Paving The Road Ahead For A Better Ride
January 4, 2021 Timothy Prickett Morgan
We always sit behind the wheel of the present as we drive to the future with our baggage from the past in the trunk.
It is with this in mind that we contemplate 2021 and the uncertainty of regional, national, and global economies as well as how the coronavirus pandemic will be handled around the world in some pretty tricky political climates. These forces will affect all IBM i customers, of course, and we are not so much interested in describing all of these complex turbulences as they intertwine. What we do want to do is provide a few ideas as we start this new year to help get the IBM i community to a place with a better climate, which we all know of out there. This is not a wish list or a New Year’s resolution as such, but rather a listing of ideas that we think can be implemented that will increase the viability and resilience of the IBM i base and therefore Big Blue’s own IBM i business as well as the many downstream software and services companies that live in the IBM i community.
The very first thing that Big Blue can do, and this is so simple, is to admit that among enterprise customers, the IBM i base is still its very largest enterprise computing customer base. There may be millions of licenses of Red Hat Enterprise Linux out there, but we would guess – and we have to because neither Red Hat nor IBM has ever given us a customer count – that there are still, at somewhere around 120,000 unique customers worldwide, more IBM i customers than there are Red Hat customers in the enterprise. (We are not counting supercomputing facilities and massive server farms at the telcos and service providers here, and besides, there are only a few thousand supercomputing centers worldwide and probably only a thousand or so relatively large service providers at that.) So what we want, right off the bat, is for Arvind Krishna, IBM’s chief executive officer since April 2020, right when the coronavirus pandemic was really kicking in, and for James Kavanaugh, its chief financial officer, to actually talk about the IBM i business and its customers like they matter to International Business Machines. Not IBM Systems customers, not Power Systems customers, not Cognitive Systems customers, but IBM i customers. Let the IBM i customer base know that they are on the CEO’s and CFO’s radar, and that they matter.
That leads me to the next thing I want IBM and its Power Systems channel partners to do. I want IBM to actually make a concerted effort to identify the workloads that are running on X86 iron at IBM i shops and figure out ways to either port this code to Power Systems through integrated runtimes or find alternatives running on Linux partitions. There is a vast amount of work that can be brought back onto Power iron, and it is amazing to me that Big Blue doesn’t even try to do this. If the market is going to shift to Linux and away from Windows Server in the long run for many kinds of workloads – and that is what it looks like to me from the trend data – then get on the front of that curve and help drive Power Systems as well as Red Hat revenues using IBM i shops as the testbed for this.
If we are driving to the future, we need to get all of the suitcases in the trunk and we need to get everybody – and I mean everybody – into the car before we leave. If IBM can support a limited edition version of IBM i 7.1 on Power9 iron atop logical partitions running on the PowerVM hypervisor – which was not initially supported on Power9 iron and which was just unveiled in November 2020 – then it can go all the way back through IBM i 6.1, i5/OS 5.4, OS/400 V5R3, and anything else older that is necessary to get every operating system release properly supported. And I do not care if there are three layers of hardware and software virtualization to make this happen because the Power9 chip has so much more oomph than a Power5, Power6, or Power7 chip that it doesn’t matter. I don’t care if 90 percent of the performance of a four-core Power9 chip is eaten by the virtualization; this would still feel like an upgrade to customers on older iron. With Power10 not coming to entry and midrange Power Systems machines running IBM i until early 2022, which we told you about back in November, we have more than a year of hardware sales desert to cross. So let’s pick every damned cactus up on the way and turn it into tequila.
Given the nature of the mission critical, system of record work that all IBM i shops do day in and day out, every – and I mean every – core IBM i system should have high availability software of some fashion running and it should not be prohibitively expensive to do this. I would argue further that every core IBM i system should have HA replication locally – either on a shadow partition on a single physical machine or on a separate machine within the datacenter or within a metro distance where synchronous replication works – with a third asynchronous replication out to a public cloud that supports IBM i on Power servers of some sort. I have long advocated that IBM use the latent extra cores on entry Power chips – which now have a dozen fat SMT8 cores on them – to provide a local replica of the entire system. If it were me planning this, I would have two local replicas: one in a partition that is absolutely synchronous with an absolutely synchronous replication journal and another one in another partition that is slightly out of phase, perhaps with somewhere between no more than a twelve-hour delay but no less than a four-hour delay, but with the full replica journal. This way, if there is a hack of some sort, there is a way to jump to a non-hacked copy of the system. And then there would be multiple images of clean machines stored out in the public cloud to fall back on. This may sound paranoid, but the Internet is not nice, and hardware is cheap if you think of it that way. I am not saying that these shadow partitions could be used without paying a fee if customers wanted to run business intelligence workloads against them. But allow this, too. The point is, make this machine not only modern, but absolutely rock-solid for all customers. Work with the major HA vendors – we are down to Maxava, Precisely, Rocket Software, HelpSystems, Shield Advanced Solutions, iSam Blue (which has a variant of the BugBusters HA that HelpSystems acquired a few years back) at this point in addition to IBM’s own PowerHA of course – to front end all of this stuff so customers don’t have to change software they already have. But somewhere around 100,000 customers do not have HA software, compared to the 20,000 or so who do, so from that point of view, it doesn’t matter. They are starting from scratch.
Db2 Mirror database replication is separate from this, but applies equally well. There should be replicated databases as well as whole systems, and at an affordable price.
Finally, marketing support for the IBM i channel partners is more important than ever. We are not experts in what IBM does and does not do, but we live downstream from some of those marketing development funds, or MDF. IBM, like other IT vendors, throws such funds around their partner channels, and the word we have gotten over the past several years is that Big Blue has turned off the spigot, or at least has turned the knob to the left a little too far. (It depends on who we ask.) Luckily for us, our vendors have been supporting us through thick and thin because we support them through thick and thin. But now, in a year when we are not getting a Power Systems upgrade cycle when we expected it for several years, it is more important than ever for IBM to be creative and supportive with its partner channel to help them generate revenue. I don’t have any specific recommendations as to how such funds should be allocated and leave this to the partners to tell IBM. I just know that it needs to happen, and with a kind of generosity that Big Blue has not exhibited – as far as we know – in recent years. And it cannot just be pushing Linux on Power, either. This has to be MDF that helps the initiatives above, and specifically as regards the IBM i customer base and partner community.
If you have any other ideas on how to make the ride in 2021 a bit easier, pipe up.
Lol… this is the elefant in the room: IBMi lacks a native graphic terminal and GUI libraries, supported fully by IBM.
So if you want to bring back x86 ERP style application give to customer an “i”ntegrated dev tool and gui and remote pixel solution, pure graphic terminals, PCoIP style or than renders something in browsers used as thin terminals for PCoIP (exactly as SEU and 5250, still used today for a reason, everybody knows them and are centrally available and a common language between devs).
Wake up, no more philosophy and more facts… you don’t need to port solution you need to attract NEW solution and killer solutions to the platform.
The only missing part is the integrated remote graphic terminal.