What does the future hold for the IBM i? There are some definite clues about the direction IBM i running on Power will take. Considering that nobody knows with 100% certainty, let’s concern ourselves only with what we know so far, and what we can predict with some degree of reliability. First, let’s get a firm handle on where we are at the moment.
Who’s using IBM i?
While hard figures are tough to come by, IBM has previously estimated the global IBM i installed base to be around 120,000 organizations. These estimates are several years old, so assuming that the installed base is shrinking at about 3% annually, the current figure is somewhere around 110,000 individual organizations using the IBM i today.
These organizations run the gamut, from midsize distributors and manufacturers to retailers and banks. Some of the biggest companies in the world, like Honda and Toyota, are IBM i users. You can also find IBM i servers in government institutions, from small county assessors in Texas to the United States Air Force. It’s used by some of the biggest names in media (Walt Disney Co.) to some of the most well-known sports teams (Chicago Cubs). The website All400s publishes a spreadsheet that contains the names of tens of thousands of known IBM i, i5, iSeries, and AS/400 users around the world. SEA Software is one of many IBM i vendors who sponsor this site. Check it out if you’d like more information about who is part of the IBM i community
Demography, geography, and IBM i installations
IBM has traditionally broken the midrange server market up into two main camps: large companies with more than 1,000 employees and small and medium-sized businesses (SMBs) with fewer than 1,000 employees. According to IBM’s recently published “IBM i Strategy And Roadmap report,” 30% of IBM i shops are large companies, while 70% are SMBs, which is the same ratio it gave back in 2016, the last time it published the report.
According to IBM, one aspect of the IBM i installed base is changing: the geographical makeup. In 2016, IBM said that 70% of its sales came from North America, Western Europe, and Japan. In 2019, that number has increased to 80%. Clearly, the continued economic expansion is bolstering sales in these established markets. Companies in Japan, which is the second largest market for IBM i gear and software, are still spending millions on this platform. In fact, at a recent IBM i conference in Japan, there was standing room only in the sessions, according to reports.
But IBM i is not defined by these three markets. IBM says that it’s seeing continued growth in emerging markets such as Latin America, Eastern Europe and the Association of Southeast Asian Nations region (ASEAN), especially in the banking and distribution sectors. “While China tends to be a growth market dominated by Unix,” IBM writes in its roadmap report, “IBM i continues its strong presence there, especially in the banking and financial services sectors.”
Churn, churn, churn
There will be churn in the IBM i installed base. It’s likely that more companies will choose to leave the platform than adopt it for the foreseeable future. But anecdotal evidence suggests the pace of migration off the platform has slowed in recent years, leaving behind a solid core of dedicated users that is relatively stable. While the midrange server’s heyday is behind it, there’s nothing to indicate that a surge of migrations off the platform is imminent.
Now let’s switch to the platform itself.
The Next and Next+1 steps to IBM i 2032
IBM is committed to supporting IBM i at least through 2032. That minimum 13-year commitment should tell any CIO that IBM is not walking away from the IBM i operating system any time soon. The three current releases of the IBM i operating system – versions 7.2, 7.3, and 7.4 –will be supported by IBM until 2022, 2023, and 2026, respectively.
The IBM i Roadmap document also commits to releasing the next release of IBM i – dubbed “IBM i Next” – in early 2022 and supporting it through 2030. The following release, “IBM i Next +1,” is currently slated to ship in early 2025 and be supported well past 2032. IBM is essentially promising customers that applications running on IBM i today will continue to run on future releases of the operating system for well over a decade. This should allow CIOs to sleep well at night.
The IBM i of tomorrow
What sorts of new features and capabilities can we expect to see in these future IBM i releases? This is obviously an area of speculation, as IBM itself, fresh off the delivery of IBM i 7.4, is now turning its eyes to the next Technology Refresh. But judging from IT industry trends, we can make some educated guesses about the types of things we’ll likely see.
For starters, don’t be surprised if IBM i at some point adopts the cloud-native virtualization technology called containers.
