The Mainframe Lives and Thrives – Part 1

There is a story that must be told since it uncovers an injustice visited upon a piece of information technology (IT) equipment called a mainframe computer. These, like many such situations displaying bigotry and ignorance, are put forward by people who do not understand the butt of their criticism.

There is a word for this type of person’s traits: ‘Ultracrepidarianism is the habit of giving opinions and advice on matters which one knows nothing about’; The word first appeared as an insult in an 1819 letter by essayist William Hazlitt. [from Howtogeek].

I decided that, since few outside IBM will would take up swords on the mainframe’s behalf, I would do it to the best of my ability. I did spend nearly a quarter of a century with IBM, much of it with customers who had mainframes.

I should say at this juncture that I have no connections with IBM except those of a normal retiree.

Short History of the Mainframe

In the 1960s, when IBM was in its prime, there was no definition of a ‘mainframe’ except it was a term applied to larger machines, usually one, which served large companies and were accessed using ‘dumb’ terminals. The latter’s claim to intelligence was a data buffer which was built up to completion before being sent to the computer for processing, plus a few other built-in communication functions. The term mainframe eventually became synonymous with large IBM systems, sometimes called big iron, big ‘mippers’ and big burners.

The layout of earlier mainframes is shown in part in Figure 2, where disk and tape control units are not shown, yielding a space occupation of about half a tennis court

Very dumb terminals, like the old DEC VTxx did not buffer keyboard input and when a character was typed, it went to the computer and back before it appeared on the user’s screen; the ‘feature’ called echoplexing. If the communications line was busy, character appeared on the screen ages after typing by which time the user had typed 10 more before seeing anything!

IBM Mainframe Birth (1964)

Prior to this significant event, IBM (and others) sold a variety of systems, many aimed at a certain computing need and one type cold not run programmes from another type or communicate effectively. A company might own a computer for commercial work and another for scientific work, perhaps more than one of each.

In April 1964 IBM announced the System/360 range of compatible computers, so-called because it covered both scientific and commercial computing, that is, across 360O of the compute spectrum. IBM effectively ‘bet its business’ on the success of the S/360 after spending a borrowed $5 bn. on its development. The rest is history.

The architecture of the /360, not the technology, was carried forward with enhancements through succeeding mainframe generations and it was IBM’s proud boast that programmes written for /360 would run on its successors. These included the S/370 (1970), S/390, 308x, 3090 and ES/9000 ranges, culminating in today’s (2020) IBM Z System.

The smaller /370 systems had 1401 (an earlier IBM ‘mainframe’) compatibility, using its microcode to translate compiled 1401 instructions into S/370 architecture instructions for seamless execution. Thus, compatibility went back further than 1964.

Mainframe Family Tree

View here

There were, as in all products, imitators, one of which was the Amdahl systems which were IBM ‘look-a-likes’, developed by Gene Amdahl, an ex-IBM /360 mainframe developer. There were others, like the Japanese equivalents, which emerged to compete with IBM. These were the mainframe wars, a precursor to the UNIX wars a decade or so later. Many smaller systems came and went, for example, Itel (sic), Sequent, Prime and DEC VAX.

System 360; From Computers to Computer Systems

View here

Mainframe Memories

I remember much of the post-/360 era with nostalgia, not just for the technology but for the team spirit and humour.

1. I remember going into customer site one day and hearing a fairground-like rendition of ‘She’ll be coming round the mountains ….’. Odd, I thought and odder it seemed when I found the sound was being generated by the computer’s printer, driven by a special programme written by the IBM engineers. I have since been told of other musical computer machines, even calculators.
2. I configured the microcode for a dual printer setup for a public utility one year and rang in to ask how they were performing. They told me they were functioning but printing alternate lines of a utility bill on a different printer!
3. This notice, often found attached to mainframes is one abiding memory of them.

Modern Mainframe Myths

I meet and read about non-computer people (the IT-unwashed) who think the world’s business is run on PCs and PC-like computers (x86). Not only that, they believe there is such a thing as a computer expert, who is a person who has learned to code or an eight-year old who is a whizz on an iPhone or PC. This is the apex of disillusionment. The workplace information technology world is bigger than ‘small pieces of tin’ and fancy graphics.

It is therefore understandable that they will never conceive that the mainframe runs a significant part of their lives, since their misunderstanding is perpetuated by current computer science (CS) courses, the media and some politicians. This misunderstanding will be corrected in this paper; whether the non-believers will recant is open to question. The facts I present will, however remain and, as Aldous Huxley noted:

‘Facts do not cease to exist because they are ignored.’

The mainframe not only lives, it is the most reliable and cost-effective way of running large scale commercial computing.

Mainframe Issues

The mainframe does have one serious drawback; not in itself but in the lack of people to develop on and support it. When I joined IBM, full training was given to myself and others; 19 weeks of computer training, one week of industry training and an edict to continue training for 22 days in each year. Today, I am told that people joining a computer company are given a company induction, a day or two about their computing and sent out into the field. Many of them are cut down when taken out of their narrow sphere of knowledge.

To cut a long story short, there is a dearth of mainframe education and training, though IBM does some good work via, for example, its Master the Mainframe initiative. The other downside is the rapid rate of retirement of old mainframe people, creating a double whammy for the mainframe support.

