VMworld 2021: A Strategic Overview: Multi-Cloud All The Things

VMware has chosen multi-cloud as the central core of its strategy, and it is using a lot of confusing all-things-to-all-people messaging in trying to sell this strategy. Looking past the rhetoric, VMware is positioning itself as the pathway to change for organisations that need both cloud and more traditional ways of working, but don’t want to build everything from scratch themselves.

Which is most of the market.

Customers have a lot of existing, successful, and important workloads running in virtual machines that used to run on bare metal and took a couple of decades to move into VMs. VMware is betting that moving all those workloads to a cloud operating model won’t happen quickly either.

However, customers do want to deploy in a cloud-like way, and VMware needs to remain relevant as the industry gradually changes. VMware wasn’t successful at building its own public cloud offering (recall the underwhelming vCloud Air that was sold off to OVH, and chose to partner rather than compete with the public clouds. This has paid off well.

Now VMware wants to be the engine of what it is calling Sovereign Clouds. This is an attempt to position VMware as the Everything Anywhere option.

VMware is essentially embracing Kubernetes as the workload orchestrator, and absorbing it into its existing VMware-style workload management suite. By exposing a Kubernetes-style interface to developers, while keeping the existing tools and processes familiar to infrastructure groups, VMware hopes to be acceptable for software developers and IT operations teams.

This dual appeal is at the heart of what VMware is trying to do. It isn’t so much a DevOps story as a Dev AND Ops story, with something to appeal to everyone, wherever they might be.

Key Points

• VMware wants to be available everywhere customers might want them, and that means multi-cloud and appealing to both developers and operators.
• Choosing VMware, ironically, means compromising between the new cloud way of doing things and the existing VMware on-site way of doing things.
• VMware is a hedge, and this straddling can’t continue forever. Either customers will adopt cloud-style operations, or they won’t. The important factor is how quickly change will actually happen.

VMware has been trying to appeal to developers for a long time, while remaining an inherently infrastructure company, and developers simply do not want to care about infrastructure. They use it, but they don’t want to know much about it. They want to deal with purely software abstractions, and the cloud has helped them to do that in ways that traditional infrastructure, even virtualised, failed to do.

Making the infrastructure disappear beneath easy-to-use abstractions is what developers actually want, and PivotNine believes VMware is failing to fully embrace this idea, at least for now.

All Things To All People

VMware is trying to appeal to traditional infrastructure teams who actually buy and run the infrastructure products that VMware sells, while simultaneously trying to appeal to the developers that are closer to the business units that have new budget to do new and interesting things. These two groups are fundamentally different, no matter what the DevOps enthusiasts claim.

By trying to be all things to all people, VMware is essentially delaying making any decisions on what to abandon until customers tell it what to do. This is the AWS approach, and we are seeing evidence of an increased AWS-ification of how VMware does things this year. VMware has announced a bewildering array of new products with evocative names that have little-to-no relation to what the products actually do.

VMware has simultaneously criticised the open source Kubernetes ecosystem for being too complex and having too much choice, while also adding lots of new options in its attempt to offer whatever customers might want. Throwing a lot of things at the wall to see what sticks could be a bold portfolio-theory approach or a complete lack of coherent strategy.

Growth From Existing Customers

We think VMware is copying Amazon’s Everything Store approach where VMware will provide at least a basic option for anything a customer might want so that they never shop anywhere else.

While this could be characterised as a defensive, stop-loss approach, we think it’s smarter than that. Most organisations are not new. Enterprises, the core customer base of VMware, already do a lot of successful things that they need to keep doing, even as they do new things that may ultimately replace the old things.

VMware needs to remain relevant enough so that it doesn’t become the bare-metal of the virtualisation era: the thing that got replaced over time. Of course, bare metal is still there; virtual machines run on physical infrastructure. But the hardware works in service to the software that runs on it, which became VMware. Now the hypervisor is disappearing beneath a new layer of abstraction: cloud, and Kubernetes.

