Why and when to perform this task
Planning ahead for high availability support is important to avoid
the risk of a failure without failover coverage. The application server infrastructure
that is managed by a high availability manager includes cells and clusters.
These components relate closely to core groups, high availability groups,
and the policy that controls the high availability infrastructure.
The high
availability manager is designed to function in all of the supported WebSphere
Application Server topologies. However, a high availability-managed environment
must comply to the following rules:
- A cell in a high availability infrastructure is partitioned into one or
more core groups. WebSphere Application Server provides a default core group
as part of the high availability manager function. Additional core groups
can be created using the administrative console.
- A core group cannot extend beyond the boundaries of a cell, and it cannot
overlap with any other core groups.
- A cluster must be a member of only one core group. All of the individual
members of that cluster must be members of the same core group. This one-to-one
relationship between a cluster and a core group exists for both static and
dynamic clusters.
- Individual application servers that are part of the high availability
environment must also be part of a core group.
- An application server can only join a core group if its JVM can communicate
with all of the other online application servers that are part of that core
group. If a single application server can not open a connection to the JVM
or send a heartbeat to it, the application server is not joined to the core
group.
The following diagram illustrates what a cell might look like in a
high availability environment:
What to do next
Important: After you set up your WebSphere Application
Server environment to comply with all of the high availability-managed environment
rules, use the default core group to control this environment. DO NOT add
additional core groups unless your environment absolutely requires them. Also,
do not change the default configurations unless you are doing so to solve
a specific problem or situation. When you do make configuration changes, such
as changing the policy for a high availability group or moving core group
members between core groups in a multi-core group environment, make sure you
fully understand the effect such changes will have on your entire environment.