How Committed Is Big Blue To The IBM Cloud?
May 23, 2022 Timothy Prickett Morgan
Before you get all nervous, I did not ask how committed is Big Blue to the Power Virtual Server. So don’t jump to the wrong conclusion. But we are beginning to wonder just how committed IBM is to the idea of operating a globe-spanning X86 server cloud that competes with the likes of Amazon Web Services, Microsoft Azure, Google Cloud, Alibaba Cloud, Tencent Cloud, and Baidu Cloud.
What got us to thinking about this was an announcement by IBM that it has signed a “strategic collaborative agreement” with AWS, which is just facing the facts that AWS is the juggernaut of cloud computing and that IBM, after nearly a decade of playing the game, most certainly is not. And it is not alone in this.
Dell and Hewlett Packard Enterprise, seeing the rise of AWS and then Azure and then the others, both committed to building their own clouds based on the open source OpenStack cloud platform, and they shut them down long ago because the amount of investment is just enormous and prior investments are what get current customers. If you didn’t start in the cloud in earnest when the Great Recession hit and if you didn’t have some kind of cash flow from which to make these huge investments, you didn’t make it.
Rackspace Hosting, the co-founder of the OpenStack project with NASA had aspired to build a cloud of clouds, with thousands of telcos and service providers all interconnecting their datacenters to create something that had the scale and scope of AWS. And when that didn’t work out, it went private, shut down its cloud, and is basically became an AWS reseller with some hefty technical skills.
VMware had a similar vision in creating interconnected “software defined datacenters” based on its ESXi hypervisor, vSphere and vRealize tools, and other cloud extensions, again with the telco and service provider partner angle as well as its own datacenters, and in the end, VMware chucked it all and did a partnership deal with AWS to run the VMware cloud stack on bare metal AWS gear. In fact, VMware had a deal with IBM Cloud to host its cloud stack on Big Blue’s cloudy infrastructure, if you will recall.
Just to give you a sense of the scale these big clouds operate on:
Google is spending $9.5 billion this year on new regions and the datacenters that comprise them as well as on the servers, storage, and switching inside of those datacenters. And all of these Google machines are driven by actual – and we think fairly predictable and growing – workload demands from Google itself and its outside customers, not a Field of Dreams approach: “If you built it, they will come.”
IBM had high hopes for being a public cloud player when it acquired SoftLayer in June 2013 for approximately $2 billion. But it took three years to get Power-based machines – in this case, entry Power8 servers with two sockets – onto the SoftLayer cloud, and when they did finally get there, they only ran Linux on top of the OpenKVM hypervisor and they did not run IBM i or AIX – the operating systems used by the vast majority of IBM’s actual Power Systems customers. These days, IBM does offer its Power Virtual Server, which has taken a step sideways and delivered IBM i, AIX, and Linux atop its PowerVM hypervisor. This is the kind of cloud offering IBM should have made available a decade and a half ago, when it was clear that AWS was going to be a thing. IBM got the job done, finally, and it is at least competing against Microsoft/Skytap in delivering IBM i to customers. Google has some Power iron in its eponymous cloud, but as far as we know it has never delivered on its promise to support IBM i on Google Cloud. And of course, there are myriad cloud providers and hosting companies that will sell you time slices of IBM i capacity on Power iron.
Under the deal, IBM’s Software group is offering its wares under subscription-based pricing on the AWS cloud, as it has already been doing on the IBM Cloud. The very existence of the deal suggests that IBM’s own customers want to run IBM’s software on AWS iron, not on what is still, essentially, the SoftLayer X86 infrastructure. The key to all of this – and it really tells you how much Big Blue’s focus has shifted – is the Red Hat OpenShift Service on AWS, what it lovingly calls ROSA. This is what IBM is really worried about: That OpenShift run easily and smoothly and effortlessly on AWS so it can be the hybrid cloud substrate on private clouds and on the AWS public cloud. (I don’t like using the term “public cloud” anymore because these are not public utilities, but very private platforms – just as proprietary as the AS/400 or the System/360 ever were.)
As part of the agreement, IBM and AWS are making joint investments to make it easier to consume IBM’s software – against, mostly OpenShift – on AWS and doing the usual integrated go-to-market strategy for sales and marketing, providing channel incentives, developer training, and vertical focus on specific industries. (Oil and gas and travel and transportation were called out as part of this deal.)
We say mostly OpenShift, but other key IBM applications and tools are part of the deal between itself and AWS, including API Connect, Db2, Observability by Instana APM, Maximo Application Suite, and Security ReaQta, Later this year, Security Trusteer, Security Verify, and Watson Orchestrate, and other tools will be deliverable on AWS later this year on the AWS Marketplace. (IBM has support for the AWS platform over 30 applications and tools already, but this one-click download and install is new.) All of this IBM code will be packaged up in CloudFormation (AWS) and Terraform (independent) templates to run inside of containers. Oh, and IBM adds that it has more than 10,000 AWS certifications across its consulting business across thirteen different competencies, and is happy to help companies deploy to its rival’s cloud.
It wouldn’t be a surprise to us to hear that IBM is shopping what is still the SoftLayer cloud to someone, but we think it will be highly unlikely that AWS, Microsoft, or Google would buy it and that IBM would not be allowed to sell it to Alibaba, Baidu, or Tencent. The wonder, if our hunch – really more of a disturbance in The Force – is correct, is that the IBM Cloud was not spun out with the application hosting and outsourcing businesses with Kyndryl. Some people within Big Blue are probably still hopeful that it can hang in there, and provide its own cloud with value to IBM i and AIX shops that also need X86 iron and maybe some GPUs as well. We have always thought IBM could have done this, and thought that was the idea from the get-go nine years ago. But the fact remains that IBM cannot invest the way the Magnificent Seven clouds can, or even the way that Oracle and other lower-tier players can. It has too many other things going on.
We shall see what IBM does over the long haul.
No matter what, we think that it will stand up Power Systems iron somewhere and sell Power Virtual Servers, where it actually has differentiation. And in conjunction with subscription-based pricing for its systems, thereby bolstering its hybrid cloud message. IBM already doesn’t sell X86 servers on premises, and there is no reason to believe that it will be the preferred vendor for X86 servers in the cloud when HPE and Dell are the biggest OEMs in the enterprise today and they are selling their own machines under their respective GreenLake and APEX cloud pricing models for on-premises and co-located iron.