IBM i Software And Power Systems Upgrades Keep Rolling Forward
January 18, 2021 Timothy Prickett Morgan
During tough economic times, I am fond of reminding people – and especially myself as I try to remain calm in a recession or whatever the heck this is we are dealing with now during the coronavirus pandemic – that business, in the aggregate, never goes to zero. No matter how bad it gets, thus far in the modern era starting perhaps with the Renaissance and maybe even longer but certainly since the agricultural and industrial revolutions in the 1700s, people still need stuff and they need people to grow or make it.
That doesn’t mean it is easy, however. But it looks like the IBM i installed base, which is a diverse group of small, medium, and large enterprises that span the globe and every industry, has done pretty well throughout the coronavirus pandemic. That’s our assessment after participating in the webinar going through the results in the seventh annual IBM i Marketplace Survey performed by HelpSystems. I listened a lot more than I talked, which is always good, and learned that it was a weird year in a lot of ways.
For every story of a company under extreme duress – perhaps in the hotel, casino, or leisure industries that have been hurt badly by the global shutdown – there are others – certainly in the healthcare, trucking, logistics, transportation, banking, and retail industries – that have seen their capacity needs skyrocket and have had to accelerate certain kinds of projects, such as contactless payments or virtual private networking into systems, to deal with the workforce and customer changes compelled by the pandemic. This was a comforting conversation in many ways, a testament to the stability and resiliency of the customers who use IBM i platforms, complementing the stability and resiliency of the platform itself. (I don’t know how much causation and correlation is going on here.) For that, I thank my colleagues on the webinar: Tom Huntington, the executive vice president of technical solutions at HelpSystems; Alison Butterill, IBM i product offering manager; Ian Jarman, business unit executive for Power Systems Lab Services; and Brandon Pederson, worldwide IBM i product marketing manager.
What we always want to know right off the bat when the IBM i Marketplace Survey comes out is where people are at as it relates to hardware and software. And then we want to know what their plans are for the coming year. In this analysis, we are going to tackle the Power Systems and IBM i installed base as reflected in the survey results. And in a follow-on story, we will drill down into the plans that companies have as they looked ahead to 2021.
The IBM i Marketplace Survey was conducted in September 2020, and there were a total of 733 responses and 472 complete responses on the survey. Developers, system administrators, and IT managers continue to be the ones taking the survey, but this year was a little different in that North America – the United States plus Canada – represented 39 percent of those polled, compared to somewhere around 60 percent for years past. There was much more participation across Europe and Central and Latin America, with 24 percent and 27 percent of respondents coming from these parts of the globe, respectively. This is a much more balanced distribution and more closely tracks the actual IBM i installed base. This has probably caused some wiggling here and there in the data, but as you will see, it does not appear to have caused any drastic or inconsistent changes with past data that was more centered on customers in the United States and Canada.
It would be interesting to actually track precise customers each year as a control, to see how specific customers are reacting and follow them individually and collectively through the years. It is debatable if this would be more representative of the overall base. That may be a much thinner data set – this is how Gartner and IDC do it for some of their survey datasets – but a much more valuable one. That said, we think the dataset that HelpSystems has built up with the help of IBM and ourselves here at IT Jungle is valuable and representative of what is going on among those customers who are active.
I always make that caveat when I talk about this and similar surveys and their accompanying reports, and it is important to do this again. I think the people who read newsletters about the IBM i platform and who take surveys about the IBM i platform are the ones who know and care about the platform in a way that a lot of customers do not. These active customers are the ones on Software Maintenance, who pay maintenance on their hardware, who are more active in their application development, and also keep their hardware and software current. This is maybe a fifth of the base, and it is distributed across small, medium, and large companies and small, medium, and large enterprises – and sometimes big companies have relatively modest systems, and vice versa, so do think there is a 1:1 correlation there, either. We cannot necessarily infer from this dataset – somewhere around 500 customers statistically representing somewhere between 20,000 and 30,000 actual customers depending on the actual size of the OS/400 and IBM i installed base, which is estimated at 120,000 to 150,000 depending on who you ask and when you ask them – that it is applicable for the other four-fifths of the customer base. They could be similar, but I strongly suspect not. I think many customers in this other part of the customer base are on older hardware, systems software, and application software but no less dedicated to their platforms – even if many of them don’t know they have an iSeries, System i, or Power Systems running IBM i platform. Many only know they have a machine running an application that their business partner set up and they just depend on it and get on with business.
