How to Disrupt IBM, Oracle, and VMware: The CloudBees Story
Sacha Labourey was one of the early skeptics when it came to cloud computing.
“For me, initially, the cloud was not that interesting. It was actually pretty boring,” he says. That’s in part because he was the chief technology officer of middleware firm JBoss, which in 2006 was bought by open source software giant Red Hat, where “everything was the server, that’s how we live and think,” he says. “That was the church there.”
Since then, cloud computing has gone from a lot of hype to reality. Much of that reality is around what startups and big companies alike are doing to migrate their software applications and data from their own servers to cloud-based systems. Indeed, Labourey’s latest startup, Woburn, MA-based CloudBees, is a testament to how far cloud computing has come—and how some big themes in information technology are playing out.
CloudBees also tells an intriguing (if early) story of a young company looking to disrupt the business of entrenched giants in its field, like Oracle, IBM, VMware, and, yes, Red Hat. Along the way, Labourey, who is CloudBees’ founder and CEO, has learned some important lessons about the evolution of the cloud, the strengths and weaknesses of open source software, the relevance of the Java programming language (versus newer languages like Ruby), and what he calls the “industrialization of IT.”
First, some back-story. Labourey (see photo, left) is a widely respected techie who is based in his native Switzerland and runs a distributed team around the world. He may have been skeptical of the cloud, but by the time he left Red Hat in 2009, the notion of cloud computing as on-demand, remote-access, pay-as-you-go storage and processing power was well established by pioneers like Amazon.com. “I started to think, how can we make something that has those qualities, but is focused on applications?” he says.
That was the genesis of CloudBees, which offers what it calls “platform as a service.” It is a cloud-computing environment where Java developers can build, store, test, and deploy their code, from start to finish. So instead of writing code and then having to get an IT department involved to secure enough machines to do large-scale testing and code management, for example, CloudBees gives companies an all-in-one system to do that. “Platform as a service is a way to simplify all of this and put developers back in charge of their applications,” Labourey says. “They become more efficient, and agility is much better.”
In case you’re wondering why bother—since Web developers these days are already doing everything in the cloud (using platforms like Heroku)—the key here is Java. Java systems pre-date cloud computing. Java runs most big companies’ database and back-end systems. In short, Java still rules—at least in enterprise software.
“All of the cool kids are trying Ruby,” Labourey says. “During the last five to seven years, you were kind of ashamed of it if you were a Java developer. You were not the cool guy anymore. You used to be cool when you were 25 at the university, and now you’re 35, and slightly fat. You make jokes but nobody laughs. That’s a Java developer.”
Yet the long-term future of Java is bright, he says. Ten to 15 years from now, Labourey sees Java-based systems “being the foundation for more and more deployments, and a flurry of languages popping up left and right on top of it.”
For now, CloudBees has grown to about 30 employees, with about a half-dozen staff based in Woburn (mostly in sales and marketing). The rest are distributed worldwide. The company got started in early 2010 and has raised $14.5 million from Matrix Partners, Lightspeed Venture Partners, and private investors including JBoss founder Marc Fleury. Its customers include Amdocs, Bullhorn, Cisco, Digg, LoseIt, Netflix, Raytheon, and Symantec.
So where does this startup fit in the grand scheme of cloud computing? If you think of the cloud as three layers, Labourey says, CloudBees sits in the middle. The bottom layer is infrastructure as a service—like Amazon’s building blocks of storage and processing. The top layer is software as a service, which has become the dominant model for software offerings (one pioneer being Salesforce.com). In between is platform as a service—which, Labourey says, to date has been focused mainly on running applications, not developing them too, as CloudBees is trying to do.
“The net effect is the developer is going to be able to supercharge the business, because they’re going to be able to create these new innovative applications layered on top of the old legacy systems,” says André Pino, who heads up CloudBees marketing.
And that brings us to the disruption that CloudBees is looking to wreak. “Delivering [platform as a service] is going to kill the existing packaged middleware business,” says Steve Harris, senior vice president of products for CloudBees. “It’s certainly a huge disruption from a business standpoint. Instead of people selling you licenses, they sell you use of their service through our platform.” (Meaning that companies are moving away from buying licenses for software built around servers and operating systems and towards a model where they buy just what they need via the Internet.)
Harris knows all about entrenched middleware vendors, being a 13-year veteran of tech giant Oracle (NASDAQ: ORCL). But it’s not just Oracle, which sells its WebLogic server software for developing cloud applications. It’s also IBM (NYSE: IBM), which has a competing product called WebSphere, and VMware’s (NYSE: VMW) Spring framework. And, of course, Red Hat (NYSE: RHT) is in CloudBees’ sights as well, with its JBoss application software. (Any entrepreneur worth his salt wants to compete with the company that bought his last startup, right?)
The big boys would probably brush off CloudBees as just another buzzing pest in their ear. Nevertheless, the two-year-old startup has reason to be hopeful, says Bob Bickel, a former JBoss and Hewlett-Packard exec who serves as an advisor to CloudBees. “From a middleware business perspective, we think the cloud brings a lot more change than open source did, in terms of lower cost.”
Take VMware’s business. “They sold $2 billion last year,” Bickel says. “That means this year they’ve got to sell $3 billion. They’re not going to be doing that by making things cheaper with their licenses. Their first generation of virtualization starts to get eclipsed by cloud virtualization. They’re going to do great and all that—they’re better prepared than your typical middleware vendors—but even a VMware is going to have trouble shifting from this licensed enterprise software model, which will get blown up over time because companies won’t be able to afford it anymore.”
It’s too early to say whether CloudBees itself will disrupt the industry, but the trend toward cloud-based services is clearly affecting the big players. Yet something else Bickel said also piqued my interest: the bit about open source’s business impact as compared to the cloud’s. After all, CloudBees’ software is not open source, even though its founding DNA comes from that movement.
That decision comes down to Labourey’s experience with JBoss and Red Hat. “I was so convinced that open source would be amazing. In some ways it is. The cloud only exists because open source existed. But I thought it would be a bigger business, be much more influential,” he says. “There’s only one public company, Red Hat. And VMware is much bigger. I think [open source] is a very hard business model.” (Companies working with OpenStack, an open source system for configuring corporate data centers as clouds, might have something to say about that. See, for example, Nebula.)
And while Labourey now sees the limitations of open source, he seems to have come around fully to the promise of cloud computing—which he places in historical context as the industrialization of IT. “It’s exactly the same process as what happened in 1890 with industrialization of electricity,” he says. “Initially you had electricity produced in factories. They were private, very costly, hard to maintain. Then you had community-owned factories.”
He recalls a map of Paris from 1910 in which different regions used different power standards for voltage and current. “It was IT as we know it today. You have an HP box with Windows, and oh, sorry—you have equipment you want to plug in, and it won’t work,” he says. “Now you have a standardized plug, you will get current, and it will always work. The new plug is HTML, the Internet is the new grid, and the producing entities are in the cloud.”