Written by Cameron Seay, Open Mainframe Project Governing Board member, co-chair of the COBOL Working Group and Adjunct Professor at East Carolina University
This content was originally published on the ASG Technologies’ blog.
What happened over time that caused US colleges and universities to stop teaching mainframe skills?
In the 1980s and 90s with the advent of personal computers there arose the misconception that you could replace the functionality of the mainframe by assigning enough commodity (cheap) processors to do the same work mainframes did. This reasoning bespoke a fundamental misunderstanding of how the mainframe worked and ignored the fact that it was input/output (I/O) that allowed the mainframe to process then tens of thousands, now hundreds of thousands, transactions per second and not just the number or even the speed of the processors. In addition, it ignored the fact that commodity (Intel, AMD) processors are much less efficient than a mainframe processor. Hence just throwing a bunch of processors at a workload does not mean you can do away with the mainframe. Transactions need to be processed in sequence, on one thread, so trying to parallel process transaction was not a solution. You need a mainframe.
But this misconception became societal in nature: it was the beginning of the myth that the mainframe is “old” technology. Stewart Alsop, a technical writer in the early 1990s, famously predicted that “the last mainframe would be unplugged on March 15, 1996.” Of course, when this date came and went with mainframes still in operation all over the world, Alsop was forced to amend his statement. Nonetheless, academic departments began to remove mainframe-related courses from their curricula in the late 1990s in anticipation of the disappearance of the mainframe. So much so that by the early 2000s mainframe technology had all but disappeared from the academic landscape in the US. As of today, very few colleges in the US have 8, 12, or 16-week courses in mainframe. In the 17 campus UNC System, only 2 campuses teach mainframe, and it is dying on one of those. COBOL, a language essential to the global economy, is virtually non-existent on US campuses, even though it remains vitally important to the domestic and global economy.
What was the impact on companies and students?
The US Dept of Education says there are over 4,300 degree granting institutions in the US. I know of no more that 10 or 15 schools that have semester long mainframe courses of any kind. There may be more, but I am not aware of them. This situation is attributable to clearly identifiable reasons. Given the nature of higher education orthodoxy, the 4-year schools have no compelling reason to teach mainframe and several reasons not to. Among the reasons not to teach it is lack of faculty expertise in the technology. This comes from the fact that mainframe expertise is, for the most part, acquired via non-academic paths. By this I mean that mainframe practitioners usually obtain their expertise on the job, and usually their first exposure to the technology is on the job, not in the classroom. Many of them either do not have a degree or have a degree in an unrelated area. Higher education is focused on newly minted Ph.D. pursuing tenure by doing research and publishing in their discipline. Because mainframe has not been a big part of Academia for a long time, few dissertations have been written related to it, and subsequently fewer and fewer academics have expertise in that area. So the departments really have no incentive to teach it.
Where are we now?
The dependence on mainframe is unabated. The advent of the cloud has created a dramatic reduction in the amount of on premises hardware, but the need for mainframe has not diminished. To be sure, many workloads have been moved off the platform, but the core applications remain because there is really no other technology that can process 500,000 transactions a second- a level of performance needed by credit card companies, large banks and large retailers. With the mainframe workforce steadily aging and retiring, we are now at a critical juncture where companies will have to come up with innovative onboarding models to maintain a mainframe workforce.
I am not seeing a marked increase in the number of schools teaching mainframe, but I am seeing considerable effort on the part of IBM and other companies to come up with innovative ways to bring newcomers to the platform. For the past 12 years IBM has hosted the Master the Mainframe Contest (MtM). MtM is a series of challenges at 3 different levels of increasing difficulty. It is both a way for a novice to be introduced to the platform and for IBM to identify potential superstars on that platform. I have had my students participate each year but the last two. The content of MtM is excellent, but the problem is that in the US you are only going to attract mostly affluent students by having students participate in a time consuming activity (which MtM is) that is unrelated to their degree. Most students have to work, so the time they can devote to MtM is limited. I have advocated for years to include MtM as part of an elective class, and some professors do this, but very few.
It has been my contention for a long time that mainframe belongs in the college curriculum. There are two places in academia where mainframe is an almost perfect fit: Community Colleges and the Historically Black Colleges and Universities (HBCUs). In a future post I will explain, in detail, why this should be done, and how it can be. Stay tuned.