Having said all that, let’s walk through the installed base, starting with the Power Systems hardware and then the IBM i systems software. I have added in the data from 2015 to get all of the seven years of information in the charts. Sorry if they are hard to read. Here is the distribution of answers for the question, Which Power Servers do you own?
It is important to note that this data doesn’t ask IBM i shops what their primary server is, but which servers of any given processor generation that they own and the numbers add up to more than 100 percent – depending in the year, it adds up to somewhere around 140 percent. That implies that many shops have one machine, but some have more than one and some have many, which is consistent with other parts of the report. This data does not show the progression of upgrades for generations of single, primary machines at the company, and that’s important to remember.
Having said that, the curve for the 2021 report (which is really late 2020 data) are pretty consistent with times gone by, and that is refreshing. Power9 machines did not get on the radar until early 2018, and finished the year with 14 percent of the machine pool (which was presented in the 2019 report and hence is under the 2019 data in the chart), and a year later that Power9 share grew by 16 points to 31 percent of customers (late 2019 data shown in the 2020 report). And despite the coronavirus pandemic, it rose by 11 percent in late 2020 and Power9 has grown to represent 42 percent of the installed hardware base in the 2021 report. If you look at the data for Power8 machines, this is absolutely consistent, and all things being equal, we expect Power9 to break through into the middle 50 percents by late 2021 and to be reported at that rate in the 2022 report put out by HelpSystems, with Power8 probably dropping by around 10 percent of the base in that time. That is if historical trends persist. But historical trends may not persist.
If the economy weakens or if customers are concerned about their IT budgets, we may see an uptick in acquisitions of certified pre-owned Power8 machinery from IBM Global Asset Recovery Services or other third party used equipment dealers. In fact, sales of such machinery may have been happening in 2020 to keep the Power8 base as high as it is. It is hard to say, but as we have been discussing in a series of articles that ran last year, a strong economic case can be made for buying a used Power8 machine compared to a new Power9 machine for certain customers who do not need to be on the bleeding edge – or cannot be. The case is harder to make for Power7 iron (or Power7+ iron that is lumped in with it in the data gathered by HelpSystems), given the age of the machines and the limited software releases available, but there are, nonetheless, some customers who can’t move ahead, usually because they have lost their application source code or they are off maintenance for a long period of time from their third party application suppliers.
Now, let’s talk about operating systems for a minute. The curves follow similar paths, but in this case the question is slightly different. It asks: What version of IBM i is your primary operating system level?
This counts one OS/400, i5/OS, or IBM i release per site. So the figures add up to 100 percent every year and the data ignores the fact that customers have multiple machines that might have multiple releases. We can’t know how many machines have each license. But we can infer that customers who have high availability clusters will usually have the same release level on both halves of the cluster. Moreover, it doesn’t apply an operating system distribution to each logical partition, which would give an accurate distribution of operating systems across the maybe 1 million to 1.5 million logical partitions I think is in the installed base of maybe 250,000 to 300,000 Power Systems machines that have IBM i as their primary operating system. (See last year’s analysis of the IBM i Marketplace Survey results for how I came to that conclusion.)
There is about a 10-year cycle for an IBM i release in this base, which stands to reason given that IBM provides support and extended support for about a decade for each release. The point is, IBM’s intent is largely met by customer actions, and that doesn’t necessarily have to be so, as is evidenced by the IBM i 6.1 and earlier releases that are still in the base. IBM i 7.2 is on the way out, IBM i 7.3 has probably peaked, and IBM i 7.4 is beginning its rise. What seems clear is that there is a lot of IBM i 7.3 and IBM i 7.4 on Power9 machinery, and probably a lot of IBM i 7.2 and 7.3 with a smattering of 7.4 on Power8 machinery, but we are inferring that from the two quasi-compatible datasets represented in the charts above.
If we had infinite time and customers had infinite patience and perfect knowledge, we would ask each survey respondent what Power Systems machines they had anywhere in their companies, how many logical partitions they have atop PowerVM, and how many of each release of IBM i, AIX, and Linux they had on those partitions. This would be very precise, and precisely zero people would take that survey.
That said, I kinda want to make that survey myself and see if you will take it. Where did I put that SurveyMonkey login. . . .