources.

• Allow more than one cluster instance to start on the same node, also known as vertical stacking. With vertical stacking, we can improve bottleneck conditions within an application. The stacking number defines how many cluster instances can start on a single node.

• Specify if cluster instances from other dynamic clusters can start on the same node, also known as dynamic cluster isolation.

Operating modes

Dynamic clusters act differently depending on the operating mode. Choose one of the following options for mode of operation:

• Manual. In manual mode, the dynamic cluster is no different from the standard application server environments with static clusters. Manual mode does not support application placement, or runtime task suggestions. The autonomic request flow manager and dynamic workload management (DWLM) can work with the cluster.

• Supervised. In supervised mode, the environment provides information about required corrective actions by generating runtime tasks. We can accept or deny the recommendations of the autonomic managers in the task management panel in the console. To manage runtime tasks, click System administration > Task management > Runtime tasks.

• Automatic. In automatic mode, the environment takes corrective actions automatically.

Important: To use dynamic application placement, click automatic or supervised as the mode of operation.

If we are using manual mode for the autonomic request flow manager (ARFM) on either the cell level or deployment target level, then we also must put the dynamic clusters in manual mode. We can also use static clusters when ARFM is in manual mode for the cell or deployment target. If ARFM is in automatic mode, we can use any of the operating modes for the cluster.

Related concepts

Dynamic cluster isolation

Dynamic application placement

Related tasks

Configure vertical stacking

Intelligent Management: static clusters versus dynamic clusters