IBM Takes A Hands Off Approach With Red Hat
July 15, 2019 Timothy Prickett Morgan
IBM has been around long enough in the IT racket that it doesn’t have any trouble maintaining distinct portfolios of products that have overlapping and often incompatible functions. The System/3, which debuted in 1969, is only five years younger than the System/360, which laid the foundation and set the pace for corporate computing when it launched in 1964. Both styles of machines continue to exist today as the IBM i on Power Systems platform and the System z.
With the $34 billion acquisition of Red Hat, which closed last week, neither of those two legacy products are under threat and IBM does not seem to be inclined whatsoever in ceasing development of the legacy operating system and middleware stacks embodied in the IBM i and System z lines.
As Arvind Krishna, senior vice president in charge of IBM’s cloud and cognitive software products, put it bluntly in a call after the deal closed, IBM’s customers expect for Big Blue to maintain its own operating systems, middleware, storage, databases, and security software in the IBM i, AIX, and System z lines, and that is precisely what Big Blue is going to do. Krisha estimated that there is only about 5 percent overlap in products between Big Blue and Red Hat – something we talked about at length when the deal was announced last October – and added that in many enterprise accounts that use both Red Hat and IBM platforms, companies invest in both sets of software for different purposes – perhaps using JBoss in one case and WebSphere in another, for instance.
We think the overlap is a lot more significant than 5 percent, particularly if you diced and sliced it by revenue streams instead of customer count. But it doesn’t matter so long as IBM remains neutral about supporting both its own code and that created by Red Hat through the open source community. For now, IBM seems keen on keeping Red Hat a Switzerland of Software, much as EMC did in the wake of its acquisition of server virtualization juggernaut VMware a decade ago and as Dell, which bought EMC and VMware a few years back, has done since then. We think that in the long run, development on those legacy IBM platforms could slow as the Red Hat stack penetrates more deeply into the enterprise – something that IBM is counting on to get its $34 billion back and then some, no matter what it says – and certainly we think that the relative cost of these platforms will continue to diverge as well. No one would argue that a mainframe platform is as cheap as a Linux on X86 platform, but conversely, Linux provides great value for mainframe shops and is the main driving of spending on capacity in the 21st century. IBM has wanted the same Linux Effect on its Power Systems line for more than two decades, and Linux is finally gaining some traction there.
The reason for this hands-off attitude for such expensive acquisitions is simple: Both VMware and Red Hat live and die by the fact that they are neutral to any particular platform. While IBM may prefer Red Hat’s various elements of the stack – the Enterprise Linux operating system, the OpenShift container system, the OpenStack cloud controller, the JBoss application server, and the Ceph block and object storage – it cannot prevent Red Hat’s vast partner network from doing what they do, which is compete against IBM and each other selling Linux stacks. To not do so would be to destroy the value of the Red Hat business, just as Dell would destroy the value of the VMware business if it had only put VMware’s virtualization and cloud software on its own PowerEdge servers.
“The problem we had back in the Unix days, and to a certain extent in the Linux days as well, was companies going with these siloed approaches,” explained Paul Cormier, president of products and technologies at Red Hat, explained in a conference call he cohosted with Krishna last week when the deal closed. “I think we both recognize that won’t work here and that won’t scale the technology. That independence is essential to assure our partners – who may be competitors to IBM – that they have an equal shot at the business out there.”
Which, of course, begs the question: Why did IBM pay $34 billion for Red Hat?
We told you this before, but it bears repeating. IBM’s own software platforms are either stable or declining, with the recent exception of IBM i and Linux on Power Systems in recent quarters. But if you average the trendlines out over decades, IBM’s legacy platforms – and I include AIX in that bucket alongside IBM i and z/OS – have shrunk tremendously even if that shrinkage has stabilized to a nominal rate now. In the 1990s and 2000s, the declines were like a bloodletting, and now the diehard customers are what is left. The Power Systems and System z lines are still worth maintaining and extending, and will be for many, many years. But what IBM really needs is a cogent and coherent story to tell Wall Street and its customers about the hybrid cloud journey, and Red Hat gives Big Blue that story and a hell of a lot more open source cred to boot. You may argue that IBM could have done a lot of things with $34 billion to help build out and promote its own platforms, but there is no stuffing the Linux genie back into the bottle. Even Microsoft knows that Linux is the only operating system platform that is growing and that, if you add in the underpinnings of the hyperscalers and cloud builders for their vast infrastructures, Linux accounts for more than 50 percent of the server footprints installed in the world. That is a position that Unix, as a collective, held in the 1990s, and that Windows had in the 2000s, and that now belongs to Linux – very likely forever because of the open, collaborative way it is created. Linux is like Ethernet now – impossible to kill, and any new idea can be adapted and adopted for them.
