Serviceguard is a tool that forms part of a complete High Availability solution. Serviceguard does not offer any form of fault tolerance. Serviceguard was designed to minimize the downtime involved in having applications processes run on a different node. If a critical resource fails, Serviceguard can automate the process of having applications run on an adoptive node. This will involve a degree of downtime for the application while the necessary checks are made in order to start up the application on another node. This is the main topic of discussion in the next chapter.
In this chapter, we looked at a simple two-node cluster. Large clusters (up to 16 nodes) offer greater flexibility when dealing with a failure. As we see in upcoming chapters, Serviceguard can be configured with a measure of intelligence when moving applications to adoptive nodes ; Serviceguard will choose the adoptive node that is currently running the smallest number of applications.
Serviceguard is available for HP-UX as well as Linux (where the concept of Quorum Servers was born). The configuration of Serviceguard on Linux is the same as on HP-UX, offering seamless migration between the two platforms.
Our cluster is currently not monitoring the status of any applications. This is a feature of Serviceguard that most administrators will want to investigate. This is the topic we discuss next.
