The OS/400 Ecosystem, Part 2
by Timothy Prickett Morgan
Last week, I could accurately blame the idea of trying to reckon the size of the OS/400 ecosystem on desperation (there was no hot news in the iSeries market) and on a fever induced by a pesky summer cold. But this week, as I continue to estimate the size and impact of the OS/400 market, I have nothing to blame but inertia and my desire to fill in a gaping hole in knowledge and market data with the best guesses I can come up with.
In the story last week, I explained how I thought the OS/400 ecosystem was about $20 billion in 1993. According to my estimates back then, the combined sales of AS/400 servers, disk subsystems, other hardware (such as main memory and other peripherals), and systems software (OS/400, DB2/400, compilers, and other tools) was about $7.7 billion, and inside IBM, the total AS/400 ecosystem (including PCs and CRTs, maintenance on AS/400s and System/3X gear, and other services) was $14 billion. I made a leap from that $14 billion to the overall $20 billion based on the fact that IBM, back then, controlled much more of the AS/400 base and all of the thousands of application software vendors were generally a lot smaller. Most OS/400 shops bought minimalist systems and did all or most of their own coding of the application software that ran on the systems. In 2004, I reckoned that the OS/400 ecosystem was about $6.5 billion or so, and based on a five-year refresh cycle for PCs among the 17 million or so end user seats out there in OS/400 land, I reckoned they spend about $3.4 billion a year on PCs. Call the ecosystem (as defined by the breakdown of data I had in 1993) about $10 billion for 2004. That's about half of what I reckon it was in 1993. If you take out the PCs and CRTs from the equation (which is valid to one way of thinking), the OS/400 ecosystem from 1993 was probably on the order of $15.5 billion, compared to $6.5 billion today, which is a decline of about 58 percent. I didn't make that point very clearly last week, which is why I am doing it now.
Of course, the OS/400 ecosystem figures cited above did not take into account other IT budget items, and, most importantly for the OS/400 platform, the cost of all that in-house programming that was done in lieu of buying applications from third parties and paying maintenance on it. That is, of course, part of the OS/400 ecosystem, too. And, as it turns out, it is the dominant piece of the ecosystem, and I think it always has been.
My projections show that while people are a big cost in the IT market, they absolutely dominate the IT budgets at OS/400 shops. And this is not a contradiction with the long-held proposition that the OS/400 platform has a lower total cost of ownership because of its integration of operating system, database, and compiler tools; because of its ease of administration; and its ease of programming. On the contrary, by being an integrated system with a strong base of managers and programmers who were intimate with the OS/400 platforms and their businesses, IBM midrange shops have invested in people rather than in technology--and continue to do so. The modern IT credos are paradoxically "do more with less" and "buy, don't build." The OS/400 credos seem to be "do it yourself" and "do it better with your own people."
I respect that.
Calculating the people costs in 1993 and 2004 for the OS/400 installed base is not a trivial thing, but we can make some rough estimates. Depending on who you ask, there are about 215,000 to 225,000 unique OS/400 sites in the world. At many small sites, there are turnkey solutions and no IT staff or there is one programmer who is effectively the entire IT staff. There are a large number of sites with a handful of people and a manager of IT operations. And then there are enterprise-class sites with hundreds or even a few thousand programmers and managers. But I think, on the whole, there are about an average of three IT employees dedicated to the IT platform at OS/400 sites. (I didn't forget the system operators, either. I think there are still some. But I will use programmer as short-hand for meaning "not manager.") When you assume an average of salary of just under $90,000 for managers and just under $60,000 for programmers (which is consistent with data supplied by iSeries salary expert Nate Viall & Associates), assume a manager to programmer ratio of 1 to 5 on average, and then throw in a 25 percent markup to cover the costs of benefits, office space, and other overhead, then in 2004 the value of those OS/400-related IT people was a stunning $54.2 billion. Yes, that is six times what OS/400 shops spent on iSeries hardware, software, and services in 2004. Put another way, OS/400 shops invest six times as much money in their people as they do in their wares.
