Well, this is a nice, simple, straightforward word, made of two simple, easily serviceable bits that have been in English as long as there has been an English: main and frame. The overt sense is “principal supporting structure” or “basic framework”. It was used first in application to heavy equipment – of course, it’s not the frame that does the glamorous part of the work, but nothing at all could happen without it.
But in 1964, two important things happened to mainframe: it gained a new meaning, and a future mainframe alpha-geek – my brother, Reg Harbeck – was born. The mainframe is now known as a kind of computer, a very big kind (these days, about 7 feet tall, 6 feet wide, and 4 feet deep – that used to be small for a computer, but it’s enormous compared to my Mac Mini. My brother, on the other hand, is about 6 feet tall and about a foot and a half wide and deep). I learned in high school that mainframes were big, old, dusty, clunky machines, hoary things that were the computer equivalent of dinosaurs. I also learned a number of other things that turned out to be false, but few so egregiously so – and yet pervasively believed – as that.
Let me quote from The Devil’s IT Dictionary:
n. 1. An obsolete device still used by thousands of obsolete companies serving billions of obsolete customers and making huge obsolete profits for their obsolete shareholders. And this year’s run twice as fast as last year’s. n. 2. A large PC peripheral
Let me put it this way: everyone you know uses a mainframe. Including you. Unless you don’t have an ATM card, a credit card, insurance, a bank account, a pension plan… or pay taxes. Getting out cash? That’s you, personally, operating a mainframe interface. What, you thought they used PCs to do that?
My brother has been so kind as to write a pocket encyclopedia entry for me on the topic (he writes quite a lot on mainframes, so this was an easy whip-off for him):
On April 7, 1964, IBM announced the System/360 mainframe. At the time, there were a number of other kinds of mainframe in use or in development. Mainframe, of course, refers to the “frames” of magnetic core memory which predated modern transistor memory. The cabinet containing these frames of memory led to the term main frame computer.
Over the past 47 years, that IBM mainframe computer has grown to become the one that powers the global economy, while the other types of “mainframe” eventually fell by the wayside after losing successive battles with IBM’s mainframe and then PCs. The IBM mainframe never blue screens or gets viruses or trojan horses. And IPLs (“initial program loads” – the mainframe version of a reboot) may be years apart. In fact, it’s common for mainframe downtime to average in the order of 5 minutes a year, i.e., 99.999% uptime.
Additionally, a given mainframe may be running hundreds – or thousands – of applications concurrently, at 99+% busy with no degradation in performance. That contrasts sharply with all other platforms still in common use, which tend to run no more than one data-intensive application per machine (office-type user-oriented applications are not intensive in comparison), and slow down to a crawl when they get over 30% busy (but average closer to 5% busy).
Interestingly, there are only about 10,000 mainframes in use in the world today, by the largest 4,000 organizations on earth – and there have never been more. The mainframe hasn’t gone away. Instead, it has worked so well that all the other computers can focus their processing power on user-oriented activity such as presentation, allowing those mainframes to take care of the massive data processing that keeps the world economy functioning, with any exceptions to that function arising from political and business, not technical, considerations.
Today’s leading-edge mainframe, IBM’s zEnterprise, is the highest-performing ever, and even has the ability to have other Unix, Linux and Windows computers “bolted on” to it as zBX Blade Centers to run all the critical applications in an entire enterprise, regardless of which platform they were written for. That way, those applications that were written for non-mainframe platforms can still benefit from the massive, secure, high-performing, reliable back-end processing of the mainframe while doing their additional work with that data and processing in a closely coupled and secure configuration.
Yes, Virgina, there is a mainframe, and it keeps the world economy running, and even plays nicely with every other platform out there, almost invisibly keeping the largest organizations on earth running their critical data processing smoothly and reliably.
But mainframes still don’t get no respect. And mainframers are typically somewhat older than the average computer person. This is in no small part because computer programmers tend to learn their craft at institutions that don’t have mainframes, so they’re used to smaller machines and thus tend to lean towards them in their careers and to push their employers towards them. This in spite of the fact that such computers are very inefficient because they were designed for very constrained applications and environments, and building a large application on such a basis is like trying to build a large castle with only ten square metres of land and a lot of cantilevers.
Or, to put it another way yet, consider this quote from The Economist in 1984: “Mainframes, which are used by big businesses for their centralised data processing, are slower than supercomputers (though still very fast).” Reg’s response? “That’s like saying a Bugatti is slower than a drag racer.” Supercomputers do one thing very well. Mainframes do tens of thousands of things very well, and they’ve always been designed to do so.
But they’re like those basic words we have from Old English, the ones that make up the core of our vocabulary. They seem unspectacular, and they don’t get a lot of respect or notice, but they perform a huge number of functions and perform them very efficiently. PC may seem sharp and crisp, and may rhyme internally, but it’s also small – those /i/ vowels are high, and it’s a tiny little initialism. Mainframe is longer, and doesn’t quite rhyme (but very nearly), but it’s warm and friendly and stable, with its easy nasals, and it has a powerful mid-front vowel. Fittingly, it’s not so outgoing (How do you know you’re talking to an extraverted mainframe sysadmin? He looks at your shoes when he’s talking to you), but it does the job.
Mainframes are often called big iron. This puts me in mind of the title of a book by Anthony Burgess: Any Old Iron. The book is about the sword Excalibur, which was embedded in a stone until King Arthur pulled it out. The layman may look at such a thing and think, “Woo-hoo, a big rock with a fancy handle. Next?” But when you put them in action, mainframes cut through many a problem quite efficiently.
Take the tour at Space Center Houston and go on the trolly to the computer building. When I went on a school field trip a gillion years ago that room was slammed full of big machines with huge reel to reel tapes. Now there are a few mainframes doing the same work and more. It was amazing to see when i took my kid on the same tour.
Was it a gillion or just umpteen? 😉
Pingback: Io | Sesquiotica
I never knew there was a younger version of Reg Harbeck, but after reading the name of the blogsite it all made sense….
For those of you who are interested in reading more on Mainframes (or, maybe more appropriate, would like to view a couple of YouTube episodes on “The History Of Big Iron”, go here: http://community.ca.com/blogs/mainframe-voice/archive/2011/08/23/big-iron-mainframe-history-part-1.aspx.
keep up the good work…
Pingback: Lamson | Sesquiotica