In this chapter, we'll combine the Heartbeat package with the LVS software to build a high-availability LVS-DR cluster with no single point of failure that will serve as the foundation for our enterprise-class cluster. (We'll add a filesystem, a method of scheduling batch jobs, and application monitoring software in later chapters.)
In order to build a high-availability cluster we need to use redundant LVS Directors, and we must be able to automatically remove real servers (cluster nodes) when they fail. Before we discuss all of the design goals of a high-availability cluster, we'll examine these two basic requirements.
Note | See the Keepalived project for another approach to building a high-availability cluster. |