The best way to stay relevant is for VMware to focus on the operations of infrastructure. It’s at the core of what VMware really is: a way to coordinate and operate infrastructure in a way that pretty much the whole industry understands. People have forgotten what a fragmented, incompatible world we had (especially in storage) before VMware managed to provide neutral ground for vendors to meet. VMware became important enough that other vendors has no choice.

Now Kubernetes is providing a new meeting ground, and deliberately so. We spoke numerous times with the late Dan Kohn about how this was the plan all along: Kubernetes was supposed to to provide a foundation for an ecosystem to build upon where there was plenty of money to go around. Kubernetes has now reached the scale of influence that it is, essentially, the new VMware.

VMware As Liminal Choice

VMware represents one pathway from an existing way of operating IT infrastructure and a new way. It is not the only path, but VMware is keen to exploit its existing relationship with customers and convince them that, essentially, better the devil you know than to rebuild everything in newfangled clouds.

This idea is not without merit.

The idea of bi-modal IT has thankfully gone away, but it’s been replaced by new buzzwords like DevOps that attempt to merge incompatible concepts. Developers do not operate infrastructure, and operators don’t write applications, not at scale, because specialisation and division of labour is what industrialised processes are all about. DevOps is about replacing warring factions with mutual understanding and empathy between groups of people who have different focuses and specialities.

Doing two different things at once is a transition phase between one and the other, and change is hard. VMware wants to make that change easier for the vast majority of organisations that are not all-in in one cloud and are unlikely ever to be, partly because no one really agrees what multi-cloud actually means.[⁵][2]

For VMware to pull this off it has no choice but to appeal to both the new world and the old world, but the appeal to developers is mostly about appeasement, not attraction. The real market is infrastructure operations teams, whatever name they go by.[⁶][3]

If you already have a substantial investment in VMware, it’s safe to stick with it. The challenge comes from what to do about the new workloads that developers want to build using public cloud primitives. Do you build them on top of VMware, or build them onto what will eventually replace VMware in your environment?

The decision comes down to how you want to operate your IT, and over what time period you want change to happen. If you want to rapidly move to a cloud-like operating model, you can be less concerned with how VMware plans to do things. Most organisations are not in this position and can afford to take their time.

The operations of Kubernetes is still quite immature when compared to how enterprise infrastructure is operated. Adopting it quickly will mean rapid change now, and for the foreseeable future, as you realign your processes to how things eventually shake out in the broader ecosystem. This is not for the faint of heart, and the risk requires an outsize benefit to justify it.

You will need to look at the people and processes in your organisation, not the technology, because the core of what VMware is offering is an operational model. Do you want to pick a few cloud or SaaS options and specialise in those tools for your business, or do you want to choose VMware to provide an abstraction layer to all possible choices?

PivotNine recommends approaching this challenge by asking:

1. How quickly do I need to change from what I’m already doing?
2. How well do I understand where I need to go? Will I need flexibility to change my mind, or am I certain I know what I need to do for the next 5-10 years?
3. What do my people think about the alternate ways of working this represents? Will they change and adapt quickly enough, or do I need new people?
4. How confident am I in guiding organisational change? What evidence do I have to support my confidence?

When you have good answers to these questions, you will be ready to start looking at what makes a good technology choice that will match your desired operating model.

1. Simon Sharwood, ‘There’s No Shame in VMware Quitting the Public Cloud’ <https://www.theregister.com/2017/04/05/vmware_sells_vcloud_air_to_ovh_analysis/>.↩
2. The term Sovereign Clouds is clumsy nationalistic rhetoric that echoes geopolitical tensions, but we won’t discuss that further in this note.↩
3. Brad Stone, The Everything Store: Jeff Bezos and the Age of Amazon (Back Bay Books/Little, Brown; Company, First Back Bay paperback edition, 2014) (‘The Everything Store’).↩
4. Abraham Lincoln, A House Divided, Speech, 16 June 1858.↩
5. We say multi-cloud is simply the use of more than one cloud, which happens if you use Microsoft365, Salesforce and ServiceNOW at the same time and never touch Infrastructure-as-a-Service from AWS directly.↩
6. IT Operations, Infrastructure Operations, SRE, or DevOps, it’s the people who build, operate, and govern the infrastructure that supports everything else.↩