Building Things That People Like

In the last fortnight, I’ve had the opportunity to read two very contrasting articles on the Internet. One from The Economist on a humbler Microsoft, and another from an EMC blogger Chuck Hollis arguing exactly the opposite on why a single stack matters more than anything else. The platform “hubris” is fading at Microsoft, and the (disruptive) abandonment of the strategy tax is proving regenerative for the company. In contrast, the EMC Federation, led by blogger strategists such as Chuck, is thumping the table on why customers and partners owe them the strategy tax for an omnipotent platform.

Excerpts of The Economist article:

IBM in the 80’s

In the mid-80s, IBM had a similar hubris about its “platform” and “account control.” Media from that time called it game-over for Oracle and others when IBM DB2 entered the relational database market. The way Oracle responded was by acting as the antithesis of a platform — by running its software on all OSes– no different than the Microsoft of today.

How Companies Strive… and Thrive

What is clear from the Microsoft and Oracle examples is that companies thrive when they don’t chest-thump with a single stack story. They thrive when they build apps that run everywhere. They thrive when they fuel an entire ecosystem of apps built by 3rd parties. They thrive when the network effects of these 3rd party apps fuel a virtuous cycle of innovation on their substrate. Like what the App Store is to Apple — notwithstanding a mediocre Apple Maps that rarely brings the same delight as a more ubiquitous Google Maps. The ability to compete honestly with 3rd party apps is what makes Apple successful.

And Now on Chuck’s Arrogance

There are some interesting arguments in Chuck’s blog, but the one that is truly inane (#5) is around Support and OEM. His claim is that because Nutanix does not OEM VMware ESXi software, customers suffer due to lack of high quality support. Nothing could be further from the truth. Our NPS scores are at a mind-boggling 89, most of which is because we help customers end-to-end, without worrying about whose problem it really is. With convergence of technologies comes a natural convergence of responsibilities around support.

If OEM were to be the only way to guarantee support, vCenter and vCOPS should be OEM’ing an Oracle or a SQL Server database with them. Or else, by his reasoning, support is compromised. Google Maps should then OEM Apple iOS, or else support is second rate. Microsoft Office should OEM Mac OS, or else support is suboptimal. Or for that matter, any application running as a virtualized workload should OEM ESXi, because support will otherwise be compromised.

We’re Just a Freakin’ App on a Hypervisor

Nutanix is simply trying to argue that storage is simply an app on the hypervisor. And just like there is no expectation from an app vendor to OEM ESXi (simply because they are running in a virtualized environment), there should be no expectation for us to OEM any hypervisor as part of our offering.

Why not OEM the hypervisor?

  • Because we wanted the customer to choose which hypervisor they wanted for their run time. There was no point hardcoding a hypervisor from the factory when it could be installed in minutes at a customer site. We’ve built some beautiful tools for that kind of bare-metal automation.
  • Because we’d then be competing with the Channel. Resellers make some margin on selling VMware ESXi in the retail market. There was no point us making margin on a product that is already difficult for the channel to make good money on. And if both us and the channel made money distributing that software, it would end up becoming more expensive for the customer. It just didn’t make any economic sense for the channel or the end customer.
  • Because we’re working on some heterogeneous hyperconverged use cases in which single environments could have multiple hypervisors! There are some powerful use cases that we will unveil shortly. Hyperconvergence will need to have the ability to mix-and-match hypervisors — just like 3-tier deployments — or else they quickly become computing islands.

Our True North

Our NPS scores have gone up in the last 18 months while we’ve grown our customer base at a rapid pace. One of the reasons why our customers love us is because we treat all problems related to a converged infrastructure — compute, storage, networking, applications — as our own. And we do that without charging a dime for things that go beyond our product. More importantly, we do it for every hypervisor, every application, and every switch that connects to us. And for that, we don’t need to OEM any hypervisor, application, or switch with our product. The customer binds us all together, our varied backgrounds notwithstanding. We all work towards a common good, i.e., customer delight. Nothing else matters. The customer is all powerful.

We build stuff that people like. We do things that customers love.