Still A Community Of Common Interest
July 7, 2014 Timothy Prickett Morgan
Whenever we come upon the anniversary of the launch of the AS/400, which passed on a week when The Four Hundred was off on hiatus, most of us are inclined to contemplate and celebrate the genius of the system and the adventurous techies inside of IBM who took a lot of cool ideas about how a system should be designed and created a platform that made it possible for midrange businesses to do sophisticated things without having a lot of propellerheads on their staff to make it all work. This is a very top-down view of the AS/400 and its progeny, and it is natural enough, but it is only one way to view the situation.
The most important thing about the IBM i platform and its history, dating all the way back to the System/3, the great-great-great granddaddy of all of the IBM Rochester RPG machines aimed at midrange shops, is of course the people. And not just the people in IBM who made the platform what it was–we have celebrated them often throughout the years–but the several million or so people who have probably worked on these platforms over the decades, creating applications and tools. Not to mention the many tens of millions of users who have sat at those applications in hundreds of thousands of companies the world over.
A million may not sound like very many people in the 21st century, with a company like Facebook boasting 1.28 billion users. But millions of techies and tens of millions of users are significant numbers. No one knows for sure how many people have worked on SSP, CPF, OS/400, i5/OS, and IBM i applications over the years, but we have taken a stab at estimating it from time to time.
At its peak, the System/36 base probably had 250,000 unique customers, averaging somewhere between one and two programmer/operators on average across the base and anywhere from dozens to hundreds of users. The System/38 was architecturally important because many of the ideas brought to full fruiting and commercialization with the AS/400, but it never had anywhere close to the adoption rate of the System/36 or the AS/400. The numbers I have seen suggest that IBM might have sold something on the order of 75,000 System/38 machines in the decade prior to the launch of the AS/400 on June 21, 1988. The System/38 was wickedly expensive compared to the System/36, and on a MIPS-for-MIPS basis, including a relational database, it was about twice as expensive as an IBM System/370, System/3080, and System/390 mainframe of the same era. There were probably only 150,000 people who worked on System/38 machines, again including a lot of people who did triple duty as application programmer, database administrator, and system administrator.
The AS/400 expanded the IBM midrange base considerably, thanks in no small part to the widespread use of homegrown applications to manage organizations of all walks of life, from manufacturers to distributors to retailers to government agencies to regional banks to insurance companies. The AS/400 peaked at somewhere north of 8,000 third-party application providers with well over 20,000 different suites of software (that is not modules, but collections of modules). The installed base peaked in 1998 at somewhere around 275,000 unique customers, right during the ERP software boom when companies were shifting away from homegrown code–more precisely a hodge-podge of homegrown RPG programs and third-party applications that had been so heavily modified they might as well have been homegrown at that time.
Remember that customers buying third-party software on System/3X and AS/400 machines always used to get access to the source code–it was part of the license because many shops needed to modify the programs to suit their needs and to integrate disparate modules from other vendors and from the internal IT staff. Plenty of companies used off-the-shelf accounting modules–general ledger, costing, accounts receivable, accounts payable, payroll, and so on–but created the other modules to run their manufacturing or distribution operations because either packages did not exist or because it was thought to be easier to start from scratch than to try to rework someone else’s code. I knew some of the programmers who created complex MRP systems, all by their lonesome, from scratch back in the early days. It could be done, and was done. People forget that sometimes.
The typical AS/400 site probably had somewhere between two and four people on staff, so at the peak, with 275,000 customers, that had to be somewhere between 550,000 and 1.1 million AS/400 experts. Let us assume the high number. I also reckoned based on some IBM data I saw about the total DB2 database installed base that IBM peaked at something around 25 million AS/400-DB2 for i end user seats compared to around 22 million DB2 for IBM mainframe end users. (All end users on System/38, AS/400, and so on out to the current IBM i machines were database users, whether they knew it or not.)
The point is, there were probably well in excess of 1 million AS/400 techies at the peak, before the ERP transition put a lot of them out of work, and quite frankly, compelled many companies to move to other platforms rather than deal with Y2K date issues that were lurking in their code.
It is hard to say for sure how big the techie community is today, here in 2014, some 26 years since the launch of the AS/400. IBM says that there are north of 150,000 unique OS/400, i5/OS, and IBM i shops in the world, and I think a lot of them are relatively larger shops with more sophisticated workloads and a deep understanding of the substantial benefits of the IBM i platform–particularly if it is the thing you know best and know how to make run efficiently and reliably. If you fully burden the cost of a transaction processing system with people and end user costs, all of the system costs are barely a rounding error in the data. If your system allows you to hire cheaper or fewer end users, that is more important than moving to a so-called cheap X86 platform to save a few bucks on the front end of a deal. Then again, Windows and Linux are the familiar and safe choices these days for most companies, and that is the real problem as well as the perception (and often the reality) that Windows and Linux machines are cheaper than IBM i, Unix, or mainframe alternatives. You can justify the IBM i premium if you can program and run the system with fewer people. I think this is the case, but IBM has done a terrible job quantifying this over the decades. Precisely why is a mystery to me. Always has been.
My best guess is that there are probably somewhere between 300,000 and 450,000 AS/400 professionals still out there in the world today. The number of companies that have actively sold and maintained applications and other tools for the IBM i platform is on the order of 2,500 to 3,500, depending on who you ask at IBM, and these companies probably have on the order of tens of thousands of employees in the aggregate (including IBM, of course) and generate several billions of dollars in aggregate revenue. It is hard to say for sure how much money beyond this, and there are very large error bars on that data. I don’t think there are more than 30,000 very active IBM i accounts, by which I mean customers who keep on current releases of hardware and software.
As for end users, the world is a lot different today than the peak in 1998. If you mean end users who have an actual log-in screen on i5/OS or IBM i, it is probably somewhere between 10 million and 15 million seats, and I would guess it to be somewhere around 12 million or 13 million if you want to be more precise. But there are other ways to access these systems today, through Internet-facing applications or even through machine-to-machine transactions. It is hard to even guess what percentage of transactions against the Web servers, application servers, and databases come from these other “end users” but I think it is safe to say that even as the AS/400 installed has shrank, the systems still in use have become even more tightly woven into the fabric of the business and the world at large.
The point is, this is still a pretty large community and one still worth belonging to.