Dynamic
cluster isolation ensures that dynamic cluster instances from different
dynamic clusters do not run on the same node, but it does not make
guarantees about how the system avoids a violation of the isolation
mode. Configuring strict isolation does not give a dynamic cluster
priority over any other dynamic clusters.
For example, you might
have an environment with one available node, and two dynamic clusters.
Each dynamic cluster has the minimum number of cluster instances set
to 1. Consider the following isolation configuration scenarios:
- Both of the dynamic clusters are configured with strict isolation.
- One of the dynamic clusters is configured with strict isolation,
and the other dynamic cluster does not have strict isolation defined.
In both of these scenarios, the application placement controller
can place a single cluster instance for one of the dynamic clusters.
Dynamic cluster instances cannot be placed for both dynamic clusters
because only one node is available. In both scenarios, no guarantee
is made about which dynamic cluster starts an instance. Even in the
second scenario, where one of the dynamic clusters has strict isolation
defined, the isolated dynamic cluster does not have priority over
the other dynamic cluster. This restriction can be problematic for
small systems, such as a single node environment, or environments
with a large number of constraints.