To 2032 . . . And Beyond!

December 7, 2020 Timothy Prickett Morgan

As you sit here in 2020, can you even imagine 2032? Can you imagine beyond that? Oddly enough, I can imagine three or four decades out from here a lot more easily than I can do a dozen years, even though I know the error bars get longer and the probabilistic clouds get fuzzier the further into the future you wander with your mind. So what does it mean, really, when Big Blue commits to support the IBM i platform at least until 2032, a mere dozen years away in a platform that, depending on how you want to cut it, dates back to 1969 with the System/3 or 1978 with the System/38?

In some ways, the support for the software is much more important than revolutions on the hardware crank every three or four years, which we detailed two weeks ago. But that cranking of Power hardware – from Power4 way back in 2001 when IBM converged its AS/400 and RS/6000 Power chip development and literally got its act together, to the expected delivery of Power10 in late 2021 in high-end machines and early 2022 in low-end machines and the Power11 chip in maybe 2025 and support for that chip at least for four years if not longer – give customers and partners the confidence that IBM is investing in the hardware and that implies that it will be investing in the software stack in a real and substantial way, not just fixing bugs and rolling in some new database features or open source packages in PASE or in Linux.

Therefore, we can see visibility, more or less, in Power hardware that will provide ample enough compute power for the vast majority of IBM i, AIX, and Linux customers that Big Blue has through 2030. And if you look at the various IBM i roadmaps, which are published in presentations and strategy documents, we can see the support to go out even longer. Here is the IBM i roadmap from 2019, to which I have made some modifications to update it:

First, I extended the light blue line for IBM i 7.1 to reflect the extended extended support that Big Blue added to this release back in October of this year, stretching its extended support by two years until the end of April 2023. That will make IBM i 7.1 the longest supported release in the history of the IBM midrange since 1969. IBM is clearly not afraid of longevity, and customers have some issues as well as some comfort. The red bars have been added to show three years of extended support after normal support is removed. So, you can see that with iNext or IBM i Next, which should coincide with the Power10 launch next year and into 2021, regular support runs to about 2030 by that roadmap from 2019 and then you can assume 2033 given extended support (which has its limits, as we have discussed often). For IBM i Next +1, which we presume coincides with Power11, it looks like extended support runs until late 2036 or early 2037, if history is any guide.

Now, to stay on that IBM i support schedule, you will have to be on reasonably current hardware, generally two generations back. So IBM i Next will require at least a Power8 system and IBM i Next +1 will require at least a Power9 processor. But IBM i 7.4 will very likely also be supported on Power10 and Power11 iron as well. But that will be the end of the road there, and IBM i 7.3 will probably only get to Power10. Again, if history is any guide.

And that is the thing. History is a guide, but not a guarantee, and IBM will tell you. The IBM i roadmap below, which was shown by Steve Sibley, vice president and offering manager for the Cognitive Systems division at IBM, at the Common Europe Online vCEC 2020 event a few weeks ago, is less precise in the dates covered but does come with some disclaimers. Take a look:

Let’s repeat in print what IBM says above so the search engines see it and old eyes with decades of looking at screens see it:

** All statements regarding IBM’s future direction and intent are subject to change or withdrawal without notice and represent goals and objectives only.
** Arrows indicate "ongoing status" and do not imply specific dates.

A roadmap, no matter who is drawing it, is only as good as the terrain you actually end up driving through, the vehicle that you take, and the fares that the riders on the platform bus are willing to pay to go where you say we all should be going. You need all three of these to actually get to where you are going.

It is important not to mix up intent with contractual agreement. A roadmap is not a promise, it is an intent, and while IBM has been very good in making good on its intentions, conditions change and so do regimes at major IT vendors. There is never a guarantee of anything, really. Just ask the 20,000-plus employees in Europe of IBM’s NewCo spinoff of its outsourcing and hosting business who are rumored to be soon handed pink slips – or rather, be given the sack if we are talking colloquially in England or renvoyé de son travail in France or essere licenziata in Italy or gefeuert zu warden in Germany, and so on. That’s around 22 percent of the 90,000-strong NewCo workforce. Conditions change, indeed.

Now, back to IBM i.

The good news is that a lot of Power Systems customers don’t need for IBM i to be doing a lot more than it is already doing. Adding support for new hardware features – we find the memory clustering technology in Power10 particularly intriguing – will be key, as will the addition of new languages and programming frameworks as time passes. We expect to see more tight integration with AI software stacks and features in IBM i that let it be more easily deployed on the public cloud and have its pricing metered more like a cloud subscription. The database and various kinds of resiliency and high availability clustering will also see improvements. But generally, IBM i Next and IBM i Next +1 will look a lot like IBM i 7.4. We are talking about fairly modest changes, at least compared to the early and middle years of OS/400, when so many things were changing.

Think about it. How much of the OS/400 stack was changed to make it suitable to host Internet protocols and workloads? Do we think there is a substantial, big change on that level coming to corporate computing? I don’t. And that means change will be less and less jarring even if it is a fairly regular pain in the neck and always carries a certain amount of risk that something can go wrong as systems software is upgraded and updated. The trick is to not get your applications trapped on an old release of IBM i. Far too many companies have done this, for a multitude of reasons, and this is why Big Blue keeps extending IBM i 7.1 out and out and even went so far as to support IBM i 7.1 on Power9 hardware only a few weeks ago in some kind of degraded mode.

We have been advocating for just this approach for many, many years, and we think IBM should go all the way back to IBM i 6.1, i/OS 5.4, and OS/400 V5R3 if necessary to get every customer on current Power9 iron. Why not? Charge for the software support and create an organization to support old releases. A lot of them. Stretch those light blue and red bars out to infinity and beyond! And charge a fair price for supporting so many releases. These customers have stable workloads, so it should be fairly easy compared to a Linux platform running AI software that changes every freaking day.

It’s a far better model than encouraging customers to move to Windows Server or Linux with Microsoft or Oracle databases running on X86 iron. IBM only sells the Linux part of that switch, and it is far less lucrative than supporting customers where they are and moving them ahead on hardware. Who cares if it takes more cycles to do this? IBM i shops and most AIX shops will have more cycles than what they want anyway in a Power9, Power10, or Power11 system.

And if IBM doesn’t want to support these heritage releases, then farm it out to a bunch of techies who know how to do it and will do it if Big Blue turns them on inside of the PowerVM hypervisor.

I’m just interested in results. There are many ways to achieve them.

This Issue Sponsored By

Table of Contents

Content archive

Recent Posts

Subscribe

Pages

Search