An architecture is a set of defined terms and rules that are used as instructions to build products. Each generation of mainframe computers has included improvements in architecture, while remaining the most stable, secure, and compatible of all computing platforms.

In computer science, an architecture describes the organizational structure of a system. An architecture can be recursively decomposed into parts that interact through interfaces, relationships that connect parts, and constraints for assembling parts. Parts that interact through interfaces include classes, components, and subsystems.

Starting with the first large machines, which arrived on the scene in the 1960s and became known as "Big Iron" (in contrast to smaller departmental systems), each new generation of mainframe computers has included improvements in one or more of the following areas of the architecture:

• More and faster processors
• More physical memory and greater memory addressing capability
• Dynamic capabilities for upgrading both hardware and software
• Increased automation of hardware error checking and recovery
• Enhanced devices for input/output (I/O) and more and faster paths (channels) between I/O devices and processors
• More sophisticated I/O attachments, such as LAN adapters with extensive inboard processing
• A greater ability to divide the resources of one machine into multiple, logically independent and isolated systems, each running its own operating system
• Advanced clustering technologies, such as Parallel Sysplex®, and the ability to share data among multiple systems.

Despite the continual change, mainframe computers remain the most stable, secure, and compatible of all computing platforms. The latest models can handle the most advanced and demanding customer workloads, yet continue to run applications that were written in the 1970s or earlier.

How can a technology change so much, yet remain so stable? It can by evolving to meet new challenges. In the early 1990s, the client/server model of computing, with its distributed nodes of less powerful computers, emerged to challenge the dominance of mainframe computers. Industry pundits predicted a swift end for the mainframe computer and called it a "dinosaur." In response, mainframe designers did what they have always done when confronted with changing times and a growing list of user requirements: They designed new mainframe computers to meet the demand. With a tip of the hat to the dinosaur naysayers, IBM®, as the leading manufacturer of mainframe computers, code-named its then-current machine T-Rex.

With the expanded functions and added tiers of data processing capabilities such as Web-serving, autonomics, disaster recovery, and grid computing, the mainframe computer is poised to ride the next wave of growth in the IT industry. Mainframe manufacturers such as IBM are once again reporting annual sales growth in the double digits.

And the evolution continues. While the mainframe computer has retained its traditional, central role in the IT organization, that role is now defined to include being the primary hub in the largest distributed networks. In fact, the Internet itself is based largely on numerous, interconnected mainframe computers serving as major hubs and routers.

As the image of the mainframe computer continues to evolve, you might ask: Is the mainframe computer a self-contained computing environment, or is it one part of the puzzle in distributed computing? The answer is that The New Mainframe is both— a self-contained processing center, powerful enough to process the largest and most diverse workloads in one secure "footprint", and one that is just as effective when implemented as the primary server in a corporation's distributed server farm. In effect, the mainframe computer is the definitive server in the client/server model of computing.