Choose a Cluster Model
The next step is to determine the cluster model you choose to use. If you decide to use a load-balancing cluster for your application, then you can skip this next step, for the cluster model only applies to failover clustering. There are three different failover cluster models:
Single-node failover cluster
Dedicated secondary node failover cluster
Distributed failover cluster
A single-node failover cluster is one server configured as the first node within a failover cluster. Even though the server is configured as if it were within a two-node cluster, there is actually only one server in use. This model is ideal for organizations that wish to build a cluster, but have a limited budget or need to produce a cluster application-testing environment without purchasing a second server. If you are building a test application or your budget is limited, you may choose the single-node failover cluster for your system.
The dedicated secondary node failover cluster is for organizations that wish to provide only one fault-tolerant application on their cluster. This is an ideal cluster model for exchange servers, SQL servers, file shares, and other types of applications that provide dynamic content. A dedicated secondary node failover cluster can be constructed using two to four servers, but only serves one application.
The distributed failover cluster is similar to the dedicated secondary node failover cluster model, but goes a step further by simultaneously providing two or more applications. The two or more applications can run on a single server, utilizing the other servers within the cluster for failover, or each application can run on an individual server within the cluster. For instance, if you wanted to run Exchange Server, SQL Server, a print server, and a file server all within the same four node cluster, you could have one node in the cluster be responsible for each application and then provide failover for the other applications as needed. You could also choose to run multiple instances of the same server within the failover cluster to distribute the load. If you have multiple applications, or multiple instances of the same application, you want to choose the distributed failover cluster so that your applications can be load-balanced between the two (or more) nodes in the cluster.