The mainframe enjoys the reputation as the most reliable server hardware on the market. One of your reasons for considering Linux on the mainframe might be the outlook of deploying highly available Linux application servers.
In contrast to what many people believe, high availability is not a matter of all or nothing. High availability is driven by trade-offs between business needs and costs. Before working towards high availability, it is important to have a clear idea of what exactly is required. For example, high-availability solutions are often not meant to cover disaster scenarios. In this chapter we examine some availability requirements in relation to what we mean by high availability.
This chapter builds on the other chapters in this part of the book. It describes configuration considerations that might help you to improve the availability characteristics of your Linux servers. We look at various configuration options of the zSeries hardware and the z/VM virtualization that can decrease the probability of an outage and speed up the recovery of a failed Linux image. The corresponding management tasks for keeping an already-running Linux server available are covered in the next part of the book.
We use examples to show how the methods that are available today allow some of the Linux on the mainframe server farms to be configured for specific application workloads, such that any likely failures would not affect the users' access to the services.
There is a rapid evolution of tools for creating and managing highly available Linux servers. In the last topic of this chapter we point to some promising developments that are in under way as we write this book.
We will explore questions such as: