Mark Shuttleworth's next mission: making private clouds affordable

Canonical/Ubuntu leader wants to make private clouds as economical as public clouds

Mark Shuttleworth is a man in demand. The South African’s business successes have almost been overshadowed by his 2002 trip that made him the first person from his continent to embark on a space mission, an icebreaker voyage, his private jet, his buccaneering quotes and his high-profile public appearances.

But there’s no doubt that for all his fascinating back story and the attention he garners every time he enters the public forum, the Cape Town-born native is a bona fide successful entrepreneur. He led a successful startup as far back as the mid-1990s, just as the first dotcom boom was building, eventually selling digital certificate authority Thawte Consulting to Verisign in 1999 for $575m.

In an unlikely twist, his stake in Thawte liberated him to become a cosmonaut but also to fund another mission that led to Ubuntu, the free and open source Linux OS that lies at the heart of the OpenStack infrastructure platform and many of the largest public clouds (including those of Amazon and Microsoft) and cloud-native giants such as Netflix and Airbnb.  

Today, Shuttleworth is having his second stint as CEO of Canonical, the company that delivers, certifies and consults on Ubuntu, having returned to the leadership position last year. The firm is perhaps only second to Red Hat in terms of firms focused on the enterprise open source space. It turned over $126m for its last financial year and over the course of a 45-minute phone conversation recently, Shuttleworth took me through his plans for the company.

 

Private ambitions

But when we chatted, Shuttleworth didn’t focus on disruptive plans, high-profile rivals nor blue-sky markets. Instead, he was keen to emphasize another string to Canonical’s bow.

The premise for our interview is new analysis by 451 Research [download requiring login, or blog here] that, he argues, illustrates a significant, under-regarded change to the way organizations can economically deploy IT applications and services.

Here’s the nub of the matter: public cloud has made it more affordable and faster to deploy IT; on-premises datacenters provide control but are very complex to run; and private cloud seems to offer an excellent combination of the above but also has a reputation for being very expensive.

“The traditional legacy datacenter is enormously inefficient and that’s why CIOs have been so keen on public cloud,” Shuttleworth says. “They’re renting from people who operate in a very efficient way [with public cloud], but what would costs line up like if you have private cloud with similar levels of automation [as public cloud] from bare metal upwards, where everything is driven by policy and code and you can run it on a fully utilized basis?”

And that’s where Shuttleworth wants Canonical to come in, using its expertise and management know-how to make private clouds as efficient, or near enough, as public clouds.  

“Canonical is in an unusual position where all the lighthouse leaders operating in public cloud or companies born on the cloud take Ubuntu as standard practice. If you are going to operate a private cloud then the key is the level of automation you can operate in your own datacenter. If you’re going to build a private cloud you should get someone else to manage it.”

And that “someone else”, he hopes, naturally, will be Canonical.

In some ways, I suggest, this is the world swinging back to outsourcing but Shuttleworth demurs.

“Cloud infrastructure is quite different to the apps that run on it. The average bank has 10,000 apps and most aren’t getting changed [because] change is when things tend to go wrong. That’s the difference with outsourcing: cloud draws a very hard line between infrastructure and application; cloud creates a commodity.”

So where Canonical wants to play is where it manages infrastructure but the apps remain the domain of the owner.

 

Horses for courses

There is no suggestion that this model of private cloud, where infrastructure is managed by a specialist, becomes de facto and other approaches shuffle off. Instead, he says it’s like using a car: you might buy a car for everyday use at home but use an Uber, Zipcar or vehicle rental service when travelling on the other side of the world. It’s a horses-for-courses, competitive market but private cloud becomes a much more attractive and economical option when the management layer is in the hands of a specialist.

These changes are pretty strategic for CIOs and don’t tend to happen overnight but Shuttleworth believes that the cloud model’s success has forced IT leaders to look at architectures and not just tolerate low levels of utilization. Nor can IT justify their traditional conservatism by overstating fears of service downtime. Today, dynamic overprovisioning, sharding of workloads and other approaches have led to an environment where services can be in the cloud, highly utilized, highly automated, highly reliable and performant... without them sitting in a privately-run datacenter.

“Today, the CIO is almost the customer of IT and the CIO’s job [has changed],” he says. “Historically, IT had a monopoly on the business and the IT team had sole say in what was possible. The public cloud has effectively broken that model. The old low levels of utilization were classic evidence of inefficiency, like a local monopoly telco in the 1990s. The reality was it was [caused by] a lack of competition.”

 

Getting smarter

Now we are a little more sophisticated and decisions have become less binary and more granular, he argues. Firms are increasingly spread across multiple clouds to win negotiating power and to assuage fears of being locked in to one vendor.

“Today it’s almost swinging back to the private datacenter but also people are asking why it was so unutilized and messy,” Shuttleworth says, comparing the odds of IT departments on their own winning public cloud-like efficiency on private clouds to a consumer buying a flat-pack DIY car from Ikea and making it run.

Managing private clouds will center on BootStack, Canonical’s version of OpenStack. Of the latter, Shuttleworth says it created “a lot of buzz and then an air of dejection” at IT shops that went it alone. But if Canonical wants to make managing private clouds a significant part of its business, will that also involve adding a substantial number of people to the business and effectively entering the consulting business? Probably not.

“Operating this with human beings is not as profitable as without,” he says. Instead, he is betting on very high levels of automation with an element of deployment and consulting assistance. Often, Canonical will “prime the pump then walk away” when that’s the right thing to do, but there are customer requests to manage more infrastructure, he adds as a caveat.

 

Other markets

There is a vast opportunity to take share away from VMware and even Red Hat by offering a superior ROI. But another estimated $100m-plus opportunity for Canonical lies in putting Ubuntu at the heart of the Internet of Things as the OS in cars, drones, robots, industrial systems and the like.

If all this comes off then an IPO could come off sooner rather than later, maybe next year (my feeling, not Shuttleworth’s).

“We’ve taken a decision to head in that direction,” he says, although that decision was not without some heartache, I detect, as he recalls that Canonical started out as a “semi-philanthropic project” and is making “a big cultural change”. That change involved the downplaying of “personally loved projects”, including Ubuntu as a desktop OS to compete head-to-head with Windows, although he still thinks that eventually “client convergence will come to fruition”.

But for now, trying to ingrain the view that the private cloud can offer superior value to other deployment models is what he is at pains to get across. If Shuttleworth and others of the same persuasion are successful then we might once more see a remarkable change in the way the world makes IT work.