The point is, Red Hat grows at around 20 percent a quarter or so, and if IBM can boost that growth to 25 percent or even 30 percent, it can get its money back – all $34 billion of it – as revenue for its software and cloud divisions by maybe 2023 or 2024. That’s not a bad bet to make, particularly for a software company that has been fairly stable in the past decade and that will be accretive to free cash flow and gross margins in its first year after the acquisition and to earning per share in the second year, as Krishna put it.
IBM has its eyes on a very long term prize, and that is the $1 trillion opportunity as companies undergo a massive wave of modernization of their workloads and systems. “We believe that hybrid cloud unlocks tremendous value and is the one way forward for our clients, who are getting ready to embark on the 80 percent of the cloud journey that has yet to happen,” explained Krishna. “Yes, you did hear that right – only 20 percent has happened so far. Some people say public cloud is the only destination for 100 percent of all workloads, and we believe that is not accurate. Given that it has been a decade and only 20 percent of the workloads have moved, there is a lot that is going to stay private and on premises, and we need a way to operate in all of these environments as opposed to having different siloes that can’t and have skills fungibility across all of them.”
The good news is that IBM has a hybrid story to tell now with IBM i and AIX, finally, and if you don’t care about Linux then you can continue to not care about it. The other good news is that IBM is going to be pushing Linux hard through Red Hat and benefitting when all of its competitors also push Linux. We suspect that just like Red Hat had a revenue bump in the quarter after the IBM deal was announced, SUSE Linux will also see a bump because some companies will want an alternative. But the SUSE Linux stack, as good as it is, is not as broad or deep as that of Red Hat, which probably has a base of 4 million servers in the world running RHEL and its CentOS clone. That’s around 10 percent of the server base, and the biggest hyperscalers and cloud builders, who create their own Linuxes, probably account for another 20 percent collectively. There’s another 20 percent or more of the base that is either running another distribution (a small piece) or are self-supported at service providers, telcos, enterprise shops, and supercomputing centers (a much bigger piece).
For now, and probably for a long time, Red Hat keeps its brand, its own facilities and product and sales teams and IBM is strictly hands off. To be precise, the Red Hat product teams are being given the freedom to decide on their own product roadmaps and development approaches, and the companies’ two partner programs will be kept separate, too. Big Blue is encouraging customers to partner with IBM and Red Hat, however. Red Hat’s sales force is not selling IBM software products and is not getting commissions or compensation for any IBM sales, either, which keeps the sales teams separate. You can bet that RHEL will be the preferred Linux on Power Systems and System z going forward, of course. But that is not a new status.
RELATED STORIES
The Impact On IBM i Of Big Blue’s Acquisition Of Red Hat
Kubernetes Container Control Comes To Power Systems
The Platform Matters More Than Ever, The Operating System Less So
IBM Will Change WebSphere To Work In A Cloudy World
The Cognitive Systems/500 2018 Edition
Big Blue Tweaks Red Hat Deal for Power Systems
Red Hat, IBM, and Friends Gang Up On VMware, Ignore PowerVM
Will Red Hat Cloudware Come to Power-Based IBM i Clouds?
Hello Timothy,
Your statements above about the IBMi being based off the System/3 is not true. You should have referenced the S/36 Environment running under IBMi as the System/3 reference.
The S/36 was the last machine of the System/3 product lines that IBM introduced in the early 80’s. That included the various System/3 models, plus the S/32, S34, and finally S/36.
IBMi started on a future system called S/38 where IBMi was originally called CPF before being renamed OS/400 on the AS400. The S/38 was a single 370 channel system and the AS400 was a 370 bus architecture, before going thru the CISC/RISC conversion in the mid 90’s.
This is when VMC was rewritten in C++.
BTW, the VMC (vertical microcode) was really VM and IBMi runs as a guest. The machine pool contains the 2 OS’s with VM booting up first and then IBMi is initialized and booted up under VM.
IBM named VM as microcode to protect the code from VM system programmers.
I started in IBM in July 1978 working in GSD division and got trained internals on the S/38 to become a PSR. We were level1/level2 in the field before the support centers started up during the mid to late 80’s. Rochester took over the field support for S/38 in the summer of 1987, almost a year before the AS400 was announced.
So the roots of the S/38 were really S/360/370.
IBMi and Microcode are mainframe OS’s, which IBM still refuses to acknowledge.
But this goes back to IBM making sure this platform never competed against the mainframe Even limiting the bells and whistles of the CPU’s Rochester could manufacture.
Need to edit to Sys/3 is 5 years younger or more recent than 360. Always interesting articles as usual, TPM. thanks Ralph
Correct. It was late and my dyslexia was kicking in.