What is a mainframe? It's a style of computing

Learn the latest in IBM mainframe technology:

Although the term mainframe first described the physical characteristics of early systems, today it can best be used to describe a style of operation, applications, and operating system facilities.

Today, computer manufacturers don't always use the term mainframe to refer to mainframe computers. Instead, most have taken to calling any commercial-use computer— large or small— a server, with the mainframe simply being the largest type of server in use today. IBM®, for example, refers to its latest mainframe as the IBM System z9® server. We use the term mainframe in this section to mean computers that can support thousands of applications and input/output devices to simultaneously serve thousands of users.

Servers are proliferating. A business might have a large server collection that includes transaction servers, database servers, e-mail servers and Web servers. Very large collections of servers are sometimes called server farms (in fact, some data centers cover areas measured in acres). The hardware required to perform a server function can range from little more than a cluster of rack-mounted personal computers to the most powerful mainframes manufactured today.

A mainframe is the central data repository, or hub, in a corporation's data processing center, linked to users through less powerful devices such as workstations or terminals. The presence of a mainframe often implies a centralized form of computing, as opposed to a distributed form of computing. Centralizing the data in a single mainframe repository saves customers from having to manage updates to more than one copy of their business data, which increases the likelihood that the data is current.

The distinction between centralized and distributed computing, however, is rapidly blurring as smaller machines continue to gain in processing power and mainframes become ever more flexible and multipurpose. Market pressures require that today's businesses continually reevaluate their IT strategies to find better ways of supporting a changing marketplace. As a result, mainframes are now frequently used in combination with networks of smaller servers in a multitude of configurations. The ability to dynamically reconfigure a mainframe's hardware and software resources (such as processors, memory, and device connections), while applications continue running, further underscores the flexible, evolving nature of the modern mainframe.

While mainframe hardware has become harder to pigeon-hole, so, too, have the operating systems that run on mainframes. Years ago, in fact, the terms defined each other: a mainframe was any hardware system that ran a major IBM operating system. This meaning has been blurred in recent years because these operating systems can be run on very small systems.

Computer manufacturers and IT professionals often use the term platform to refer to the hardware and software that are associated with a particular computer architecture. For example, a mainframe computer and its operating system (and their predecessors) are considered a platform; UNIX® on a Reduced Instruction Set Computer (RISC) system is considered a platform somewhat independently of exactly which RISC machine is involved; personal computers can be seen as several different platforms, depending on which operating system is being used.

So, let's return to our question now: "What is a mainframe?" Today, the term mainframe can best be used to describe a style of operation, applications, and operating system facilities. To start with a working definition, a mainframe is what businesses use to host the commercial databases, transaction servers, and applications that require a greater degree of security and availability than is commonly found on smaller-scale machines.

Early mainframe systems were housed in enormous, room-sized metal boxes or frames, which is probably how the term mainframe originated. The early mainframe required large amounts of electrical power and air-conditioning, and the room was filled mainly with I/O devices. Also, a typical customer site had several mainframes installed, with most of the I/O devices connected to all of the mainframes. During their largest period, in terms of physical size, a typical mainframe occupied 2,000 to 10,000 square feet (600 to 3000 square meters). Some installations were even larger than this.

Starting around 1990, mainframe processors and most of their I/O devices became physically smaller, while their functionality and capacity continued to grow. Mainframe systems today are much smaller than earlier systems— about the size of a large refrigerator.

In some cases, it is now possible to run a mainframe operating system on a PC that emulates a mainframe. Such emulators are useful for developing and testing business applications before moving them to a mainframe production system.

Clearly, the term mainframe has expanded beyond merely describing the physical characteristics of a system. Instead, the word typically applies to some combination of the following attributes:

Compatibility with mainframe operating systems, applications, and data.
Centralized control of resources.
Hardware and operating systems that can share access to disk drives with other systems, with automatic locking and protection against destructive simultaneous use of disk data.
A style of operation, often involving dedicated operations staff who use detailed operations procedure books and highly organized procedures for backups, recovery, training, and disaster recovery at an alternative location.
Hardware and operating systems that routinely work with hundreds or thousands of simultaneous I/O operations.
Clustering technologies that allow the customer to operate multiple copies of the operating system as a single system. This configuration, known as Parallel Sysplex®, is analogous in concept to a UNIX cluster, but allows systems to be added or removed as needed, while applications continue to run. This flexibility allows mainframe customers to introduce new applications, or discontinue the use of existing applications, in response to changes in business activity.
Additional data and resource sharing capabilities. In a Parallel Sysplex, for example, it is possible for users across multiple systems to access the same databases concurrently, with database access controlled at the record level.

As the performance and cost of such hardware resources as central processing unit (CPU) power and external storage media improve, and the number and types of devices that can be attached to the CPU increase, the operating system software can more fully take advantage of the improved hardware. Also, continuing improvements in software functionality help drive the development of each new generation of hardware systems.