Mainframes have one of the longest histories of any kind of computing technology that is still used today. In fact, mainframe history is far too long to pack into a single blog post. But I’m going to try anyway. Keep reading for a (very) brief history of mainframe computing.
Before diving in, let’s take a moment to appreciate just how long mainframes have been around.
Very few of the computer scientists who worked on the first mainframes are still working with mainframes today. That makes mainframes different than virtually all other computing technologies that we use today.
For example, plenty of people who work as Windows and Linux server admins today can remember when Windows came out in the 1980s, or when System V was the hottest new technology in the Unix world. Most people who support cell phones remember when the first cell phones appeared. Data scientists who work with Hadoop or Spark can certainly remember when those platforms came out; they’re still quite new compared to mainframes.
But few of the people who work with mainframes today can recall when the first mainframes came out.
Milestones in Mainframe History
That’s why it’s worth surveying major milestones in the development of mainframes over the years. They give you valuable perspective on just how significantly mainframes have evolved to become the lean, mean computing workhorses that they are today.
Major developments in mainframe history include:
By most measures, the first mainframe computer was the Harvard Mark I. Developed starting in the 1930s, the machine was not ready for use until 1943. It weighed five tons, filled an entire room and cost about $200,000 to build – which is something like twenty-eight million in 2017 money. (Keep that in mind the next time you complain about the cost of your iPhone.)
World War II sparked the creation of the next famous mainframe computer, the ENIAC (Electronic Numerical Integrator and Computer). Although the ENIAC was not actually completed until a year after the war ended, it signaled the start of heavy government investment in mainframe development. The ENIAC remained in service for a decade.
While early mainframes were based on vacuum tubes for storing data, a major innovation came to the mainframe world with the development of what was called core memory. In place of vacuum tubes, core memory stores information magnetically. Your magnetic 5400-RPM hard disk it was not, but core memory provided a faster and more reliable way to store and retrieve data than vacuum tubes. Core memory was first used in 1953 and soon replaced vacuum tubes entirely.
COBOL, the programming language most closely associated with mainframes, debuted all the way back in 1959 and remains in widespread use today. That makes COBOL one of the oldest continuously used programming languages. It beats even C, which originated in the early 1970s.
It’s worth noting that some of the companies involved in early mainframe development. IBM is the name most closely associated with mainframes. But historically, the mainframe commercial ecosystem was more diverse. More than half-dozen companies – including Univac, General Electric, and RCA – also sold mainframes during the first few decades of mainframe computing.
Linux for mainframes
Also worth noting is that, while mainframes for the first decades of their history ran on special mainframe operating systems, by the late 1990s this changed. Starting in 1998, IBM began developing a Linux-based operating system that could run on mainframes in place of mainframe-native systems.
Today, mainframes are much smaller and more agile than they were decades ago.
IBM mainframes are now about the size of a refrigerator. Modern mainframes don’t weigh more than a (very heavily packed) car.
They cost a lot less, too. Although IBM mainframes still had price tags in the million-dollar range in the early 2000s, today you can pick one up for closer to $100,000.
Over the past half-century of mainframe history, the machines have evolved rapidly and remarkably. Modern mainframes are hardly the huge, crazily expensive, unwieldy machines of yore. They readily integrate legacy with modern technologies, allowing you to do things like run COBOL apps on z/OS alongside Docker containers on Linux (using z/VM) on the same physical machine.
And they remain powerhouses for number crunching and data analytics; in fact, IBM’s latest mainframes boast the most powerful processors in the world.
Want to see examples of just how useful mainframes remain today? Check out Syncsort’s latest eBook Mainframe and Machine Learning for IT Service Intelligence.