Hardware availability

The example ZOS Company data center has components that contribute to hardware availability.

Switches or routers. Two network switches, sometimes referred to as the core network switches, with multiple paths to each other and to the four OSA cards. Each switch would be able to handle the inbound and outbound data on its own to handle scheduled and unscheduled maintenance. The routers will be capable of running the OSPF routing protocol.
OSA cards. Two OSA cards per CPC. These cards are defined and shared by both CPC LPARs. In most organizations there will be many more cards. LPAR1 and LPAR2 share OSA-E1 and OSA-E2 on CPCA. Each LPAR will have a unique IP interface address defined within the TCP/IP stack for OSA-E1 and OSA-E2 cards. The ability for the OSA cards to be shared provides flexibility should a card or switch fail. OSPF also plays a role here, as it can detect link state changes and switch to an alternate path, if required.
CPC. Two mainframe CPCs to allow for any scheduled or unscheduled maintenance. The mainframe processors are very reliable, but some microcode updates require a power-on reset. Some changes, such as a new z/OS upgrade, by implication might involve swapping CPCs while some testing in isolation is carried out.
Coupling Facility. The Coupling Facility is also normally duplicated, but a duplicate is not shown in the example. The CF is used for communication between z/OS LPARs in a sysplex.