10.1 Total system design approach to high availability

Window NT Clustering software is only one component of a data processing system designed for achieving high availability. In order to design the best possible high-availability environment for your data center you must take a "total system" design approach to achieve your goal. By that we mean to achieve the highest level of availability you must analyze the effect that every possible component or subsystem has on total system availability. In fact, even if you never install a cluster, you should still consider the following list of requirements for your data processing system. The following list is really made up of "no-brainer" recommendations for system administrators. These recommendations set the foundation for your cluster design. They are suggestions that, as an administrator, you should consider adopting and including in your standard operating procedures. You would be surprised by the number of computer systems we see that are not protected by these common-sense procedures and recommendations.

Figure 10.1 represents the challenge that you face when trying to achieve high availability for your data processing system. You will notice that only one of these items in this list is a software solution. The rest of them are tried-and-true procedures that have been used for years in the minicomputer and mainframe world just the things that you would expect to be implemented in a large corporate data center. The game plan here is to eliminate all single points of failure if possible. It should not be...