OpenStack, Open Source

Angel Diaz, Vice President of Cloud Architecture and Technology at IBM, recalls the meeting in which he convinced CEO Sam Palmisano that IBM should support OpenStack. “It didn’t make sense to make another open source competitor,” he says. “That does no good to anybody. And Sam looked at me and said, ‘That’s great. Make it bigger than Linux’.”

Several years later, OpenStack may have already reached Palmisano’s objective. The OpenStack Foundation now includes 800 corporations and organizations and 2,000 code contributors. As Alan Clark, a director at SUSE who currently chairs the OpenStack Foundation board, points out, these numbers are comparable to those of the Linux kernel, prompting Clark to call OpenStack “The Linux of the Cloud.”

However, the comparison goes deeper than numbers. Foundation leaders attribute OpenStack’s success directly to Linux and open source software and their example. “Technological innovation, especially in the last twenty or thirty years, has been driven by the innovation of communities,” says Diaz. “Linux was a great example of that.” In fact, free software is viewed as so central to the continuing success of OpenStack that Diaz adds matter-of-factly, “If you’re not building your cloud so that it’s open by design, you’re building a dead-end cloud. You cannot move at the pace that you have to in order to be in the market.”

Choosing Open Source

From an organizational perspective, open source was a logical choice for OpenStack. Referring to the 2001 campaign that announced IBM’s involvement with open source, Diaz jokes that the current approach was chosen because of “peace, love, and happiness.” However, the deeper reasons, he quickly adds, were practical ones.

According to Diaz, even in the early days of the Foundation, it was obviously going to be an alliance of different interests. “We wanted to move OpenStack from a single vendor or a small number of vendors, into an open meritocracy where everyone could contribute, where you could have an ecosystem of vendors, and where the community goes is based on the technical community, not on the will of a single entity. What we wanted to do was build a cloud infrastructure that wold be the foundation of the innovation that is occurring now and that will certain occur in the future.” Examples of such communities had already proved their effectiveness in such projects as the Linux kernel, the Apache Foundation, and the Eclipse Foundation, so following their example was a natural choice.

Such a model was easy to adopt because, unlike the situation at the turn of the millennium, when open source was a new concept, it had proved itself in the years since to be a useful tactic in the main priorities of business. As Clark points out, “Companies have a high pressure to innovate, and at the same time they have high pressure to control their costs. Plus they need to get to market faster. So they’ve got these pressures, and are finding that they can go out and collaborate and reduce those pressures. At the same time, all the innovation is happening in open source. If you want to keep up with all the latest initiatives, they’re happening in open source.” Clark describes the situation as “a perfect storm,” in which all the motivations and needs are answered by open source.

Getting Buy-In

For many companies, such as IBM, SUSE, and Ubuntu, open source expertise was easily transferable to OpenStack. For instance, Peter Chadwick, Senior Product Manager Cloud Infrastructure at SUSE, says that, having decided that OpenStack already had the most developed ecosystem among cloud platforms, “what we wanted to understand was how SUSE could best bring its open source experience from Linux to making things better for our customers in OpenStack.” This orientation helped SUSE choose its initial contributions: support for Xen hypervisors and adding awareness of hypervisors to the OpenStack scheduler. In the same way, Ubuntu’s Canonical Software could leverage its open source expertise to become a leader in the OpenStack market.

Other Openstack members, though, have been slower to accept open source fully. Part of their reluctance originated with copyleft licenses, such as the GNU General Public License, which generally require that any open source code retains its license. This reluctance was overcome by the decision to use the more permissive Apache 2.0 license, which makes easier the use of open source code alongside proprietary code. Peter Chadwick, Senior Product Manager, Cloud Infrastructure, at SUSE, suggests that, “settling on Apache was the right thing to do because it was easier to get buy-in from the proprietary vendors, because the boundaries are a little more flexible.”

Still, even with the concession for the license, traditional methods of doing business die hard. Clark notes that some members of the OpenStack Foundation “are just starting to recognize that they’ve been doing things in a proprietary fashion, and it’s like they’re not getting traction. Now, they’ve started saying, ‘Oh. The way to get the buy-in is to make things open source’. That’s been happening for a couple of years, and I’ve actually had people come to the conference with the intent of learning how to do open source, so they can take it back to the company and put it into practise.”

Despite these delays among some members of the Foundation, open source has become central to OpenStack’s culture. Given OpenStack’s runaway popularity, even proprietary companies see, at a minimum, the need to provide OpenStack drivers. Among all the startups displaying at the recent OpenStack Summit was a scattering of long-established proprietary companies proclaiming their new allegiance. VMware, for example, had a banner over its booth proclaiming it “The best platform for OpenStack.” Similarly, Cisco Systems was proclaiming that it was “Committed to OpenStack. Committed to You.”

Such slogans might suggest a greater trust in marketing than in open source; yet, more importantly, they also indicate that proprietary companies often perceive themselves as needing to catch up in the move toward acceptance of open source.

The Negative Contribution

Not all the lessons learned from open source are positive. As an open source veteran, Diaz notes that OpenStack’s version of open source is “not your father’s or grandfather’s open source. Things have changed. When IBM was doing Linux and the web, it was really a lot of academics – which was good, and we got a lot of value from that. The difference now is that the best open source is a combination of users and academics and developers.” According to Diaz, IBM had 175 developers at the recent OpenStack Summit, “and they’re all talking with the users about what the users need.”

Diaz is referring to open source’s historical tendency to ignore users in favor of the developers’ preferences. In contrast, he describes OpenStack as learning from this attitude and involving users in the development process. Like the web before it, OpenStack is moving toward greater simplicity. Five years from now, Diaz predicts, OpenStack is “going to be very similar to the web. Most humans don’t think about the web when they type in ‘http’ or ‘ibm.com’ any more. Back in the day they did, because it was all, ‘Omigod, what does this mean?’ I think it will come to a point where [OpenStack] will all be so well-connected that people will be focusing on the digital innovation it allows. It’s not the combination of cloud and data that people will be talking about. It’s the applications. Which is great! We not only sell expertise – we create value by making it easier for users. And when you create a wealth of skills, the pie gets bigger.”

But this direction, like so much of OpenStack’s success to date, is due primarily to the example of open source. “OpenStack is getting traction faster than Linux did,” Chadwick observes, “because Linux had to overcome all that ‘what the hell is open source?’ stuff. And if you look at how long it took Linux to go from a science project to something that people would put some workloads on or an enterprise workload, it was fifteen years for that evolution. But if you look to open source hypervisors, it took something like ten years. And I think OpenStack is going to be much faster than that.”

However, wherever OpenStack goes in the next few years, one fact seems certain: One way or the other, open source will continue to play a major influence. Already, open source is so much a part of the OpenStack culture, it is impossible to imagine any alternative.