Availability, which is the degree to which a system is ready when needed to process data, is key in providing around the clock services. The network--and particularly a network attached to a mainframe--is considered critical, and availability is mandatory for the continuity of business processes. Designers of large networks enhance availability by introducing redundant communication lines, routers, and switches, and implementing server clusters.

Maximizing redundancy has a high price tag and the network designer, together with management, must decide on the risks and impact of an outage. This will determine the availability level that suits the application and the organization. The level of reliability and redundancy introduced in mainframes is in the range of 99.999% (the five 9's), which still leaves unplanned outage of about 5.3 minutes a year. To achieve even higher availability, IBM introduced a clustering technology called Parallel Sysplex.