VMware and AWS -- Harder Than It Looks

One of the bigger pieces of industry news last week was VMware announcing they intend to eventually offer an infrastructure service built on top of a new AWS bare metal offering.

My corner of the internet broke for a while as a result. Before long, I was inevitably getting pinged by my Twitter brethren for a personal view.

I spent a few days thinking about it from various angles.  Sorry to say, I come away with more negatives than positives.

Fair warning: I left EMC and VMware because I didn't see a viable cloud strategy. I joined Oracle on the strength of their cloud strategy. So far, it's played out exactly as I thought it would.  

Not much has changed with my perspective since then.  Including this latest press release.

What A Long, Strange Trip It's Been

The story of VMware trying to come up with a viable public cloud strategy is a long and torturous one.

vCloud Director. vCHS (also known as vCheese). vCloud Air 1. vCloud Air 2. EMC buying Virtuastream and mistakenly positioning it as "the answer", then backing down. The IBM deal. The loose affiliation of the vCloud Air Network.

And now this. Forgive me if I come across as just a teeny bit skeptical based on the story so far.

At a macro level, what we're seeing is obvious. The world is moving to a public cloud model. VMware's core product -- vSphere -- wasn't designed for public clouds. VMware -- the company -- is not structured as a public cloud provider. Hence the history of thrashing around, trying to come up with some sort of viable answer.

There are certainly a lot of heavy vSphere users that are hoping (praying?) that VMware can deliver a workable solution. So far, no dice.

I'll remind people: hope is not a strategy.

The Big Positives

The most positive thing here are the industry optics.

VMware, the king of data center virtualization, inks an agreement with AWS, the king of public cloud infrastructure.

VMware gets the potential of a credible public cloud option, which it doesn't have today, AWS gets a potential entry into the enterprise IT world which it doesn't have today.

Everyone wins, birds sing and all is right with the world. Or is it?

The Big Negatives

First, there's the vague delivery schedule.

Sometime next year, we are told, there will be something to look at. Having been a behind-the-scenes veteran of multiple industry announcements, long and vague schedules usually mean significant amounts of work, technical risk, or -- usually -- both.

If they had something to show that would help establish credibility, they'd already be showing it. Trust me.

While we're waiting, we'll have to be content with powerpoint slides and optimistic blog posts.

Second, there's a vexing question.

VMware couldn't offer a viable cloud service when they were building infrastructure to their specifications (vCloud Air). What makes us think they can do better when running on someone else's infrastructure? Yes, they'll be using a new bare-metal service from AWS, but the devil is in the details.

And we won't know the details for quite a while.

Finally, there's time lost.

I know that there will be more than a few VMware faithful who will take this announcement to their leadership team and say "see, we have a plan for cloud now!". Maybe they do, maybe they don't, but it will take a year to figure out either way.

Meanwhile, their progress forward will be stalled due to a "market freeze" (e.g. wait, don't do anything, there's something big coming!). And that's unfortunate.

The Strategic View

VMware intends to offer this as their premiere cloud offering: sold and supported by VMware. Which means, of course, all the other infrastructure and IaaS vendors can move to the back of the line. Not exactly the most partner-friendly move. I can only imagine the conversations.

And VMware's current business model is heavily dependent on those partners today.

The intent of the service appears to precisely replicate on-premises VMware functionality. There's no new significant functionality being discussed as part of the service.

The problem? One of the big motivations to move to a public cloud is easy access to new functionality. That doesn't seem to be the case here. All we get is a new way to consume familiar VMs.

I feel a bit shortchanged. That means it's going to be all about the economics of the new service compared to doing things on-premises.

Amazon leaves a bit on the table. They've been investing in various software services (e.g. Redshift et. al.) in an attempt to differentiate from simple IaaS. None of that stuff runs in vSphere, nor has there ever been an inking that AWS would ever invest in doing that.

And, of course, there are opposing forces at work.  VMware wants the world to use their stack, AWS wants the same for their technology.  Not exactly what I would call a long-term stable situation.

I wonder how long before AWS offers yet another "migrate your vSphere VMs to native AWS" promotion?

(update October 25th: not too long, it seems.)

VMware will have to teach and incentivize their salespeople to sell subscriptions and metered consumption vs. traditional software licenses. Not impossible, but not easy either. The new "market freeze" will also apply to their traditional on-premises license business, and that's going to show up in the revenue line before long.

Additionally, if VMware fails to make the new service financially attractive to their existing partners (who incidentally sell boatloads of on-premises hardware), that's going to completely stall the venture. Not a lot of margin to work with, either.

That ain't FUD folks, that's how the world works.

