Overview

A group is a logical collection of space members. Groups are used to define policies for replication, failover, and load-balancing.

A cluster must contain at least one group. If there is exactly one group, it should contain all the spaces in the cluster. If there is more than one, each group must have a unique name, and may contain as many spaces as you like. Groups cannot contain sub-groups.
A group can also have one or more types of policies. For example, a certain group can define both a replication and a load balancing policy. However, it is illegal to define more than one policy of the same type for the same group in the same cluster (e.g. a group cannot have two different replication policies).
Groups with different types of policies may overlap – for example, a space belonging to a replication group can also belong to a failover group. However, it is illegal for a space to belong to more than one group with the same policy type.

The following example shows how groups are defined in the cluster configuration file:

<cluster-config>
  ...
  <groups>
    <group>
      <group-name>repl-example-group</group-name>
      <group-members>
        <member-name>container-name:sp1</member-name>
        <member-name>container-name:sp2</member-name>
      </group-members>
      ...
    </group>
  </groups>
</cluster-config>

It is important to note that the <groups> tag is at the same level as the <cluster-members> tag, which contains the unique names and access URLs of all cluster spaces. The member names listed under <groups> refer to spaces listed under <cluster-members>. It follows that each space name is listed twice (or more) in the cluster configuration file, and each occurrence must match exactly.

The <group> tag can contain additional tags, besides <group-name> and <group-members>, that define the policy of the group.