7.4.1 Environment configuration

Figure 7-4 shows a high level representation of a hypothetical customer configuration that contains two System p servers (Server 1 and Server 2) in separate data centers.

Figure 7-4 LPAR distribution on two servers (each in a separate data center)

Server 1 contains the production environment, with four large LPARs. Server 2, located in a remote data center, contains a preproduction environment with LPARs for the appservers as large as in production, but with a smaller LPAR for the deployment manager. In addition, Server 2 hosts four test environments, each on its own LPAR, which are used by developers to unit test their code.

This customer has decided that it cannot afford to have a "full DR environment" standing idle in its second data center. In the event of a true disaster, where the first data center is wiped out, the customer would be left frantically reconfiguring hardware and software in order to somehow get back up and running.

For simplicity, the applications running on this particular environment are assumed to be stateless. The configuration described here is a small subset used only to demonstrate our example.

Note: A real customer configuration would probably have more environments, more physical servers for redundancy purposes, and so on. A real disaster recovery configuration would also have to take in to account more than just the WebSphere infrastructure; the entire wider environment would need to be replicated correctly. This could also be achieved by using System p LPARs and techniques similar to those described here.