Both server and NLB clusters support a range of applications, although each one, as discussed, handles them differently. This section concentrates on setting up clusters of the more popular types of applications.
First we will take a look at the applications that are not cluster aware. It is difficult to provide general rules for clustering such applications, because each needs to be analyzed separately and any decision needs to be based on a thorough review of its behavior and use of system resources. Typically, however, clustering can be accomplished by using Generic Application or Generic Script resources (assuming that clustering is possible at all).
A large number of applications depend on transactional and messaging features provided by the Windows operating systems. Microsoft developed clustered resources for two components offering these essential services: Distributed Transaction Coordinator and Message Queuing. They can serve both cluster non-aware and cluster-aware (such as SQL Server 2005) applications. Two subsections are dedicated to the description of the clustering procedures for these components.
High availability of client applications can be accomplished by installing them on a Terminal Server farm. This approach involves additional considerations, such as keeping track of the user sessions across the farm. The procedure for setting up such an environment and the way to maintain a session state is described in the next subsection.
Fortunately, the number of cluster-aware applications is growing. Microsoft SQL Server, which is covered in the following subsection, serves as one of the best examples.
This section concludes with a description of clustering procedures for a highly available Web farm.
Was this article helpful?