Figure 39-3 shows typical continental clusters, consisting of a cluster in Dallas
(standby) and another in Boston (production). The two clusters can be an
active-passive (asymmetric) or active-active (symmetric) pair. Let’s assume an
asymmetric configuration where the Boston cluster is running mission-critical
applications with servers configured as a local cluster. Applications failover
from node to node as necessary. The data in Boston is replicated over the WAN
to storage connected to hosts in the Dallas cluster. The Dallas cluster is also
configured with the same applications as on the Boston cluster. However, the
nodes in Dallas do not run the applications under normal circumstances. If
none of the servers in the Boston cluster can run the applications, or if there is
a site-wide power outage (or any other disaster) in Boston, one of the servers
in the Dallas cluster starts the applications. The two clusters can also run in
active-active mode, where they monitor each other for failures and data is
replicated in both directions. In addition, the cluster servers in Dallas and
Boston may run applications that are not part of the cluster configuration.