Modeling Clustering
The modeling of clustering is based on the same principle proposed for the virtualization. A cluster is a group of servers that share common resources to run an expected application faster than each individual server is able to do. For the end-user, there is only one server and only one application he wants to use.
In this context, the modeling approach is the following:
1. Create all the physical servers that are grouped into a cluster. Each server inherits from the Server artifact.
2. Create the cluster as an artifact inheriting from the Server artifact.
3. Include all the primary physical servers in the cluster as Artifact Components.
4. Add the applications hosted by the cluster that are available for the end user.
5. Add the potential applications of each individual server that are useful for the clustering.
6. Create the links between the physical servers that explain what the communication means are and what is exchanged in the context of the cluster.
Figure 7: Example of Clustering Model.
In the example above, the cluster server XYZ hosts the application Invoice from the end-user perspective. Depending on the current load, the application is balanced either to Server A or Server B but this is totally transparent for the end-user. At least the model explains the architecture and then it is possible to deduce that the application Invoice B can be impacted by a failure of the physical server if the architecture describes some communications channel between the application and the servers.