Now, because I am a curious person--many people have said that to me, and often to my face--I could not just leave these numbers at that. What I really wanted to know is how this compares to the IT industry at large. So, last week, in my fever-induced fugue state, I built a table that compared the overall, worldwide IT ecosystem to the OS/400 ecosystem, line by line. You can take a look at the table, which is called "Global IT Ecosystem Versus OS/400 Ecosystem, Table A" by clicking here.
In the top part of the table, I compiled some rough statistics on the site of what I call the IT wares industry--hardware, software, and peopleware (which is what services really is, after all). I hear vendors and IT market watchers rattle off figures about different market sizes all the time, and this table is an amalgam of some of those figures rounded to one or two significant digits. Getting more precise in the numbers is not particularly useful, even if it were possible. I am merely just trying to show you the shape of the IT industry, not its precise dimensions down to the last dollar. This is a big picture, so the pixel size can be large, too.
I reckon that in 2004, PCs represented about $200 billion in sales worldwide, and about half of these ended up in corporate IT departments. Servers represented about $50 in sales, and other hardware (storage, exotic networking gear, and such) came to another $100 billion or so. The ERP and core enterprise software market is, according to the people in the industry I talk to, was a $100 billion business in 2004, so that number was easy. Systems software, by which I mean operating systems and databases, was probably another $25 million (not including the software installed on consumer PCs), and other software from the bazillion of pieces of code that companies buy worldwide probably came to another $125 billion. Services came to a stunning $600 billion--peopleware ain't cheap, is it?--and telecom costs associated directly with the IT department--meaning some voice and a lot of high-speed data links--came to maybe $100 billion. That's $1.2 trillion, and that is consistent with the jibberish the big IT consultancies and IT vendors are talking about the size of the IT market. This model may not be perfect, but it has the right shape.
When you look at this, the first thing that jumps out is that services is half of the pie chart for IT wares. In the IT market, only a slice of the services revenue is for maintenance--it is hard to say how much, but it is probably less than 20 percent of the total. On the OS/400 platform, by contrast, services represent only about $1.6 billion, and I figure about $1.2 billion of that, or about 75 percent, is just maintenance on the systems in the installed base--for those companies that pay maintenance. (I think many do not.)
Now look at the OS/400 side of the table. If I have done my PC estimates right, PCs represent a big portion of the OS/400 ecosystem, but OS/400 spending on PCs is less than half of what it is in the corporate market at large. Comparatively speaking, OS/400 shops seem to spend more money on servers than does the IT industry at large (which is consistent with the fact that iSeries servers are more expensive than other servers), and they even invest a lot more in packaged software (as a percent of overall expenditures on various IT wares). When you do the math, the OS/400 ecosystem is just under a percent of the total IT ecosystem for hardware, software, and services, and if you remove PCs from the equation (as I did in the table to make this point), the OS/400 ecosystem represented 6/10ths of one percent of the IT ecosystem in 2004.
But when you add in people, you get a radically different picture. I have seen industry statistics that say there are about 10 million programmers in the world. If you add in managers, and system operators, and other IT professionals into the mix, and then take out the IT people who work on the development side of the IT vendors, then I think there may be as many as 15 million IT professionals in the world. (Again, not counting those who work for big consultancies. I do not want to double count the services spending, which is by and large an outsourced people cost.) If you assume about the same salary for the overall IT ecosystem as you have in the OS/400 ecosystem, that works out to $1.25 trillion dollars, which means overall IT spending--wares and people--comes to $2.45 trillion worldwide. About half of the total is IT personnel and about a quarter is services. The rest is hardware and software with a sliver of telecommunications costs. By contrast, about 84 percent of the OS/400 ecosystem is represented by people. Each OS/400 site might have few people, and they may get a lot done, but it adds up across the installed base of around 650,000 IT professionals. (The number could be bigger, and if you argue that it is, you are only proving my point that OS/400 shops invest heavily in people.)