In the X86 world, companies are rapidly adopting Docker, a containerization runtime, and Kubernetes, which is used to orchestrate containers. Together, Docker and Kubernetes make it much easier for companies to create, scale, move, and destroy server operating system instances without regard to the underlying hardware or the runtime environment. Because everything that’s needed to run a given application – including the operating system, databases, frameworks, and all other code dependencies — is bundled into a container, it dramatically simplifies the handoff between developers and operations professionals. A similar technology could be implemented in the IBM i world.
The cognitive and AI areas are heating up, and will likely impact IBM i at some future point.
The IBM i is primarily used to run transaction-oriented workloads, but IBM is very interested in adding cognitive and AI capabilities to the mix in a hybrid manner, what’s commonly called hybrid transaction/analytical processing (HTAP) or “translytics.” IBM CEO Ginni Rometty recently pointed out that 80% of the world’s data sits behind corporate firewalls, and IBM is eager to help customers tap into that treasure trove with its extensive Watson lineup of AI and cognitive capabilities. IBM i already supports Python, the most popular language for data science at the moment, and it added support for R in i 7.4, which is also heavily used by data scientists. As open source data science software matures and technologies become more productized and self-service in nature, don’t be surprised to see machine learning routines augmenting traditional programming and advanced analytics embedded into traditional applications, even in the cloistered IBM i world.
IBM Power systems hardware is changing
Things are also moving along at a nice clip over on the hardware front.
Work is well under way on the next generation of Power10 processors, which are expected to be introduced in late 2020 or possibly early 2021. There are also some rumblings coming out of IBM about Power11, which is in the formative stages at the moment. If past release timelines around Power generations repeat themselves, we should see Power11 come to market in the 2024 timeframe. There isn’t much talk yet about Power12, but there’s no reason to think that it won’t be delivered, probably around the year 2027 or 2028.
We will likely see more exotic processors types offered in Power Systems in the future. IBM has a partnership with Nvidia for graphics processor units (GPUs), which were initially developed to power high-end graphics for video games but are increasingly utilized as math co-processors in supercomputers. Now GPUs are being adopted to power production-grade machine learning programs, including the advanced new “deep learning” systems that are being used for image recognition and natural language processing (NLP) workloads. We’ll also see other specialized processors, like field programmable gate arrays (FPGAs), added to the hardware mix.
IBM’s Power Systems architecture already holds performance advantages over Intel X86 offers via its Coherent Accelerator Processor Interface (CAPI) and NVlink interfaces, and IBM is looking to expand that lead with advanced new storage capabilities that will be added in the future. In the X86 world, Intel and Micron are leading development of so-called storage-class memories, dubbed 3D Point, which enables NVMe-class storage devices (non-volatile memory express) to be utilized as either main memory or long-term storage. The main advantage of this approach is a vast speedup in I/O compared to traditional SSDs. While IBM is not talking much about it publicly yet, it has confirmed that it’s working on its own storage-class memory for Power Systems that it claims will be superior to Intel’s Optane. It’s also worth noting that IBM’s i 7.4 announcement contained a statement of direction (SOD) to add native IBM i NVMe support to the operating system. So, NVMe-class storage will probably be available for IBM i servers in the near future.
IBM i’s Cloudy Forecast
But the biggest change coming to IBM i could be the cloud.
The IBM i community has had many private cloud options available for many years. But it’s about to get the first options for native public clouds from IBM, Skytap, and Google. These clouds will allow users to create IBM i instances that they can scale up and scale down as needed, all from the comfort of a Web interface. As these public IBM i cloud options come online, they will have a dramatic impact on the way companies consume IBM i resources. Grouped together with future technologies like containers and cognitive capabilities delivered via specialized processors, the potential for the types of applications that can be served from IBM i will grow dramatically. This will also be a boon for vendors offering applications via the software as a service (SaaS) delivery method. However, there will need to be big changes made to traditional IBM i licensing agreements, which generally are not compatible with the way business is done on the cloud.
While the hardware and software roadmap for IBM i and Power Systems is solid, there are still substantial challenges facing the platform and its long-term viability. One of the biggest concerns is the fairly large contingent of IBM i shops running old and unsupported versions of the operating system, often on old and unsupported hardware. Estimates of the size of this group range from 50% of the overall IBM i installed base, to two thirds. Whether or not these customers become active in upgrading their systems and modernizing their applications could have a sizable impact on the direction the platform takes in the future.