The Real IBM i Legacy Is The People
February 7, 2022 Timothy Prickett Morgan
I will never for the life of me understand why the word “legacy” has such a bad connotation in the IT business when in all other aspects of life it means something good. We talk about legacy systems, usually systems of record, a lot here at The Four Hundred, because we are focused on the AS/400 platform and its successors over the past three and a half decades. But the applications – and the people who extend and support them – that run on modern IBM i iron and some vintage predecessor systems have a heritage that extends back even longer.
I recently participated in the webinar going over the 2022 edition of the IBM i Marketplace Survey from HelpSystems, as I have done for many years now, and we are gradually going through the significant amount of data that is presented in this annual survey. But my own thoughts are going to start with something that you don’t see in the downloaded report, but rather the spot survey of the participants in the webinar that asked them how long they have been in the IBM i midrange market.
This survey question was first asked in 2019, if memory serves, in a year when there were two distinct webinars, one for the market in North America and another for customers in Europe and Asia. I averaged the data from both polls together and added my thoughts about it in Are You Experienced? IBM i Users Weigh In. Here a summary chart of the answers on the poll:
At the time, I pointed out that I wished the survey poll had taken into account the full history of the IBM midrange platform, which reaches all the back to 1969 and the System/3 minicomputer that came out of the Rochester, Minnesota, labs and also included the System/32 and the System/34. In this year’s participant poll, the earlier history was taken into account and the System/36 and the System/38 were collapsed down into a single era, which is somewhat valid.
The System/38 previewed the AS/400 architecture and the System/36 developed a much larger installed base, and the AS/400 brought these two together into a single platform that, to a certain extent, annoyed both parties but gave them a much better price/performance curve to ride with RPG II and RPG III applications than the System/36 and the System/38 would have offered along. (On a per unit of capacity of transaction processing performance, the System/38 was much more expensive than even IBM mainframes of the time (about twice as expensive), but the AS/400 was essentially the System/38 with a System/36 emulator and was much less expensive than the IBM mainframes of its time (about half as expensive). That bang for the buck jump is what converted the System/36 based to be AS/400 shops – and kept the System/38 customers in the fold.
Anyway, in this year’s participant poll, HelpSystems added earlier generations and condensed the System/36 and the System/38 eras, so we took the red and yellow data from the 2019 survey report and turned it orange in the 2022 survey report. Here is what the poll data looks like:
The data gathered at the end of 2021 and presented in the 2022 report is very similar to that gathered back in late 2018 and presented in the survey report in early 2019. And as you can see, 55 percent of the people on the call started their IBM i careers in 1987 or earlier, a little more than a quarter of those polled came from the AS/400 era of 1988 through 1999, and a surprisingly skinny part of the base comes from the iSeries era of 2000 through 2007. About 13 percent of those polled actually come from the modern IBM i era, which started in 2008.
That latter statistic is a good one, and important. Clearly, companies did not train a lot of staff or add a lot of staff during the iSeries era, which is also when the dot-com boom had just went bust and the second wave of Internet technologies had taken IT by storm, including Java and microservices and such. But in the modern era, companies have been investing and a whole bunch of newer and younger blood has come to the IBM i platform.
Assuming this data is reflective of the base at large – the poll data is about five times thinner than the survey data, so it is arguable if it is statistically significant – then we can do some interesting analysis on top of this poll. And we did in fact, which is presented in this table:
To make things concrete in our heads, we wanted to first calculate the age ranges of the people in the IBM i base. Assuming that the average start age is 22 – which means after four years of college – then the average age ranges for the date ranges encapsulated in the poll data from HelpSystems is shown in the table above. We also calculated the average “vintage” of the people in this tier, and it is meant to be the midpoint of each era, which are roughly a decade apart, give or take a little.
That age distribution is a bit alarming, but not surprising. And in fact, it looks like a much smaller portion of the IBM i people base is at retirement age or beyond based on the official retirement ages set by the U.S. Social Security Agency’s guidelines as to when you can start retirement. (You are not forced to retire by Uncle Sam, at least not yet.)
Here is when it gets interesting: When we put actual numbers of people on these shares. Then you can feel the true weight of the brain drain that could be coming the way of the IBM i base in the next five to 10 years. Historically, the average IBM i shop has had somewhere north of three programmer/analyst/administrator employees, with some having hundreds of such people and some having one. If you assume this data can be extrapolated across 120,000 IBM i and OS/400 customers, and an average of 3.2 people per site (which is what we have calculated in the past given other survey data over the many decades we have been tracking this base), then there are around 384,000 people in the installed base. Of those, nearly 27,000 people are at retirement age (meaning 67 or older), and this is significant because it took 14 years during the IBM i era to add twice as many people. Assuming everyone has the bare minimum of people they need to maintain existing apps and develop new ones when necessary, then it would take 28 years – nearly three decades – to replace these people at the same rate.
But it doesn’t stop there. If you look at the System/3X part of the pie – the orange slice – then every year or so another 18,000 or so people are going to enter retirement age every year. (One tenth of the System/3X people pool will pass through 67 each year.) So the market will also have to keep replacing these people, too. At the current rate, if this all works out as a steady state flow problem, we will bleed out nearly 27,000 people fairly quickly and then another 18,000 a year after that, and the replacement rate is around 3,600 people per year. Five years from now, the IBM i people base would be around 283,000 people, adding around 18,000 people, but bleeding off around 119,000. And in ten years, as another group of people reach retirement age, the IBM i people base will shrink down to about 209,000 people. For this to all stay in balance, the installed base would have to shrink to about 65,000 sites. Which we do not think will happen in the next decade. With a 2 percent attrition rate – which is high at this point in the history of the IBM midrange – the base would shrink to around 100,000 unique sites. With a 1 percent attrition rate, it goes down to around 110,000 unique sites. The reality is probably in the middle of these two numbers.
Now, not everyone is going to retire at 67, of course. Some people will work much later, and buy the base more time to train people. And we think this coming people shortage will indeed drive a slew of managed services demand for system monitoring and maintenance so companies can share this burden across fewer numbers of experts. Automation will be key here as well, and so will the adoption of DevOps tools to make the IBM i platform attractive to newer generations of programmers and make their programming operations more efficient and create better programs faster.
We shall see how this all plays out, but the vendors we know are all lining up to tackle this problem. There are lots of opportunities, as you can see. We all have a legacy to protect, and we all want to teach our successors. So, let’s get on with it. . . .