As far as backcasting to 1993 goes, it is hard to say with any kind of precision how big the IT personnel budgets were. Back then, there were more personnel within IT shops because OS/400 did not include as many automation features as it does today. The programming tools were a lot less sophisticated, but then again, so was business in general and the applications that encapsulated business processes. What Nate Viall does know for sure is that OS/400 salaries for managers, programmers and operators were a lot smaller--a little over $56,000 for managers on average and a little over $36,000 programmers. If the IT personnel count remained constant, the people side of the IT ecosystem would represent about $32 billion. But I think headcount has gone down at OS/400 shops, in part because of the adoption of third-party ERP software, in part from data center consolidations and corporate mergers, and in part because there are fewer OS/400 sites than there used to be. In the late 1990s, when the Y2K boom was going, I saw a statistic that said there were 900,000 RPG and DB2/400 programmers out there in the world. Back in 1993, that number would have been smaller--it is hard to say how much smaller. But when you throw in managers and operators and other staff from the 1993 OS/400 ecosystem, you might have as many as 1 million people. Even with the lower salaries, that comes to around $50 billion.
That means despite all the arduousness of the OS/400 environment in the past 12 years, OS/400 companies have more or less maintained their commitment, in terms of dollars, in people. I find that stunning, and quite frankly, I am proud of it. The Fortune 500, the Fortune 1000, the Global 2000, and the Global 10000 can't say the same thing, now can they? There's something to be said for smart people working for a paycheck and loyal to a company. Who do third-party ERP application programmers work for--the customer or the vendor? You know the answer: mostly the vendor.
Now, look at the next table I built, which is called "Global IT Ecosystem Versus OS/400 Ecosystem, Table B". I have taken the PCs out of the equation--incidentally, just as I am writing this sentence, my young son Henry (almost four) has dumped a glass of water into my wife's laptop and I am in big IT trouble on the home front thanks to that little bit of synchronicity--and have moved the peopleware/services costs from the section of the prior table where the hardware and software costs were and put them down with the other people costs. And this time, I am also calculating the percentages for the share of all IT spending and share of OS/400 spending not just across the top part of the table (the "wares" section), but across all spending.
What's interesting in rejiggering the data in this manner--taking out PCs and putting services down in the people costs--is that it shows what we all know intuitively: the IT market invests more in third-party solutions than the OS/400 market, and the OS/400 market invests more in people to do IT themselves than does the IT market as a whole. The IT market as a whole spends a lot of dough on services--about half as much as it spends on its own people, which is a lot--and that the total people costs make up almost 79 percent of total IT costs (again, if you consider services as a people cost). The OS/400 ecosystem as I have estimated it for 2004 has devoted more than 91 percent of total expenditures on people, and the vast majority of that money is spent on employees, not contractors or outsourcing.
The OS/400 ecosystem is, in this regard, very old school. And I think that in this case, this is something to be admired. There is absolutely no evidence to suggest that moving to big, off-the-shelf software suites is better than building your own code, particularly if you have been at it for decades as many OS/400 shops have been. This is not a good strategy perhaps for small businesses who don't want to invest in a programmer, but for those who already have programmers and budget for them, all I can say is that it seems to be working. Of course, the buy-versus-build choice depends on the business, the current state of the code, and the quality of the programmers and the managers, and whether or not they are going to retire in a few years. In a relatively stable, niche business--like the kinds of midrange companies that are the typical OS/400 shop--staying away from low-end and high-end ERP solutions and paying some smart people to create and maintain applications seems to be the chosen way (even if it may not be the best way) to go about creating OS/400 applications. Even after two decades of thousands of independent software vendors telling them they ought to stop doing this.
Maybe we should all think about that for a week or so before I compare the OS/400 ecosystem to those of other platforms.
[Editor's note: I took the peripherals out of the laptop, shook out as much water as I could, used compressed air to blow out some more, and hung the machine upside down in the data closet, where our servers kick out lots of heat. (I have also dried fruit and sometimes wet laundry in there, too.) In the morning, the laptop booted and Henry, who went to bed quite sad because he thought he had killed a computer, was much relieved. Me, too. Laptops don't grow on trees, you know. I think Henry understands that now, too.]
How Big Is the OS/400 Ecosystem?
TFH Flashback: Critical Mass, November 1993 issue