The IBM i Base Not As Jumpy As It Has Been
March 22, 2017 Timothy Prickett Morgan
The backwards compatibility of RPG and COBOL applications on new hardware and new operating systems in the IBM midrange is unheralded in the IT sector, and perhaps is only rivalled by the longevity of applications running in Big Blue’s System z mainframes. Somewhere out there in the world are applications running on IBM i platforms that could be running code that stretches all the way back to 1969 with the System/3.
Change is measured in the IBM i base, and with good reason. Small and medium businesses are conservative by nature because they don’t want to run any unnecessary risks, particularly with the transaction processing systems that distribute their products and services and, in a very real sense, literally represent their businesses. In general, they are also sensible people, and they are not apt to follow fashion for fashion’s sake. Many IBM midrangers have been around a long time, and they have literally seen it all in the datacenter. This is a good quality for people who are in control of the systems that count the money.
With everything being computerized these days, it is almost ludicrous to talk about what is mission critical and what is not. But I think it is safe to say that the important systems that cannot go down and that cannot have corrupted data are those that keep track of the money. In a sense, IBM founder Thomas Watson moved from selling cash registers at NCR back at the end of the 19th century to selling more glorified money counters in the 1960s, which still live on in the 21st century as the System z and IBM i platforms.
We shall see if Google and Amazon Web Services and Microsoft have such a long and illustrious run. I know I am watching it all like a hawk, with keen interest.
One gauge to how mission critical the IBM i platform is to the 120,000 or so companies who make use of it is to simply ask them how much of their business application portfolio they deploy on the system. As part of its 2017 IBM i Marketplace survey, that is precisely what HelpSystems, perhaps the largest provider of software in the IBM i market and we think generating well north of $100 million in revenues, asked customers as part of that survey. Here is how the distribution of responses looked:
This kind of distribution chart is interesting in that it gives us a feel for how much of the IBM i base is how dependent on the IBM i platform by application count. But it is a rough feel, and it clearly shows that some businesses are quite dependent on the platform.
But I like to stretch this kind of data and try to model the base at large. So, I got out my trusty Excel spreadsheet and did a little modeling. Let’s assume that there are 120,000 customers and that this data reflects what is happening in the base at large, and let’s try to figure out just how many applications we are talking about. To do that, I took the average of the application shares reported, which I am calling a weighted average. It is literally the midpoint of all of those ranges, and the data, when drilled down, may not be the midpoint of all of those ranges. But if you accept that, you can get a share of the sites and an average share of the total business critical apps at the IBM i shops. Then, you to get a sense of the application base, you have to assume some average number of applications per site. I picked ten because I think that is a pretty good average number for most IBM i sites. I then reckoned the total number of applications at these sites and the total number of IBM i applications, and it looks like this:
Of the base of 1.2 million applications across those 120,000 IBM i shops worldwide, over 750,000 of them are running on IBM i, which works out to just under 63 percent of the applications across the entire IBM i base.
For fun, I jacked up the weighted average by 10 percent on each of the tiers and jacked up the application count to 15 apps per site on average, and when you do that, you get 1.23 million IBM i apps across a base of 1.8 million total business critical apps in this base of customers, or 69.2 percent of the total application portfolio. If we adjust the weighted application share downwards by 10 percent and keep the average application count at 15, you have that same total application base of 1.8 million applications at the IBM i shops in the world, and over 1 million of them are running on IBM i itself, for a share of 56.6 percent.
My point is this: there are probably at least 1 million, and maybe many millions if the average is higher, applications running on IBM i in the world, and the share that IBM i has is well above 50 percent and possibly as high as 70 percent.
That is what those couple hundred thousand RPG, COBOL, Java, and PHP programmers worldwide are doing every workday day, and they don’t work for free so there must be a lot of value in this. And as I have pointed out in a prior story, the majority of sites in this same survey say they have homegrown applications, data which finally backs up what we all knew to be the case.
Given all of this investment in applications and detailed knowledge of a particular industry and a specific business, it makes perfect sense, at this point in the history of the IBM i platform, that most IBM i shops are not trying to get off the platform and most expect to increase their investments.
In fact, if you look at the data in the 2017 IBM i Marketplace survey, a quarter of the customers are increasing their investments in the platform and only 7.3 percent reported that they had a plan to move all of their applications to a new platform. And within those customers, a little more than half say that are going to do it in two years, just under a quarter say they will do it in five years, and another chunk expect to take ten years or more. Another 12.3 percent say they will migrate some applications off the platform to a different one.
If you look at the data over the three years of the IBM i Marketplace survey, the share of customers who are saying they are migrating has gone down every year. That could be because some sites have indeed left so they are no longer in the data. But I think it is because those who could leave – or more likely were compelled to leave because of a change in CEO or CIO or a merger or acquisition where IBM i comes up against Oracle or Microsoft or some other database platform – have long since left. After all, a lot of customers have left. The AS/400 base peaked at 275,000 shops in 1998, just as the ERP wave was under way and before the dot-com bust and 9/11 wrecked the economies of the world. That is a loss, on average of just under 8,200 shops per year over those 19 years from the peak of the base to today, and we think the bleed rate was very high in the period from 2002 through 2005, then slowed until the Great Recession in 2008, when it accelerated again until maybe 2012 or so. I would be very surprised if more than a few percent of the IBM i base actually leaves each year, and we think IBM is adding customers, particularly in Asia. Losing 2.5 percent of the base per year is 1,800 sites, and IBM could be adding this many each year, keeping us in stasis. The bleed rate is probably a percent or two higher, but that is just a guess on my part.
The point is, we think the base is more stable in 2017 than it even looks like in this survey data. People talk about moving more than they actually move. So the IBM i base is not as jumpy as it looks, mostly because such moves are tough to justify considering the deep skill and considerable investment that companies already have with IBM i.