Increasing Server Sophistication, Availability, and Reliability

As servers began to assume more and more mission-critical information processing and handling roles in businesses, hardware manufacturers began to develop various ways to ensure maximum uptime and thereby increase server availability and reliability. This began at the I/O level, with the introduction of drive duplexing, where dual controller-disk drive chains could be mirrored so that a person could keep working even in the event of a drive or controller failure in either chain.

By the mid-1990s, servers with dual redundant power supplies were available, and companies such as Novell, Microsoft, and Vinca offered a variety of clever technology solutions to permit pairs or groups of servers to function in lock-step, so that if one server failed, the other server(s) would keep running, and services would continue unabated (though sometimes slowed down, especially when architectures with all components running nominally could distribute workloads and thereby deliver performance gains).

Machine clustering options extended this capability to groups of servers. By the year 2000, Microsoft and Novell offered various server clustering extensions to their network operating systems, and companies such as Hewlett-Packard, IBM, and Sun Microsystems also offered various hardware products with necessary software included to make turnkey clustering a reality. Today, it's not at all uncommon for clustered machines to serve high-traffic websites, transaction processing systems, and other high-demand, high-traffic applications. In some cases, elements of the same cluster may even be widely geographically dispersed so that a disaster or failure of an entire site does not compromise server availability.