How OpenStack became the big dogs' game – and why it's still for you
Get rich or die tryin'
If you did have any doubts, recent events should have assuaged them, the division between the public and private cloud is here – and widening.
In what might seem to be the industry’s quickest act of consolidation and M&A in less than five years, almost all the originators of OpenStack and the early entrants are gone – snapped up by companies decades older and orders of magnitude larger.
Network giant Cisco gobbled Piston Cloud, whose brains and technology will be enrolled in Cisco’s OpenStack Private Cloud.
Piston was home to one of the brains behind the NASA Nebula Cloud Computing Platform that became the compute portion of OpenStack in 2010. Piston had offered cloudOS, a layer of software used to manage clusters of commodity, x86 servers as a single pool of compute resources. Cisco was one of those who’d poured up to $22.4m into Piston since the firm was founded, in 2011.
Also gone this week is Blue Box, on the management side of OpenStack: it sold one-click deployment. Blue Box is being added to IBM’s Bluemix platform as a service.
OpenStack and the bursts of buyouts
Nebula, like Piston, was also home to some of the talent behind OpenStack in 2010. It shut up shop a few weeks back. Oracle hired Nebula’s 40-odd staff. In quick succession, Metacloud and CloudScaling fell to Cisco and EMC last year.
Mirantis is one of the last of those first pioneers in OpenStack. The firm offers to set up and run the full OpenStack cloud using its software. Customers include DirectTV, Ericsson, NASA, NTT Docomo, Paypal and Workday.
What is going on?
Born in 2010, OpenStack was supposed to be the “Linux for the cloud.” It was a rejection of the proprietary AWS and quickly passed Eucalyptus to become the thing “everybody” could meet on. The idea was, presumably, to start a flowering of a million OpenStack public service clouds and projects.
Linux did give birth to distros and companies selling and supporting them. Today, there’s two or three successful Linux distros and fewer big companies because many of those early young hopefuls either went out of business or remained small.
The dynamic in cloud seems to be different, where demand has proved the undoing of that Linux-like dream.
The project hasn’t progressed in a controlled way, like the Linux kernel. Rather, it’s become a case of classic open source: complicated and incomplete with people focusing on the bits they felt most interested in.
This should have produced opportunity for start ups selling customer-grade product. In the end, the scale of the task was simply too great.
Nebula was founded by Chris Kemp, who at NASA was one of the creators of OpenStack. His firm got $40m from investors who included Kleiner Perkins Caufield & Buyers. In its farewell blog, Nebula summed up where we are:
The enterprise OpenStack market will take several more years to mature, they wrote. "As a venture-backed startup, we did not have the resources to wait."
And it’s the money that’s the secret to what’s really going on: despite the hype about the manifest destiny of cloud, none of those acquired were actually funded to any degree that matched. $40m in today’s money it’s a barely single round in the big data field, less than that if you’re a virtual taxi company with an app.
CloudScaling, founded in 2006, got $14m in two rounds. Remember Piston? It's co-founder Christopher McGowan was chief architect of the NASA Nebula project that became part of OpenStack and worked with Kemp. Piston got $22.4m in six rounds from seven investors. That's $3m each, split evenly.
From the start it seemed they'd be relatively funding-lite, and now – it appears – these firms have called time. Also, you’re looking at some companies that took wrong decisions: Nebula sold an OpenStack cloud appliance, but selling hardware is difficult. Costly and risky for the maker, as customers take some convincing.
None of the enterprises I speak to are interested in how a cloud is made. They are interested in backing a strong and reliable product that comes at an affordable price. Today, that’s AWS with Microsoft’s Azure coming a relatively remote second.
VCs fund a winner, or – possibly – a close number two. Neither position is open to any of those backing the OpenStack horse against AWS or Microsoft. But that doesn’t mean there isn’t a place for OpenStack: it’s just not as a mass-market alternative to AWS.
Its real future is as a private cloud.
That’s something which Oracle, Cisco and IBM clearly believe in. Businesses are interested in the cost savings and flexibility that cloud offers, but not all of them are as keen to heft their applications and data on to AWS as fans of public cloud insist. And it’s the enterprise where Oracle, Cisco, IBM and others play.
And while the exits of Piston, Nebula, Blue Box, Metacloud and CloudScaling concentrated the mind around thoughts of consolidation and shake out, there remains plenty of other smaller firms selling OpenStack products and services.
Piston’s Josh McKenty, a early mover in OpenStack, recently bemoaned OpenStack has lost its heart: the project was supposed to be a community driven open-source affair, like Linux, but has since become full of IT vendors trying to get rich.
But making a buck from OpenStack will be precisely what drive its adoption in the enterprise and also what helps improve it. An important part of what’s been holding back OpenStack is the fact it’s difficult to build and run; customers lack the skills and technologies to overcome this. By absorbing early OpenStack firms the giants are in a position to do what they do best: to “productize”.
The last few days confirm OpenStack’s position as a foundation for private clouds, either hosting corporate services or by providing a kind of super-elastic middleware.
This future is not about acting as service provider to others; rather, it's about serving as a new way of wrangling servers, storage and compute for your own internal users and apps, with the data centre acting as a super operating system.
The only challenge now is to ensure the money men’s drive to productise doesn’t produce code schisms and forks that turn OpenStack into the next Unix. ®