VMware’s Cloud Adventure

I like VMware. They’re a solid company with lots of good people (With the exception of whoever is responsible for product names – VMware vCloud Air Virtual Private Cloud OnDemand? Seriously, what is wrong with you?) and tech.

I’ve been using their products for fifteen years and still remember how magical it felt the first time I loaded VMware Workstation and had Windows running inside of Windows. I remember calling someone over to my desk and telling them “Look how cool this is.” I also remember them saying “I guess, but what would you ever use that for?”

Today, you’d be hard pressed to find a corporate datacenter that isn’t running some piece of VMware tech. They took virtualization mainstream, and built the foundation of the cloud tech that everyone is using today.

Misdirected myopia

Unfortunately, the world that they built is now eating them. Hypervisors became commodity, where “good enough” is an acceptable target. Hyper-V, Xen, KVM: they all became good enough for ecosystems to be built around them, followed by orchestration and the “cloud”.

VMware seemed completely oblivious to the scale and pace of what was happening. They sat on the sidelines while the world they helped build was being transformed. Maybe the thought was “This cloud stuff is for startups who
will grow into our enterprise products once they get tired of playing with toys.” or “Enterprise customers will move to the cloud… in twenty years.”

Sort of like Steve Ballmer’s “There’s no chance that the iPhone is going to get any significant market share.”

Too late, too little

Amazon launched AWS in 2006. Microsoft launched Azure in 2010. VMware vCloud Hybrid Service (now vCloud Air), didn’t launch until 2013 – seven years after AWS, fourty-nine years in dog/tech years.

Given their late entry, they took some shortcuts. Instead of building a native cloud platform from scratch, they pushed their on-premise, datacenter products into the cloud. That sort of works, but only at a very base level. It is heavily reliant on legacy VMware tools for management and orchestration. It doesn’t scale. It requires specific web browsers and plugins.

It isn’t cloud. It looks a lot like what would happen if a VP went to a tech conference, then came back to the office and told their engineers “Everyone has a cloud. We need a cloud. BUILD ME A CLOUD!”

Pointy-haired boss is the reason we can’t have nice things.

But it started getting better in hiccuping bursts. They added DR options and Database-as-a-Service. They stripped out a truly terrible VM backup solution. The people who’ve worked on vCloud Air should be proud of that.

Cutting losses

Then during yesterday’s VMware earnings call:

“CEO Pat Gelsinger said the service (vCloud Air) will have a “narrower focus” going forward and that money invested in it already is considered “adequate for our needs.”
In other words: “vCloud Air is dead.”

What we’ll probably see
vCloud Air might have been able to carve out a significant niche if it had been:

  • a) built as a cloud native platform or
  • b) given more focus.

But that didn’t happen, so it’s probably OK for it to die.

The biggest use case that VMware seemed to be addressing was DR – the “we need somewhere else for this stuff to go” problem. vCloud Air, especially under its original “Hybrid Cloud” moniker, mostly existed as a remote target for the tools you were already running onsite.

So we’ll probably see VMware do two things that they’ve been alluding to:

  1. Expand options for leveraging third-party clouds like AWS, Azure, and Google from within the vSphere and vCloud Director toolsets – for burst and DR.
  2. Refactor existing solutions like ESXi and Horizon to work on third-party clouds. They’re already doing this successfully with their NSX virtual networking product.

Both of these paths help VMware customers get into the cloud, but I’m not sure they’re good for VMware in the long-term. Once a customer is running in the cloud they’ll start asking “We’re here now. Do we really need these legacy toolsets anymore?” In many cases the answer will be “no”.

There’s a hybrid datacenter->public cloud model that VMware can succeed within, but that market will only get smaller as businesses replace applications and go directly to the cloud.

What do?

The onus is on VMware to make their tools more flexible, powerful, and compelling than the tools that AWS and others are building in-platform. If they can narrow and re-direct their focus on future needs (where the puck is headed), and think about cloud in a broader, multi-vendor context they might be able to.

They’ve already started that with containers. Docker orchestration platforms like Kubernetes and Mesos are still rough around the edges and there is room in the space for VMware to get in and leverage the benefit of their size and engineering bench.

Server-less code platforms like AWS Lambda will compete with container-based workloads. I hesitate to say that VMware needs to be in that space too, but it would be worthwhile to consider that containers are not the only future. Going 100% in on containers seems prone to getting stuck in the rut of “follower”, as Microsoft has been until their recent turnaround. Envisioning and pursuing alternate futures is how companies lead and innovate, but requires having leaders who are comfortable with false-starts and failure – or deep pockets and lots of acquisitions(meh).

VMware is doing some great things with end-user compute(EUC) and is starting down the path of meaningful integration between Horizon, Airwatch, and NSX in the datacenter. The story they tell about EUC is compelling and coherent (unlike their cloud narrative). If they can pull off their plans for end-to-end security and consistent access to corporate apps and data across all devices, they will likely trounce Microsoft and others competitors in that space.

They have the pieces and the people – they just need a clear direction and internal consensus to move forward.

Image Credit: Zooey