Though IBM no longer does its own client training (with some exceptions), they have partnered with training companies (such as ProTech Training (www.protechtraining.com) to provide both IBM licensed training as well as independently created mainframe courses.

Incidentally, there is also a lack of training on important computing techniques such as Project Definition, Delphi Technique, SWOTs analysis and a dozen other methods of developing IT projects. Project management is quite well covered but is useless if the initial project specification, as developed by these or other techniques, is flawed.

Anatomy of the Mainframe

What follows are a number of myth-busting facts about the mainframe ecosphere, culled from various sources too numerous to list here. If threatened with jail in the Tower of London, I can retrace my research steps and reproduce them.

The IT-unwashed View of Mainframes:

• They are old, along with their software and peripherals and are deemed legacy or heritage, along with the pyramids and Stonehenge
• They only support old, dumb terminals for access to systems
• They are totally inflexible and cocooned from modern technology
• They are cloud unfriendly, a bad trait today
• They are expensive to purchase run, slow to develop on, lack development agility and, in a word, use ‘old hat’ technology
• Their use in the workplace is dwindling fast and has been for years
• They don’t support Windows or Linux
• Nobody wants to work on them

• Mainframe jobs pay peanuts compared with modern ‘trendy’ jobs
• All work is moving to the cloud environment and they run anything but mainframes.
• In essence, it should be consigned to the dustbin of IT history and a model preserved in a museum somewhere, along with wax models of the old geeks who worked on them (see Figure 3).

The Intelligent View of Mainframes

These are myths, gathered by me from a Google search on ‘mainframe myths’ but the links are too numerous to list here. Below, are the facts about this ‘electronic corpse’.

Today, mainframes run on, support or have:

• It is the most reliable and resilient piece of computer hardware on the planet, complemented by its software; Z Systems and LinuxONE hardware and an array of software between them are second to none in providing high availability and resilient computing. In fact, “Z” in Z Systems stands for zero down time! Expected system availability is 99.999% (excluding finger trouble etc.).
• The above is necessary when you look at the list of major companies which rely on the mainframe for their business
• 87% of all credit card transactions
• 30 billion transactions per day
• Nearly $8 trillion per year
• 29 billion ATM transactions per year (95% of them)
• 92 of the top 100 banks
• 23 of the 25 top airlines
• 23 out of 25 top US retailers
• 10 of the world’s top 10 insurers, 84% of Top 25 Insurance Organizations
• 71 percent of Fortune 500 companies (355 out of 500)
• More transactions per day than Google searches (1.3 million/second on CICS transaction processing system vs. 68,542/second on Google)
• 55% of all enterprise transactions
• 80% of the world’s corporate data originates on them
• 72% of firms noted their customer-facing applications are completely or very reliant on mainframe processing
• Handle 68 % of the world’s production IT workloads, yet they account for only 6 % of IT costs.
• IBM’s Z series mainframe sales are up 70% year-over-year (2018 on)
• Mainframe workloads are increasing; 57% to 64% mainframe usage (2017).
• 90% of all mainframe applications will be running in 2023
• For larger workloads, the mainframe is the most cost-effective system
• They are used today by cloud vendors in their offerings
• Mobile users each perform about 37 transactions daily and 91% of their apps communicate via the mainframe.
• They have the extra ‘muscle’ -equal to 150o or more x86 systems – to handle increasing volumes of remote devices without resorting to multiple servers with their attendant costs
• They still have an emphasis on RAS (reliability, availability, serviceability) for mission critical workloads
• It supports Linux OS and modern programming languages such as Python, Java, JavaScript, and C++.
• The data centre lives too and rumours of its demise in favour of the cloud are somewhat exaggerated. See the section Career Aspects & Prospects later on the dangers of throwing the baby out with the bathwater.

Many data-center workloads staying on premises, Uptime Institute finds.

Still not dead: The mainframe hangs on, sustained by Linux and hybrid cloud.

‘Don’t automatically dismiss IBM Z when making strategic decisions about your current and future IT environment. It continues to have significant market and competitive strength, and IBM makes sure it works with the latest technologies, including cloud computing and open-source tools. It’s a key innovation player that drives better ROI on core infrastructure while supporting the latest advancements in technology and user experience.’

The Rockstar in Your Data Center: The IBM Z Mainframe.

‘Although the global cloud computing market size is expected to grow from USD 371.4 billion in 2020 to USD 832.1 billion by 2025, at a compound annual growth rate of 17.5%, on-premise servers still operate in 98% of businesses.

Industry “started with mainframes, then we went to standalone servers, and then cloud was a big thing,” said Peter Tsai, a senior technology analyst at Spiceworks, an IT community forum. “Now it seems like we’re pulling computing resources back, closer to where they’re needed.” ‘

6 Reasons Why Internal Data Centers Won’t Disappear

View here

• You don’t need the old, dumb 3270 screens to access it
• The mainframe today can take part in all the ‘technology races’ and is not simply a lonely COBOL silo, surviving on memories of the ‘great days’
• Despite all this, few, if any, computer science courses even acknowledge the existence of the mainframe, let alone teach anything about it.

“Why don’t you care about such a thing [mainframes]? Because you’ve been taught not to. Schools teach you that mainframes don’t matter, if they are mentioned at all. Well guess what! Not only do they matter, everything you do, your family does, your government does, relies on them.” [quotation from a mainframe article 2015]

Next time: Let’s Boot the Mainframe Out??