100 Introduction to the New Mainframe: Large-Scale Commercial Computing
The Coupling Facility is central to the operation of Parallel Sysplex. It is an LPAR
or standalone CPC with a specialized operating system, called the Coupling
Facility Control Code (CFCC), which is a Licensed Internal Code (LIC). The CF is
connected via high-speed links to all systems, and it acts as a shared data
repository for all members of the sysplex.
The Sysplex Timer is a hardware device that is connected to all members of the
sysplex. It provides a common time source for synchronizing time among the
members of a sysplex.
How Parallel Sysplex works
If an LPAR fails in a Parallel Sysplex, the other systems can take over the
workload of the failing system using a component called WorkLoad Manager
(WLM), which manages work running on all the systems in a sysplex. WLM acts
as a load balancer, routing work to whichever system in the sysplex can best
handle the new workload. WLM allows the systems in a sysplex to work together
as though they were a single system.
Using CBU, the hardware on the running systems in a sysplex can be
dynamically and nondisruptively upgraded with more processor power.
The System Logger is a facility that merges the information from logs on
individual systems in the sysplex, and from individual subsystems operating on
different systems in the sysplex, into coherent structures that represent a
sysplex-wide view of the contents of each log.
For more information about Parallel Sysplex, refer to 3.3.3, “Parallel Sysplex” on
5.5 z/OS elements for availability
With IBM mainframes, the availability design-point focuses on the application.
Why? Because applications do not simply rely on server hardware. Instead, they
require an integrated environment in which hardware, firmware, operating
systems and middleware work together to provide application and data
Approximately one-third of the z/OS code base provides rich RAS functionality
delivering reliability, availability and serviceability–often resulting in outage
events being completely masked from applications—and in severe cases,
resulting in graceful degradation rather than complete failure. And concurrent
maintenance capabilities, supported by both the hardware as well as operating
systems, help to mask planned outage events from the application as well.