From the technology side, more concerns. Hypervisors have already become largely commoditized. Containers and container management is now a real thing, making vSphere's relevance in the cloud less assured. Proven solutions already exist today to encapsulate those VMs, strip out vSphere and move them unmodified to the public cloud of your choice, and saving big money in the process.

In addition to all the other challenges, VMware has fight a completely separate battle on maintaining their technology relevance in a world of public cloud.

Remember, we talking about easily commoditized IaaS here. Not PaaS. Not SaaS.

A Final Thought, Or Two

Yes, this was certainly a big bold move on the part of VMware. The world is quickly moving to cloud. They now know that delivering a viable public cloud service isn't as easy as it looks. So they are partnering their way in, and have landed a big one with AWS.  All credit where credit is due.

The harder part over the next year is making it work: making it work from a technology perspective, and making it work from a go-to-market perspective. And not cratering the business that pays the bills in the process.

None of this is obvious to me.

I'll check back in a year, and let you know what I think.  

----------------------------------------------------

Like this blog post?  Why not subscribe?

 

Comments

[Dan Fisher]

Interesting post and as usual I love the insights. However, I think your argument is based on a false assumption. VMware is not attempting to offer a Public Cloud platform, instead they are extending their Private Cloud platform to Public IaaS providers. Although VMware made the announcement, the real change is AWS is offering to be Hosted Private Cloud platform in addition to being a Public Cloud platform. That's big news. First IBM, now AWS. I suspect Azure is right around the corner.

Why is this significant? Well, as you've posted in the past, there is a difference between Public and Private Clouds. Chiefly, in how they are built and maintained. Both offer the three hallmarks of a cloud: self-service catalog; flexible, low maintenance infrastructure; and measures and metrics for cloud management. In Private Clouds you have to build all three. Select the infrastructure components, design and build your service catalog, program and orchestrate the services, and set up the measures and metrics to ensure all goes smoothly. You’ve also blogged that this approach is fraught with problems and pitfalls that usually leads to failure. No argument. Building a Private Cloud is not for the faint of heart.

Public Clouds are much simpler. All of the work is done for you so you just have to select the services you want and off you go. The down side is you pay more for the pre-built services.

Hosted Private Clouds are somewhere in the middle. In a Hosted Private Cloud, the infrastructure is already selected and built for you. So are the measures and metrics. That leaves just the service catalog to contend with. This is a much simpler endeavor than trying to do all of it yourself. Still not easy, but simpler.

The VMware announcement tells me one now has choice in building a Private Cloud. Through this partnership with AWS, one can extend their Private Cloud into a Hosted platform and accelerate their service catalog development. It does not invalidate the partnerships with on-premises infrastructure providers, but it does highlight their ‘commodity’ nature. I do agree with you that AWS sees a possible windfall as Private Cloud owners may not only take advantage of the their hosting services, but may migrate to full Public Cloud if the service catalog becomes unmanageable and would opt for something pre-built and tested – and are willing to pay for the privilege.

[Sukh Grewal]

EMC and its customers: (1) For the largest data centers that were within reach of becoming a cloud, EMC did not give them the tools they needed to grow into a real cloud (2) For the smaller data centers, it kept the VMWare faithful from making the move to the real cloud. Imagine a small town with its twenty server VMWare installation.
Another issue, at this point less impactful,has been EMC storage. EMC did not provide the efficiencies needed to compete with the cloud. They were often late with offerings, for example to counter NetApps when it began over a decade ago. The cost difference was not 10 or 20% - it was 100 and 200%.
Unfortunately, those who bought the full menu from EMC have paid the price. The idea that they will offer something on AWS may be the transition that may work for their faithful and extend the revenue stream with the believers - but it is unlikely to expand their customer base.
As a long time AWS customer (going on six years), it is difficult to imagine EMC getting anything more than crumbs from this deal. At this time AWS is firing on all cylinders and is a talent magnet to boot.
Chuck leaving EMC was more ominous than was previously believed.

[MDLptown]

No AWS SW goodness?
What is so special about providing infrastructure services with legacy VMW SW on top?

[DCLacoste]

I agree that this announcement is mostly about optics, but also agree with Dans comment: VMwares Cloud strategy is not to compete with AWS for public cloud, it is to stretch their private cloud to leverage commodity hosting resources. We'll see how well they manage the services there, or if the adoption rate even supports the service.

[Michael Endrizzi]

Agree that there is currently a lot of smoke. Agree that VMware's NSX solution has a history of unstable releases that further invalidates their credibility to deliver. But on paper this is the solution that will enable a seamless private-to-public cloud transition because the back-end infrastructure is already in place to support the applications.

[Peter Wilson]

Great article, but you'd have to say Oracle isn't any better. Heck, I'd give IBM more credit with Bluemix/Softlayer and more of their products being offered up as SaaS